mbox series

[RFC/PATCH,net-next,0/9] net/dim: Support for multiple implementations

Message ID 20190306084832.57753-1-talgi@mellanox.com (mailing list archive)
Headers show
Series net/dim: Support for multiple implementations | expand

Message

Tal Gilboa March 6, 2019, 8:48 a.m. UTC
net_dim.h lib exposes an implementation of the DIM algorithm for dynamically-tuned interrupt
moderation for networking interfaces.

We need the same behavior for any block CQ. The main motivation is two benefit from maximized
completion rate and reduced interrupt overhead that DIM may provide.

Current DIM implementation prioritizes reducing interrupt overhead over latency. Also, in
order to reduce DIM's own overhead, the algorithm might take take some time to identify it
needs to change profiles. For these reasons we got to the understanding that a slightly
modified algorithm is needed. Early tests with current implementation show it doesn't react
fast and sharply enough in order to satisfy the block CQ needs.

I would like to suggest an implementation for block DIM. The idea is to expose the new
functionality without the risk of breaking Net DIM behavior for netdev. Below are main
similarities and differences between the two implementations and general guidelines for the
suggested solution.

Performance tests over ConnectX-5 100GbE NIC show a 200% improvement on tail latency when
switching from high load traffic to low load traffic.

Common logic, main DIM procedure:
- Calculate current stats from a given sample
- Compare current stats vs. previous iteration stats
- Make a decision -> choose a new profile

Differences:
- Different parameters for moving between profiles
- Different moderation values and number of profiles
- Different sampled data

Suggested solution:
- Common logic will be declared in include/linux/dim.h and implemented in lib/dim/dim.c
- Net DIM (existing) logic will be declared in include/linux/net_dim.h and implemented in
  lib/dim/net_dim.c, which will use the common logic from dim.h
- Block DIM logic will be declared in /include/linux/block_dim.h and implemented in
  lib/dim/blk_dim.c.
  This new implementation will expose modified versions of profiles, dim_step() and dim_decision()

Pros for this solution are:
- Zero impact on existing net_dim implementation and usage
- Relatively more code reuse (compared to two separate solutions)
- Readiness for future implementations

Tal Gilboa (6):
  linux/dim: Move logic to dim.h
  linux/dim: Remove "net" prefix from internal DIM members
  linux/dim: Rename externally exposed macros
  linux/dim: Rename net_dim_sample() to net_dim_create_sample()
  linux/dim: Rename externally used net_dim members
  linux/dim: Move implementation to .c files

Yamin Friedman (3):
  linux/dim: Add completions count to dim_sample
  linux/dim: Implement blk_dim.h
  drivers/infiniband: Use blk_dim in infiniband driver

 MAINTAINERS                                   |   3 +
 drivers/infiniband/core/cq.c                  |  75 +++-
 drivers/infiniband/hw/mlx4/qp.c               |   2 +-
 drivers/infiniband/hw/mlx5/qp.c               |   2 +-
 drivers/net/ethernet/broadcom/bcmsysport.c    |  20 +-
 drivers/net/ethernet/broadcom/bcmsysport.h    |   2 +-
 drivers/net/ethernet/broadcom/bnxt/bnxt.c     |  13 +-
 drivers/net/ethernet/broadcom/bnxt/bnxt.h     |   2 +-
 .../net/ethernet/broadcom/bnxt/bnxt_debugfs.c |   4 +-
 drivers/net/ethernet/broadcom/bnxt/bnxt_dim.c |   7 +-
 .../net/ethernet/broadcom/genet/bcmgenet.c    |  18 +-
 .../net/ethernet/broadcom/genet/bcmgenet.h    |   2 +-
 drivers/net/ethernet/mellanox/mlx5/core/en.h  |   8 +-
 .../net/ethernet/mellanox/mlx5/core/en_dim.c  |  12 +-
 .../ethernet/mellanox/mlx5/core/en_ethtool.c  |   4 +-
 .../net/ethernet/mellanox/mlx5/core/en_main.c |  22 +-
 .../net/ethernet/mellanox/mlx5/core/en_txrx.c |  10 +-
 include/linux/blk_dim.h                       |  56 +++
 include/linux/dim.h                           | 126 +++++++
 include/linux/irq_poll.h                      |   7 +
 include/linux/net_dim.h                       | 338 +-----------------
 include/rdma/ib_verbs.h                       |  11 +-
 lib/Kconfig                                   |   7 +
 lib/Makefile                                  |   1 +
 lib/dim/Makefile                              |  14 +
 lib/dim/blk_dim.c                             | 114 ++++++
 lib/dim/dim.c                                 |  92 +++++
 lib/dim/net_dim.c                             | 193 ++++++++++
 lib/irq_poll.c                                |  13 +-
 29 files changed, 778 insertions(+), 400 deletions(-)
 create mode 100644 include/linux/blk_dim.h
 create mode 100644 include/linux/dim.h
 create mode 100644 lib/dim/Makefile
 create mode 100644 lib/dim/blk_dim.c
 create mode 100644 lib/dim/dim.c
 create mode 100644 lib/dim/net_dim.c

Comments

Bart Van Assche March 6, 2019, 4:15 p.m. UTC | #1
On Wed, 2019-03-06 at 10:48 +0200, Tal Gilboa wrote:
> net_dim.h lib exposes an implementation of the DIM algorithm for dynamically-tuned interrupt
> moderation for networking interfaces.
> 
> We need the same behavior for any block CQ. The main motivation is two benefit from maximized
> completion rate and reduced interrupt overhead that DIM may provide.

What is a "block CQ"? How does net_dim compare to lib/irq_poll? Which approach
results in the best performance and lowest latency?

Thanks,

Bart.
Sagi Grimberg March 7, 2019, 1:56 a.m. UTC | #2
>> net_dim.h lib exposes an implementation of the DIM algorithm for dynamically-tuned interrupt
>> moderation for networking interfaces.
>>
>> We need the same behavior for any block CQ. The main motivation is two benefit from maximized
>> completion rate and reduced interrupt overhead that DIM may provide.
> 
> What is a "block CQ"?

There is no such thing... Also, this has no difference
if a block/file/whatever is using the rdma cq.

The naming should really be something like rdma_dim as it accounts
for completions and not bytes/packets.

> How does net_dim compare to lib/irq_poll?

Its orthogonal, its basically adaptive interrupt moderation for
RDMA devices. Its sort of below the irq_poll code. It basically
configures interrupt moderation based on stats collected by
the rdma driver.

> Which approach results in the best performance and lowest latency?

I guess it depends on what is the test case. This approach tries to
apply some time or completion count limit to when the HW should fire
an interrupt based on the load in an adaptive fashion.

The scheme is to try and detect what are the load characteristics and
come up with a moderation parameters that fit. For high interrupt rate
(usually seen with small size high queue-depth workloads) it configures
the device to aggregate some more before firing an interrupt - so less
interrupts, better efficiency per interrupt (finds more completions).
For low interrupt rate (low queue depth) the load is probably low to
moderate and aggregating before firing an interrupt is just added
latency for no benefit. So the algorithm tries to transition between a
number of pre-defined levels according to the load it samples.

This has been widely used by the network drivers for the past decade.

Now, this algorithm while trying to adjust itself by learning the load,
also adds entropy to the overall system performance and latency.
So this is not a trivial trade-off for any workload.

I took a stab at this once (came up with something very similar),
and while for large queue-depth workloads I got up to 2x IOPs as the
algorithm chose aggressive moderation parameters which improved the
efficiency a lot, but when the workload varied the algorithm wasn't very
successful detecting the load and the step direction (I used a variation
of the same basic algorithm from mlx5 driver that net_dim is based on).

Also, QD=1 resulted in higher latency as the algorithm was dangling
between the two lowest levels. So I guess this needs to undergo a
thorough performance evaluation for steady and varying workloads before
we can consider this.

Overall, I think its a great idea to add that to the rdma subsystem
but we cannot make it the default and especially without being able
to turn it off. So this needs to be opt in with a sysctl option.

Moreover, not every device support cq moderation so you need to check
the device capabilities before you apply any of this.
Max Gurtovoy March 14, 2019, 11:45 a.m. UTC | #3
On 3/7/2019 3:56 AM, Sagi Grimberg wrote:
>
>>> net_dim.h lib exposes an implementation of the DIM algorithm for 
>>> dynamically-tuned interrupt
>>> moderation for networking interfaces.
>>>
>>> We need the same behavior for any block CQ. The main motivation is 
>>> two benefit from maximized
>>> completion rate and reduced interrupt overhead that DIM may provide.
>>
>> What is a "block CQ"?
>
> There is no such thing... Also, this has no difference
> if a block/file/whatever is using the rdma cq.
>
> The naming should really be something like rdma_dim as it accounts
> for completions and not bytes/packets.

Sagi,

I think that in the future we could use it in nvme since there is an 
option to set the interrupt coalescing in NVMe spec.

This might improve performance for NVMe driver.

We already see some bottlenecks in performance (maybe driver ones) while 
developing the NVMe SNAP feature in Bluefield (NVMe emulation using 
Smart NIC).

We're trying to get 2.5-2.7 MIOPs OOB from 1 controller and it's not 
trivial for today's driver.

So let's take this into consideration when we set the naming.


>
>
>> How does net_dim compare to lib/irq_poll?
>
> Its orthogonal, its basically adaptive interrupt moderation for
> RDMA devices. Its sort of below the irq_poll code. It basically
> configures interrupt moderation based on stats collected by
> the rdma driver.
>
>> Which approach results in the best performance and lowest latency?
>
> I guess it depends on what is the test case. This approach tries to
> apply some time or completion count limit to when the HW should fire
> an interrupt based on the load in an adaptive fashion.
>
> The scheme is to try and detect what are the load characteristics and
> come up with a moderation parameters that fit. For high interrupt rate
> (usually seen with small size high queue-depth workloads) it configures
> the device to aggregate some more before firing an interrupt - so less
> interrupts, better efficiency per interrupt (finds more completions).
> For low interrupt rate (low queue depth) the load is probably low to
> moderate and aggregating before firing an interrupt is just added
> latency for no benefit. So the algorithm tries to transition between a
> number of pre-defined levels according to the load it samples.
>
> This has been widely used by the network drivers for the past decade.
>
> Now, this algorithm while trying to adjust itself by learning the load,
> also adds entropy to the overall system performance and latency.
> So this is not a trivial trade-off for any workload.
>
> I took a stab at this once (came up with something very similar),
> and while for large queue-depth workloads I got up to 2x IOPs as the
> algorithm chose aggressive moderation parameters which improved the
> efficiency a lot, but when the workload varied the algorithm wasn't very
> successful detecting the load and the step direction (I used a variation
> of the same basic algorithm from mlx5 driver that net_dim is based on).
>
> Also, QD=1 resulted in higher latency as the algorithm was dangling
> between the two lowest levels. So I guess this needs to undergo a
> thorough performance evaluation for steady and varying workloads before
> we can consider this.
>
> Overall, I think its a great idea to add that to the rdma subsystem
> but we cannot make it the default and especially without being able
> to turn it off. So this needs to be opt in with a sysctl option.

We can add flag in create_cq command that will 
try_coalescing_is_possible instead of module parameter of course.

Storage ULPs can set it to True.

Also in the internal review Yamin added a table that summarize all the 
testing that were done using NVMeoF (I guess it somehow didn't get to 
this RFC).

I guess we can do the same for iSER to get more confidence and then set 
both to create modifiable cq (if HCA supports, of course).

Agreed ?

>
>
> Moreover, not every device support cq moderation so you need to check
> the device capabilities before you apply any of this.

for sure.
Sagi Grimberg March 14, 2019, 9:53 p.m. UTC | #4
>>> What is a "block CQ"?
>>
>> There is no such thing... Also, this has no difference
>> if a block/file/whatever is using the rdma cq.
>>
>> The naming should really be something like rdma_dim as it accounts
>> for completions and not bytes/packets.
> 
> Sagi,
> 
> I think that in the future we could use it in nvme since there is an 
> option to set the interrupt coalescing in NVMe spec.
> 
> This might improve performance for NVMe driver.

That would require changing the spec to make moderation config per-queue
and not controller-wide. This does not apply specifically to block so
naming it with blk does not make sense.

>> Overall, I think its a great idea to add that to the rdma subsystem
>> but we cannot make it the default and especially without being able
>> to turn it off. So this needs to be opt in with a sysctl option.
> 
> We can add flag in create_cq command that will 
> try_coalescing_is_possible instead of module parameter of course.
> 
> Storage ULPs can set it to True.

The point is that it can't be universally on.

> Also in the internal review Yamin added a table that summarize all the 
> testing that were done using NVMeoF (I guess it somehow didn't get to 
> this RFC).
> 
> I guess we can do the same for iSER to get more confidence and then set 
> both to create modifiable cq (if HCA supports, of course).
> 
> Agreed ?

Sure.
Yamin Friedman March 18, 2019, 9:24 a.m. UTC | #5
On 3/14/2019 1:45 PM, Max Gurtovoy wrote:
>
> On 3/7/2019 3:56 AM, Sagi Grimberg wrote:
>>
>>>> net_dim.h lib exposes an implementation of the DIM algorithm for 
>>>> dynamically-tuned interrupt
>>>> moderation for networking interfaces.
>>>>
>>>> We need the same behavior for any block CQ. The main motivation is 
>>>> two benefit from maximized
>>>> completion rate and reduced interrupt overhead that DIM may provide.
>>>
>>> What is a "block CQ"?
>>
>> There is no such thing... Also, this has no difference
>> if a block/file/whatever is using the rdma cq.
>>
>> The naming should really be something like rdma_dim as it accounts
>> for completions and not bytes/packets.
>
> Sagi,
>
> I think that in the future we could use it in nvme since there is an 
> option to set the interrupt coalescing in NVMe spec.
>
> This might improve performance for NVMe driver.
>
> We already see some bottlenecks in performance (maybe driver ones) 
> while developing the NVMe SNAP feature in Bluefield (NVMe emulation 
> using Smart NIC).
>
> We're trying to get 2.5-2.7 MIOPs OOB from 1 controller and it's not 
> trivial for today's driver.
>
> So let's take this into consideration when we set the naming.
>
>
I agree that blk is not the most successful name, we were trying to find 
something that would work for general storage applications. I think 
rdma_dim would work as it is completion based but then when we want to 
use it for nvme it will probably require code duplication.

>>
>>
>>> How does net_dim compare to lib/irq_poll?
>>
>> Its orthogonal, its basically adaptive interrupt moderation for
>> RDMA devices. Its sort of below the irq_poll code. It basically
>> configures interrupt moderation based on stats collected by
>> the rdma driver.
>>
>>> Which approach results in the best performance and lowest latency?
>>
>> I guess it depends on what is the test case. This approach tries to
>> apply some time or completion count limit to when the HW should fire
>> an interrupt based on the load in an adaptive fashion.
>>
>> The scheme is to try and detect what are the load characteristics and
>> come up with a moderation parameters that fit. For high interrupt rate
>> (usually seen with small size high queue-depth workloads) it configures
>> the device to aggregate some more before firing an interrupt - so less
>> interrupts, better efficiency per interrupt (finds more completions).
>> For low interrupt rate (low queue depth) the load is probably low to
>> moderate and aggregating before firing an interrupt is just added
>> latency for no benefit. So the algorithm tries to transition between a
>> number of pre-defined levels according to the load it samples.
>>
>> This has been widely used by the network drivers for the past decade.
>>
>> Now, this algorithm while trying to adjust itself by learning the load,
>> also adds entropy to the overall system performance and latency.
>> So this is not a trivial trade-off for any workload.
>>
>> I took a stab at this once (came up with something very similar),
>> and while for large queue-depth workloads I got up to 2x IOPs as the
>> algorithm chose aggressive moderation parameters which improved the
>> efficiency a lot, but when the workload varied the algorithm wasn't very
>> successful detecting the load and the step direction (I used a variation
>> of the same basic algorithm from mlx5 driver that net_dim is based on).
>>
>> Also, QD=1 resulted in higher latency as the algorithm was dangling
>> between the two lowest levels. So I guess this needs to undergo a
>> thorough performance evaluation for steady and varying workloads before
>> we can consider this.
>>
>> Overall, I think its a great idea to add that to the rdma subsystem
>> but we cannot make it the default and especially without being able
>> to turn it off. So this needs to be opt in with a sysctl option.
>
> We can add flag in create_cq command that will 
> try_coalescing_is_possible instead of module parameter of course.
>
> Storage ULPs can set it to True.
>
> Also in the internal review Yamin added a table that summarize all the 
> testing that were done using NVMeoF (I guess it somehow didn't get to 
> this RFC).
>
> I guess we can do the same for iSER to get more confidence and then 
> set both to create modifiable cq (if HCA supports, of course).
>
> Agreed ?
>
I think that adding a flag in create_cq will be less clean as it will 
require more work for anyone writing applications that should not have 
to consider this feature.

Based on the results I saw during testing I would set it to work by 
default as I could not find a use case where it significantly reduces 
performance and in many cases it is a large improvement. It should be 
more of an opt out situation.

Performance improvement (ConnectX-5 100GbE, x86) running FIO benchmark over
     NVMf between two equal end-hosts with 56 cores across a Mellanox switch
     using null_blk device:

     IO READS before:
     blk size | BW      | IOPS | 99th percentile latency
     512B     | 3.2GiB  | 6.6M | 1549  usec
     4k       | 7.2GiB  | 1.8M | 7177  usec
     64k      | 10.7GiB | 176k | 82314 usec

     IO READS after:
     blk size | BW      | IOPS | 99th percentile latency
     512B     | 4.2GiB  | 8.6M | 1729   usec
     4k       | 10.5GiB | 2.7M | 5669   usec
     64k      | 10.7GiB | 176k | 102000 usec

     IO WRITES before:
     blk size | BW      | IOPS | 99th percentile latency
     512B     | 3GiB    | 6.2M | 2573  usec
     4k       | 7.2GiB  | 1.8M | 5342  usec
     64k      | 10.7GiB | 176k | 62129 usec

     IO WRITES after:
     blk size | BW      | IOPS  | 99th percentile latency
     512B     | 4.2GiB  | 8.6M  | 938   usec
     4k       | 10.2GiB | 2.68M | 2769  usec
     64k      | 10.6GiB | 173k  | 87557 usec

It doesn't really make a difference to me how the option is implemented 
but I think it makes more sense to have it dealt with by us such as in a 
module parameter and not something like a flag that has a larger radius 
of effect.

>>
>>
>> Moreover, not every device support cq moderation so you need to check
>> the device capabilities before you apply any of this.
>
> for sure.
>
>
Max Gurtovoy March 18, 2019, 11:08 a.m. UTC | #6
On 3/18/2019 11:24 AM, Yamin Friedman wrote:
> On 3/14/2019 1:45 PM, Max Gurtovoy wrote:
>> On 3/7/2019 3:56 AM, Sagi Grimberg wrote:
>>>>> net_dim.h lib exposes an implementation of the DIM algorithm for
>>>>> dynamically-tuned interrupt
>>>>> moderation for networking interfaces.
>>>>>
>>>>> We need the same behavior for any block CQ. The main motivation is
>>>>> two benefit from maximized
>>>>> completion rate and reduced interrupt overhead that DIM may provide.
>>>> What is a "block CQ"?
>>> There is no such thing... Also, this has no difference
>>> if a block/file/whatever is using the rdma cq.
>>>
>>> The naming should really be something like rdma_dim as it accounts
>>> for completions and not bytes/packets.
>> Sagi,
>>
>> I think that in the future we could use it in nvme since there is an
>> option to set the interrupt coalescing in NVMe spec.
>>
>> This might improve performance for NVMe driver.
>>
>> We already see some bottlenecks in performance (maybe driver ones)
>> while developing the NVMe SNAP feature in Bluefield (NVMe emulation
>> using Smart NIC).
>>
>> We're trying to get 2.5-2.7 MIOPs OOB from 1 controller and it's not
>> trivial for today's driver.
>>
>> So let's take this into consideration when we set the naming.
>>
>>
> I agree that blk is not the most successful name, we were trying to find
> something that would work for general storage applications. I think
> rdma_dim would work as it is completion based but then when we want to
> use it for nvme it will probably require code duplication.

agreed for rdma_dim.

Yamin/Idan,

let's discuss internally regarding Sagi's proposal for 
moderation/coalescing per NVMe queue (and not per controller).

maybe need to update specification.


>
>>>
>>>> How does net_dim compare to lib/irq_poll?
>>> Its orthogonal, its basically adaptive interrupt moderation for
>>> RDMA devices. Its sort of below the irq_poll code. It basically
>>> configures interrupt moderation based on stats collected by
>>> the rdma driver.
>>>
>>>> Which approach results in the best performance and lowest latency?
>>> I guess it depends on what is the test case. This approach tries to
>>> apply some time or completion count limit to when the HW should fire
>>> an interrupt based on the load in an adaptive fashion.
>>>
>>> The scheme is to try and detect what are the load characteristics and
>>> come up with a moderation parameters that fit. For high interrupt rate
>>> (usually seen with small size high queue-depth workloads) it configures
>>> the device to aggregate some more before firing an interrupt - so less
>>> interrupts, better efficiency per interrupt (finds more completions).
>>> For low interrupt rate (low queue depth) the load is probably low to
>>> moderate and aggregating before firing an interrupt is just added
>>> latency for no benefit. So the algorithm tries to transition between a
>>> number of pre-defined levels according to the load it samples.
>>>
>>> This has been widely used by the network drivers for the past decade.
>>>
>>> Now, this algorithm while trying to adjust itself by learning the load,
>>> also adds entropy to the overall system performance and latency.
>>> So this is not a trivial trade-off for any workload.
>>>
>>> I took a stab at this once (came up with something very similar),
>>> and while for large queue-depth workloads I got up to 2x IOPs as the
>>> algorithm chose aggressive moderation parameters which improved the
>>> efficiency a lot, but when the workload varied the algorithm wasn't very
>>> successful detecting the load and the step direction (I used a variation
>>> of the same basic algorithm from mlx5 driver that net_dim is based on).
>>>
>>> Also, QD=1 resulted in higher latency as the algorithm was dangling
>>> between the two lowest levels. So I guess this needs to undergo a
>>> thorough performance evaluation for steady and varying workloads before
>>> we can consider this.
>>>
>>> Overall, I think its a great idea to add that to the rdma subsystem
>>> but we cannot make it the default and especially without being able
>>> to turn it off. So this needs to be opt in with a sysctl option.
>> We can add flag in create_cq command that will
>> try_coalescing_is_possible instead of module parameter of course.
>>
>> Storage ULPs can set it to True.
>>
>> Also in the internal review Yamin added a table that summarize all the
>> testing that were done using NVMeoF (I guess it somehow didn't get to
>> this RFC).
>>
>> I guess we can do the same for iSER to get more confidence and then
>> set both to create modifiable cq (if HCA supports, of course).
>>
>> Agreed ?
>>
> I think that adding a flag in create_cq will be less clean as it will
> require more work for anyone writing applications that should not have
> to consider this feature.


As we discussed, let's check with RDMA maintainers if it's better to 
extend alloc_cq API or create alloc_cq_dim API function.

Sagi/Christoph,

how about adding a module param per ULP ? as we use register_always 
today, create a use_dimm module param for iSER/NVMe-RDMA ?

>
> Based on the results I saw during testing I would set it to work by
> default as I could not find a use case where it significantly reduces
> performance and in many cases it is a large improvement. It should be
> more of an opt out situation.
>
> Performance improvement (ConnectX-5 100GbE, x86) running FIO benchmark over
>       NVMf between two equal end-hosts with 56 cores across a Mellanox switch
>       using null_blk device:
>
>       IO READS before:
>       blk size | BW      | IOPS | 99th percentile latency
>       512B     | 3.2GiB  | 6.6M | 1549  usec
>       4k       | 7.2GiB  | 1.8M | 7177  usec
>       64k      | 10.7GiB | 176k | 82314 usec
>
>       IO READS after:
>       blk size | BW      | IOPS | 99th percentile latency
>       512B     | 4.2GiB  | 8.6M | 1729   usec
>       4k       | 10.5GiB | 2.7M | 5669   usec
>       64k      | 10.7GiB | 176k | 102000 usec
>
>       IO WRITES before:
>       blk size | BW      | IOPS | 99th percentile latency
>       512B     | 3GiB    | 6.2M | 2573  usec
>       4k       | 7.2GiB  | 1.8M | 5342  usec
>       64k      | 10.7GiB | 176k | 62129 usec
>
>       IO WRITES after:
>       blk size | BW      | IOPS  | 99th percentile latency
>       512B     | 4.2GiB  | 8.6M  | 938   usec
>       4k       | 10.2GiB | 2.68M | 2769  usec
>       64k      | 10.6GiB | 173k  | 87557 usec
>
> It doesn't really make a difference to me how the option is implemented
> but I think it makes more sense to have it dealt with by us such as in a
> module parameter and not something like a flag that has a larger radius
> of effect.
>
>>>
>>> Moreover, not every device support cq moderation so you need to check
>>> the device capabilities before you apply any of this.
>> for sure.
>>
>>
Max Gurtovoy March 18, 2019, 3:05 p.m. UTC | #7
On 3/18/2019 1:08 PM, Max Gurtovoy wrote:
>
> On 3/18/2019 11:24 AM, Yamin Friedman wrote:
>> On 3/14/2019 1:45 PM, Max Gurtovoy wrote:
>>> On 3/7/2019 3:56 AM, Sagi Grimberg wrote:
>>>>>> net_dim.h lib exposes an implementation of the DIM algorithm for
>>>>>> dynamically-tuned interrupt
>>>>>> moderation for networking interfaces.
>>>>>>
>>>>>> We need the same behavior for any block CQ. The main motivation is
>>>>>> two benefit from maximized
>>>>>> completion rate and reduced interrupt overhead that DIM may provide.
>>>>> What is a "block CQ"?
>>>> There is no such thing... Also, this has no difference
>>>> if a block/file/whatever is using the rdma cq.
>>>>
>>>> The naming should really be something like rdma_dim as it accounts
>>>> for completions and not bytes/packets.
>>> Sagi,
>>>
>>> I think that in the future we could use it in nvme since there is an
>>> option to set the interrupt coalescing in NVMe spec.
>>>
>>> This might improve performance for NVMe driver.
>>>
>>> We already see some bottlenecks in performance (maybe driver ones)
>>> while developing the NVMe SNAP feature in Bluefield (NVMe emulation
>>> using Smart NIC).
>>>
>>> We're trying to get 2.5-2.7 MIOPs OOB from 1 controller and it's not
>>> trivial for today's driver.
>>>
>>> So let's take this into consideration when we set the naming.
>>>
>>>
>> I agree that blk is not the most successful name, we were trying to find
>> something that would work for general storage applications. I think
>> rdma_dim would work as it is completion based but then when we want to
>> use it for nvme it will probably require code duplication.
>
> agreed for rdma_dim.
>
> Yamin/Idan,
>
> let's discuss internally regarding Sagi's proposal for 
> moderation/coalescing per NVMe queue (and not per controller).
>
> maybe need to update specification.
>
>
>>
>>>>
>>>>> How does net_dim compare to lib/irq_poll?
>>>> Its orthogonal, its basically adaptive interrupt moderation for
>>>> RDMA devices. Its sort of below the irq_poll code. It basically
>>>> configures interrupt moderation based on stats collected by
>>>> the rdma driver.
>>>>
>>>>> Which approach results in the best performance and lowest latency?
>>>> I guess it depends on what is the test case. This approach tries to
>>>> apply some time or completion count limit to when the HW should fire
>>>> an interrupt based on the load in an adaptive fashion.
>>>>
>>>> The scheme is to try and detect what are the load characteristics and
>>>> come up with a moderation parameters that fit. For high interrupt rate
>>>> (usually seen with small size high queue-depth workloads) it 
>>>> configures
>>>> the device to aggregate some more before firing an interrupt - so less
>>>> interrupts, better efficiency per interrupt (finds more completions).
>>>> For low interrupt rate (low queue depth) the load is probably low to
>>>> moderate and aggregating before firing an interrupt is just added
>>>> latency for no benefit. So the algorithm tries to transition between a
>>>> number of pre-defined levels according to the load it samples.
>>>>
>>>> This has been widely used by the network drivers for the past decade.
>>>>
>>>> Now, this algorithm while trying to adjust itself by learning the 
>>>> load,
>>>> also adds entropy to the overall system performance and latency.
>>>> So this is not a trivial trade-off for any workload.
>>>>
>>>> I took a stab at this once (came up with something very similar),
>>>> and while for large queue-depth workloads I got up to 2x IOPs as the
>>>> algorithm chose aggressive moderation parameters which improved the
>>>> efficiency a lot, but when the workload varied the algorithm wasn't 
>>>> very
>>>> successful detecting the load and the step direction (I used a 
>>>> variation
>>>> of the same basic algorithm from mlx5 driver that net_dim is based 
>>>> on).
>>>>
>>>> Also, QD=1 resulted in higher latency as the algorithm was dangling
>>>> between the two lowest levels. So I guess this needs to undergo a
>>>> thorough performance evaluation for steady and varying workloads 
>>>> before
>>>> we can consider this.
>>>>
>>>> Overall, I think its a great idea to add that to the rdma subsystem
>>>> but we cannot make it the default and especially without being able
>>>> to turn it off. So this needs to be opt in with a sysctl option.
>>> We can add flag in create_cq command that will
>>> try_coalescing_is_possible instead of module parameter of course.
>>>
>>> Storage ULPs can set it to True.
>>>
>>> Also in the internal review Yamin added a table that summarize all the
>>> testing that were done using NVMeoF (I guess it somehow didn't get to
>>> this RFC).
>>>
>>> I guess we can do the same for iSER to get more confidence and then
>>> set both to create modifiable cq (if HCA supports, of course).
>>>
>>> Agreed ?
>>>
>> I think that adding a flag in create_cq will be less clean as it will
>> require more work for anyone writing applications that should not have
>> to consider this feature.
>
>
> As we discussed, let's check with RDMA maintainers if it's better to 
> extend alloc_cq API or create alloc_cq_dim API function.
>
> Sagi/Christoph,
>
> how about adding a module param per ULP ? as we use register_always 
> today, create a use_dimm module param for iSER/NVMe-RDMA ?

another option is to add it to each ULP user space utility (nvme-cli, 
iscsiadm, etc...)

thoughts ?


>
>
>>
>> Based on the results I saw during testing I would set it to work by
>> default as I could not find a use case where it significantly reduces
>> performance and in many cases it is a large improvement. It should be
>> more of an opt out situation.
>>
>> Performance improvement (ConnectX-5 100GbE, x86) running FIO 
>> benchmark over
>>       NVMf between two equal end-hosts with 56 cores across a 
>> Mellanox switch
>>       using null_blk device:
>>
>>       IO READS before:
>>       blk size | BW      | IOPS | 99th percentile latency
>>       512B     | 3.2GiB  | 6.6M | 1549  usec
>>       4k       | 7.2GiB  | 1.8M | 7177  usec
>>       64k      | 10.7GiB | 176k | 82314 usec
>>
>>       IO READS after:
>>       blk size | BW      | IOPS | 99th percentile latency
>>       512B     | 4.2GiB  | 8.6M | 1729   usec
>>       4k       | 10.5GiB | 2.7M | 5669   usec
>>       64k      | 10.7GiB | 176k | 102000 usec
>>
>>       IO WRITES before:
>>       blk size | BW      | IOPS | 99th percentile latency
>>       512B     | 3GiB    | 6.2M | 2573  usec
>>       4k       | 7.2GiB  | 1.8M | 5342  usec
>>       64k      | 10.7GiB | 176k | 62129 usec
>>
>>       IO WRITES after:
>>       blk size | BW      | IOPS  | 99th percentile latency
>>       512B     | 4.2GiB  | 8.6M  | 938   usec
>>       4k       | 10.2GiB | 2.68M | 2769  usec
>>       64k      | 10.6GiB | 173k  | 87557 usec
>>
>> It doesn't really make a difference to me how the option is implemented
>> but I think it makes more sense to have it dealt with by us such as in a
>> module parameter and not something like a flag that has a larger radius
>> of effect.
>>
>>>>
>>>> Moreover, not every device support cq moderation so you need to check
>>>> the device capabilities before you apply any of this.
>>> for sure.
>>>
>>>
Sagi Grimberg March 18, 2019, 9:32 p.m. UTC | #8
> I agree that blk is not the most successful name, we were trying to find
> something that would work for general storage applications. I think
> rdma_dim would work as it is completion based but then when we want to
> use it for nvme it will probably require code duplication.

Lets worry about reuse NVMe when it is actually applicable.

>> Also in the internal review Yamin added a table that summarize all the
>> testing that were done using NVMeoF (I guess it somehow didn't get to
>> this RFC).
>>
>> I guess we can do the same for iSER to get more confidence and then
>> set both to create modifiable cq (if HCA supports, of course).
>>
>> Agreed ?
>>
> I think that adding a flag in create_cq will be less clean as it will
> require more work for anyone writing applications that should not have
> to consider this feature.
> 
> Based on the results I saw during testing I would set it to work by
> default as I could not find a use case where it significantly reduces
> performance and in many cases it is a large improvement. It should be
> more of an opt out situation.

By detailed performance results I meant:
1. Full latency histogram for QD=1 both for single queue and multi-queue
(including max, 99.99% and 99.999% percentiles)
2. latency vs. IOPs graph/table for both single queue and multi-queue
3. At least some measurement/analysis of how well the algorithm is
    adapting to workload change dynamically and how quickly.
4. Test also with real NVMe devices.

Also, we need to separate the host side moderation and the target
side moderation to understand if/how they effect each other.

Its very easy to show that for high stress workloads you can get an
improvement as obviously there is a clear win for interrupt moderation,
however, if this was the only metric that is interesting, we wouldn't
need it to be adaptive.

As I said before, this adds entropy to the equation which in certain use
cases can make more harm than good, and we need to quantify where is the
impact and understand how important they are compared to the extremely
niche use-case of a single host pushing 2M-8M IOPs.

> Performance improvement (ConnectX-5 100GbE, x86) running FIO benchmark over
>       NVMf between two equal end-hosts with 56 cores across a Mellanox switch
>       using null_blk device:
> 
>       IO READS before:
>       blk size | BW      | IOPS | 99th percentile latency
>       512B     | 3.2GiB  | 6.6M | 1549  usec
>       4k       | 7.2GiB  | 1.8M | 7177  usec
>       64k      | 10.7GiB | 176k | 82314 usec

I've seen this before, why are we not getting 100Gb/s for 4k with CX5?
I recall we used to get it with CX4.

>       IO READS after:
>       blk size | BW      | IOPS | 99th percentile latency
>       512B     | 4.2GiB  | 8.6M | 1729   usec
>       4k       | 10.5GiB | 2.7M | 5669   usec
>       64k      | 10.7GiB | 176k | 102000 usec
> 
>       IO WRITES before:
>       blk size | BW      | IOPS | 99th percentile latency
>       512B     | 3GiB    | 6.2M | 2573  usec
>       4k       | 7.2GiB  | 1.8M | 5342  usec
>       64k      | 10.7GiB | 176k | 62129 usec
> 
>       IO WRITES after:
>       blk size | BW      | IOPS  | 99th percentile latency
>       512B     | 4.2GiB  | 8.6M  | 938   usec
>       4k       | 10.2GiB | 2.68M | 2769  usec
>       64k      | 10.6GiB | 173k  | 87557 usec

The fact that the 64k 99% latency is substantially higher (20+
milliseconds) without any BW benefit, while its not a very interesting
measurement, gives me an indication that a more detailed analysis needs
to be made here to understand where are the trade-offs.

> It doesn't really make a difference to me how the option is implemented
> but I think it makes more sense to have it dealt with by us such as in a
> module parameter and not something like a flag that has a larger radius
> of effect.

I was suggesting a sysctl global parameter for global behavior and of
someone wants to override it it can add a CQ flag (which follows the
common net params exactly).
Sagi Grimberg March 18, 2019, 9:34 p.m. UTC | #9
> As we discussed, let's check with RDMA maintainers if it's better to 
> extend alloc_cq API or create alloc_cq_dim API function.
> 
> Sagi/Christoph,
> 
> how about adding a module param per ULP ? as we use register_always 
> today, create a use_dimm module param for iSER/NVMe-RDMA ?

I would say that its better (and simpler) to do a global sysctl knob for
it. No need for a per-ULP param for a starting point.
Max Gurtovoy March 20, 2019, 9:17 a.m. UTC | #10
On 3/18/2019 11:34 PM, Sagi Grimberg wrote:
>
>> As we discussed, let's check with RDMA maintainers if it's better to 
>> extend alloc_cq API or create alloc_cq_dim API function.
>>
>> Sagi/Christoph,
>>
>> how about adding a module param per ULP ? as we use register_always 
>> today, create a use_dimm module param for iSER/NVMe-RDMA ?
>
> I would say that its better (and simpler) to do a global sysctl knob for
> it. No need for a per-ULP param for a starting point.

for sure it's simpler, but in this case all should agree that in case we 
run more than 1 ULP in a server, all of them will have the same 
configuration (no QoS).

are you suggesting something like:

/proc/sys/rdma/mlx5_0/dim

Jason/Leon/Doug,

thoughts about the best way to configure this feature from your 
perspective ?
Leon Romanovsky March 20, 2019, 11:10 a.m. UTC | #11
On Wed, Mar 20, 2019 at 11:17:36AM +0200, Max Gurtovoy wrote:
>
> On 3/18/2019 11:34 PM, Sagi Grimberg wrote:
> >
> > > As we discussed, let's check with RDMA maintainers if it's better to
> > > extend alloc_cq API or create alloc_cq_dim API function.
> > >
> > > Sagi/Christoph,
> > >
> > > how about adding a module param per ULP ? as we use register_always
> > > today, create a use_dimm module param for iSER/NVMe-RDMA ?
> >
> > I would say that its better (and simpler) to do a global sysctl knob for
> > it. No need for a per-ULP param for a starting point.
>
> for sure it's simpler, but in this case all should agree that in case we run
> more than 1 ULP in a server, all of them will have the same configuration
> (no QoS).
>
> are you suggesting something like:
>
> /proc/sys/rdma/mlx5_0/dim
>
> Jason/Leon/Doug,
>
> thoughts about the best way to configure this feature from your perspective
> ?

It doesn't sound reasonable that we have per-ULP feature (DIM is
per-ULP) to be configured globally. Especially given the fact that
users can find themselves running different workloads with different
requirements on the same system.

Currently each ULP has some sort of tool to configure itself and I think
that once ULP is converted to use DIM, it should have on/off knob in the
tool used by their users.

Thanks

>
Sagi Grimberg March 20, 2019, 6:34 p.m. UTC | #12
> It doesn't sound reasonable that we have per-ULP feature (DIM is
> per-ULP) to be configured globally. Especially given the fact that
> users can find themselves running different workloads with different
> requirements on the same system.
> 
> Currently each ULP has some sort of tool to configure itself and I think
> that once ULP is converted to use DIM, it should have on/off knob in the
> tool used by their users.

Its not any different than socket options that can have a global sysctl
knob that can be overridden by individual socket consumers.
Yamin Friedman March 21, 2019, 5:53 a.m. UTC | #13
On 3/18/2019 11:32 PM, Sagi Grimberg wrote:
>
>> I agree that blk is not the most successful name, we were trying to find
>> something that would work for general storage applications. I think
>> rdma_dim would work as it is completion based but then when we want to
>> use it for nvme it will probably require code duplication.
>
> Lets worry about reuse NVMe when it is actually applicable.
>
>>> Also in the internal review Yamin added a table that summarize all the
>>> testing that were done using NVMeoF (I guess it somehow didn't get to
>>> this RFC).
>>>
>>> I guess we can do the same for iSER to get more confidence and then
>>> set both to create modifiable cq (if HCA supports, of course).
>>>
>>> Agreed ?
>>>
>> I think that adding a flag in create_cq will be less clean as it will
>> require more work for anyone writing applications that should not have
>> to consider this feature.
>>
>> Based on the results I saw during testing I would set it to work by
>> default as I could not find a use case where it significantly reduces
>> performance and in many cases it is a large improvement. It should be
>> more of an opt out situation.
>
> By detailed performance results I meant:
> 1. Full latency histogram for QD=1 both for single queue and multi-queue
> (including max, 99.99% and 99.999% percentiles)
> 2. latency vs. IOPs graph/table for both single queue and multi-queue
> 3. At least some measurement/analysis of how well the algorithm is
>    adapting to workload change dynamically and how quickly.
> 4. Test also with real NVMe devices.
>
> Also, we need to separate the host side moderation and the target
> side moderation to understand if/how they effect each other.
>
> Its very easy to show that for high stress workloads you can get an
> improvement as obviously there is a clear win for interrupt moderation,
> however, if this was the only metric that is interesting, we wouldn't
> need it to be adaptive.
>
> As I said before, this adds entropy to the equation which in certain use
> cases can make more harm than good, and we need to quantify where is the
> impact and understand how important they are compared to the extremely
> niche use-case of a single host pushing 2M-8M IOPs.


I ran extensive tests between two hosts using FIO and NVMeoF with and 
without DIM on each side separately and together. I am attaching the 
results of the tests. The main point I would like to make is that when 
used on both sides especially we see improvements across the board in 
both IOPS and latency.

>
>> Performance improvement (ConnectX-5 100GbE, x86) running FIO 
>> benchmark over
>>       NVMf between two equal end-hosts with 56 cores across a 
>> Mellanox switch
>>       using null_blk device:
>>
>>       IO READS before:
>>       blk size | BW      | IOPS | 99th percentile latency
>>       512B     | 3.2GiB  | 6.6M | 1549  usec
>>       4k       | 7.2GiB  | 1.8M | 7177  usec
>>       64k      | 10.7GiB | 176k | 82314 usec
>
> I've seen this before, why are we not getting 100Gb/s for 4k with CX5?
> I recall we used to get it with CX4.

It depends on the host server and whether or not it causes backpressure 
on the PCI because of the amount of interrupts.

>
>>       IO READS after:
>>       blk size | BW      | IOPS | 99th percentile latency
>>       512B     | 4.2GiB  | 8.6M | 1729   usec
>>       4k       | 10.5GiB | 2.7M | 5669   usec
>>       64k      | 10.7GiB | 176k | 102000 usec
>>
>>       IO WRITES before:
>>       blk size | BW      | IOPS | 99th percentile latency
>>       512B     | 3GiB    | 6.2M | 2573  usec
>>       4k       | 7.2GiB  | 1.8M | 5342  usec
>>       64k      | 10.7GiB | 176k | 62129 usec
>>
>>       IO WRITES after:
>>       blk size | BW      | IOPS  | 99th percentile latency
>>       512B     | 4.2GiB  | 8.6M  | 938   usec
>>       4k       | 10.2GiB | 2.68M | 2769  usec
>>       64k      | 10.6GiB | 173k  | 87557 usec
>
> The fact that the 64k 99% latency is substantially higher (20+
> milliseconds) without any BW benefit, while its not a very interesting
> measurement, gives me an indication that a more detailed analysis needs
> to be made here to understand where are the trade-offs.
>
>> It doesn't really make a difference to me how the option is implemented
>> but I think it makes more sense to have it dealt with by us such as in a
>> module parameter and not something like a flag that has a larger radius
>> of effect.
>
> I was suggesting a sysctl global parameter for global behavior and of
> someone wants to override it it can add a CQ flag (which follows the
> common net params exactly).
With DIM							Without DIM
Type=read,Size=512B,NumJobs=1,depth=1				Type=read,Size=512B,NumJobs=1,depth=1
   read: IOPS=56.7k, BW=27.7MiB/s (29.0MB/s)(277MiB/10001msec |	   read: IOPS=47.0k, BW=22.0MiB/s (24.1MB/s)(230MiB/10001msec
    slat (nsec): min=2087, max=43251, avg=2462.19, stdev=1301 |	    slat (nsec): min=1981, max=53196, avg=3048.29, stdev=2391
    clat (nsec): min=623, max=108655, avg=14747.79, stdev=468 |	    clat (nsec): min=442, max=107156, avg=17658.37, stdev=647
     lat (usec): min=12, max=110, avg=17.25, stdev= 5.05      |	     lat (usec): min=13, max=109, avg=20.78, stdev= 7.18
     |  1.00th=[12992],  5.00th=[12992], 10.00th=[13120], 20. |	     |  1.00th=[13632],  5.00th=[13760], 10.00th=[13888], 20.
     | 30.00th=[13120], 40.00th=[13248], 50.00th=[13248], 60. |	     | 30.00th=[14144], 40.00th=[14144], 50.00th=[14400], 60.
     | 70.00th=[13504], 80.00th=[13888], 90.00th=[16768], 95. |	     | 70.00th=[17024], 80.00th=[19072], 90.00th=[27776], 95.
     | 99.00th=[35584], 99.50th=[36608], 99.90th=[42240], 99. |	     | 99.00th=[40704], 99.50th=[45824], 99.90th=[51456], 99.
     | 99.99th=[53504]					      |	     | 99.99th=[58624]
Type=read,Size=512B,NumJobs=1,depth=16				Type=read,Size=512B,NumJobs=1,depth=16
   read: IOPS=299k, BW=146MiB/s (153MB/s)(1459MiB/10001msec)  |	   read: IOPS=334k, BW=163MiB/s (171MB/s)(1629MiB/10001msec)
    slat (nsec): min=1634, max=45257, avg=1958.38, stdev=657. |	    slat (nsec): min=1564, max=28780, avg=2162.48, stdev=588.
    clat (usec): min=38, max=171, avg=51.28, stdev=13.39      |	    clat (usec): min=17, max=118, avg=45.50, stdev= 6.30
     lat (usec): min=40, max=172, avg=53.27, stdev=13.41      |	     lat (usec): min=20, max=120, avg=47.70, stdev= 6.64
     |  1.00th=[   40],  5.00th=[   41], 10.00th=[   41], 20. |	     |  1.00th=[41728],  5.00th=[42752], 10.00th=[42752], 20.
     | 30.00th=[   42], 40.00th=[   43], 50.00th=[   46], 60. |	     | 30.00th=[43264], 40.00th=[43776], 50.00th=[43776], 60.
     | 70.00th=[   57], 80.00th=[   62], 90.00th=[   70], 95. |	     | 70.00th=[44288], 80.00th=[44288], 90.00th=[46336], 95.
     | 99.00th=[   97], 99.50th=[  109], 99.90th=[  127], 99. |	     | 99.00th=[72192], 99.50th=[73216], 99.90th=[77312], 99.
     | 99.99th=[  149]					      |	     | 99.99th=[86528]
Type=read,Size=512B,NumJobs=1,depth=64				Type=read,Size=512B,NumJobs=1,depth=64
   read: IOPS=370k, BW=181MiB/s (189MB/s)(1806MiB/10001msec)  |	   read: IOPS=344k, BW=168MiB/s (176MB/s)(1682MiB/10001msec)
    slat (nsec): min=1601, max=43074, avg=1927.33, stdev=232. |	    slat (nsec): min=1607, max=30650, avg=2093.56, stdev=233.
    clat (usec): min=56, max=378, avg=170.80, stdev= 4.52     |	    clat (usec): min=19, max=238, avg=183.42, stdev= 2.22
     lat (usec): min=57, max=380, avg=172.77, stdev= 4.55     |	     lat (usec): min=21, max=241, avg=185.55, stdev= 2.23
     |  1.00th=[  167],  5.00th=[  167], 10.00th=[  167], 20. |	     |  1.00th=[  180],  5.00th=[  182], 10.00th=[  182], 20.
     | 30.00th=[  169], 40.00th=[  169], 50.00th=[  169], 60. |	     | 30.00th=[  182], 40.00th=[  184], 50.00th=[  184], 60.
     | 70.00th=[  172], 80.00th=[  172], 90.00th=[  176], 95. |	     | 70.00th=[  184], 80.00th=[  186], 90.00th=[  186], 95.
     | 99.00th=[  184], 99.50th=[  194], 99.90th=[  221], 99. |	     | 99.00th=[  190], 99.50th=[  192], 99.90th=[  196], 99.
     | 99.99th=[  245]					      |	     | 99.99th=[  210]
Type=read,Size=512B,NumJobs=16,depth=1				Type=read,Size=512B,NumJobs=16,depth=1
   read: IOPS=917k, BW=448MiB/s (469MB/s)(4478MiB/10002msec)  |	   read: IOPS=907k, BW=443MiB/s (465MB/s)(4431MiB/10001msec)
    slat (nsec): min=2003, max=73317, avg=2440.64, stdev=588. |	    slat (nsec): min=1925, max=69082, avg=2505.79, stdev=869.
    clat (nsec): min=597, max=172124, avg=14561.65, stdev=210 |	    clat (nsec): min=476, max=112521, avg=14672.56, stdev=182
     lat (usec): min=11, max=174, avg=17.05, stdev= 2.20      |	     lat (usec): min=12, max=131, avg=17.23, stdev= 2.08
     |  1.00th=[12096],  5.00th=[13120], 10.00th=[13376], 20. |	     |  1.00th=[11584],  5.00th=[13248], 10.00th=[13376], 20.
     | 30.00th=[13888], 40.00th=[14016], 50.00th=[14144], 60. |	     | 30.00th=[13888], 40.00th=[14016], 50.00th=[14272], 60.
     | 70.00th=[14528], 80.00th=[14912], 90.00th=[15680], 95. |	     | 70.00th=[14784], 80.00th=[15296], 90.00th=[16512], 95.
     | 99.00th=[27008], 99.50th=[28288], 99.90th=[33536], 99. |	     | 99.00th=[22400], 99.50th=[24448], 99.90th=[32384], 99.
     | 99.99th=[35584]						     | 99.99th=[35584]
Type=read,Size=512B,NumJobs=16,depth=16				Type=read,Size=512B,NumJobs=16,depth=16
   read: IOPS=4627k, BW=2259MiB/s (2369MB/s)(22.1GiB/10002mse |	   read: IOPS=4055k, BW=1980MiB/s (2076MB/s)(19.3GiB/10003mse
    slat (nsec): min=1620, max=95729, avg=2408.03, stdev=483. |	    slat (nsec): min=1572, max=239521, avg=2836.25, stdev=282
    clat (usec): min=22, max=466, avg=52.52, stdev= 8.65      |	    clat (usec): min=9, max=616, avg=59.85, stdev=55.40
     lat (usec): min=24, max=474, avg=54.98, stdev= 8.95      |	     lat (usec): min=15, max=630, avg=62.74, stdev=57.98
     |  1.00th=[   42],  5.00th=[   43], 10.00th=[   44], 20. |	     |  1.00th=[   43],  5.00th=[   44], 10.00th=[   44], 20.
     | 30.00th=[   48], 40.00th=[   52], 50.00th=[   53], 60. |	     | 30.00th=[   45], 40.00th=[   46], 50.00th=[   46], 60.
     | 70.00th=[   57], 80.00th=[   59], 90.00th=[   61], 95. |	     | 70.00th=[   48], 80.00th=[   52], 90.00th=[   60], 95.
     | 99.00th=[   73], 99.50th=[   79], 99.90th=[  139], 99. |	     | 99.00th=[  326], 99.50th=[  343], 99.90th=[  367], 99.
     | 99.99th=[  281]					      |	     | 99.99th=[  519]
Type=read,Size=512B,NumJobs=16,depth=64				Type=read,Size=512B,NumJobs=16,depth=64
   read: IOPS=4878k, BW=2382MiB/s (2497MB/s)(23.3GiB/10001mse |	   read: IOPS=3451k, BW=1685MiB/s (1767MB/s)(16.5GiB/10001mse
    slat (nsec): min=1611, max=181900, avg=2357.92, stdev=159 |	    slat (nsec): min=1569, max=204726, avg=3322.74, stdev=274
    clat (usec): min=28, max=934, avg=207.19, stdev=38.57     |	    clat (usec): min=5, max=1543, avg=292.82, stdev=220.59
     lat (usec): min=29, max=944, avg=209.60, stdev=39.02     |	     lat (usec): min=14, max=1561, avg=296.20, stdev=223.03
     |  1.00th=[  172],  5.00th=[  174], 10.00th=[  174], 20. |	     |  1.00th=[  176],  5.00th=[  178], 10.00th=[  180], 20.
     | 30.00th=[  178], 40.00th=[  190], 50.00th=[  202], 60. |	     | 30.00th=[  188], 40.00th=[  227], 50.00th=[  243], 60.
     | 70.00th=[  215], 80.00th=[  233], 90.00th=[  251], 95. |	     | 70.00th=[  249], 80.00th=[  285], 90.00th=[  685], 95.
     | 99.00th=[  314], 99.50th=[  322], 99.90th=[  338], 99. |	     | 99.00th=[ 1270], 99.50th=[ 1319], 99.90th=[ 1385], 99.
     | 99.99th=[  783]					      |	     | 99.99th=[ 1450]
Type=read,Size=512B,NumJobs=56,depth=1				Type=read,Size=512B,NumJobs=56,depth=1
   read: IOPS=2311k, BW=1128MiB/s (1183MB/s)(11.0GiB/10002mse |	   read: IOPS=2351k, BW=1148MiB/s (1204MB/s)(11.2GiB/10002mse
    slat (usec): min=2, max=860, avg= 3.68, stdev= 2.24	      |	    slat (nsec): min=1744, max=502891, avg=3525.21, stdev=199
    clat (nsec): min=604, max=4342.7k, avg=19742.21, stdev=70 |	    clat (nsec): min=456, max=781278, avg=19563.02, stdev=418
     lat (usec): min=11, max=4366, avg=23.53, stdev= 7.52     |	     lat (usec): min=11, max=785, avg=23.18, stdev= 4.79
     |  1.00th=[   14],  5.00th=[   16], 10.00th=[   17], 20. |	     |  1.00th=[13120],  5.00th=[14912], 10.00th=[15936], 20.
     | 30.00th=[   18], 40.00th=[   19], 50.00th=[   20], 60. |	     | 30.00th=[17792], 40.00th=[18304], 50.00th=[19072], 60.
     | 70.00th=[   21], 80.00th=[   22], 90.00th=[   24], 95. |	     | 70.00th=[20608], 80.00th=[21632], 90.00th=[23680], 95.
     | 99.00th=[   35], 99.50th=[   39], 99.90th=[   49], 99. |	     | 99.00th=[31872], 99.50th=[35072], 99.90th=[41728], 99.
     | 99.99th=[  330]					      |	     | 99.99th=[56064]
Type=read,Size=512B,NumJobs=56,depth=16				Type=read,Size=512B,NumJobs=56,depth=16
   read: IOPS=8152k, BW=3981MiB/s (4174MB/s)(38.9GiB/10002mse |	   read: IOPS=7234k, BW=3532MiB/s (3704MB/s)(34.5GiB/10002mse
    slat (nsec): min=1658, max=2406.5k, avg=5086.35, stdev=72 |	    slat (nsec): min=1588, max=13073k, avg=5296.03, stdev=602
    clat (usec): min=10, max=3565, avg=104.02, stdev=51.20    |	    clat (usec): min=9, max=13311, avg=117.66, stdev=77.29
     lat (usec): min=23, max=3568, avg=109.21, stdev=53.58    |	     lat (usec): min=18, max=13322, avg=123.07, stdev=80.61
     |  1.00th=[   64],  5.00th=[   68], 10.00th=[   69], 20. |	     |  1.00th=[   63],  5.00th=[   67], 10.00th=[   69], 20.
     | 30.00th=[   73], 40.00th=[   74], 50.00th=[   76], 60. |	     | 30.00th=[   72], 40.00th=[   76], 50.00th=[   81], 60.
     | 70.00th=[   91], 80.00th=[  163], 90.00th=[  188], 95. |	     | 70.00th=[  108], 80.00th=[  184], 90.00th=[  247], 95.
     | 99.00th=[  221], 99.50th=[  231], 99.90th=[  359], 99. |	     | 99.00th=[  338], 99.50th=[  383], 99.90th=[  494], 99.
     | 99.99th=[  799]					      |	     | 99.99th=[  594]
Type=read,Size=512B,NumJobs=56,depth=64				Type=read,Size=512B,NumJobs=56,depth=64
   read: IOPS=8218k, BW=4013MiB/s (4208MB/s)(39.2GiB/10002mse |	   read: IOPS=7769k, BW=3793MiB/s (3978MB/s)(37.1GiB/10002mse
    slat (nsec): min=1673, max=1678.1k, avg=5096.07, stdev=74 |	    slat (nsec): min=1583, max=3785.4k, avg=5319.20, stdev=81
    clat (usec): min=34, max=2534, avg=430.13, stdev=202.15   |	    clat (usec): min=2, max=5212, avg=455.02, stdev=334.71
     lat (usec): min=61, max=2582, avg=435.33, stdev=204.56   |	     lat (usec): min=17, max=5214, avg=460.45, stdev=338.64
     |  1.00th=[  277],  5.00th=[  285], 10.00th=[  289], 20. |	     |  1.00th=[  265],  5.00th=[  273], 10.00th=[  281], 20.
     | 30.00th=[  302], 40.00th=[  306], 50.00th=[  310], 60. |	     | 30.00th=[  293], 40.00th=[  297], 50.00th=[  302], 60.
     | 70.00th=[  322], 80.00th=[  725], 90.00th=[  758], 95. |	     | 70.00th=[  310], 80.00th=[  437], 90.00th=[ 1074], 95.
     | 99.00th=[  816], 99.50th=[  824], 99.90th=[  848], 99. |	     | 99.00th=[ 1401], 99.50th=[ 1991], 99.90th=[ 2245], 99.
     | 99.99th=[  889]					      |	     | 99.99th=[ 2442]
Type=read,Size=4K,NumJobs=1,depth=1				Type=read,Size=4K,NumJobs=1,depth=1
   read: IOPS=52.8k, BW=206MiB/s (216MB/s)(2065MiB/10001msec) |	   read: IOPS=54.4k, BW=213MiB/s (223MB/s)(2126MiB/10001msec)
    slat (nsec): min=2019, max=42477, avg=2671.11, stdev=1827 |	    slat (nsec): min=2044, max=28781, avg=2556.03, stdev=1441
    clat (nsec): min=605, max=96758, avg=15743.80, stdev=5503 |	    clat (nsec): min=621, max=70566, avg=15359.54, stdev=5250
     lat (nsec): min=13458, max=99251, avg=18489.01, stdev=61 |	     lat (nsec): min=13302, max=73823, avg=17966.25, stdev=56
     |  1.00th=[13248],  5.00th=[13376], 10.00th=[13376], 20. |	     |  1.00th=[13120],  5.00th=[13248], 10.00th=[13248], 20.
     | 30.00th=[13504], 40.00th=[13632], 50.00th=[13632], 60. |	     | 30.00th=[13376], 40.00th=[13376], 50.00th=[13376], 60.
     | 70.00th=[13888], 80.00th=[14912], 90.00th=[19584], 95. |	     | 70.00th=[13632], 80.00th=[14144], 90.00th=[19328], 95.
     | 99.00th=[36608], 99.50th=[38656], 99.90th=[48384], 99. |	     | 99.00th=[35584], 99.50th=[36608], 99.90th=[45312], 99.
     | 99.99th=[69120]					      |	     | 99.99th=[53504]
Type=read,Size=4K,NumJobs=1,depth=16				Type=read,Size=4K,NumJobs=1,depth=16
   read: IOPS=306k, BW=1197MiB/s (1255MB/s)(11.7GiB/10001msec |	   read: IOPS=334k, BW=1306MiB/s (1370MB/s)(12.8GiB/10001msec
    slat (nsec): min=1621, max=44989, avg=2060.53, stdev=583. |	    slat (nsec): min=1696, max=45538, avg=2168.76, stdev=208.
    clat (usec): min=36, max=179, avg=49.85, stdev=11.62      |	    clat (usec): min=18, max=148, avg=45.39, stdev= 1.47
     lat (usec): min=38, max=181, avg=51.95, stdev=11.64      |	     lat (usec): min=20, max=192, avg=47.59, stdev= 1.48
     |  1.00th=[   42],  5.00th=[   42], 10.00th=[   43], 20. |	     |  1.00th=[43776],  5.00th=[44288], 10.00th=[44288], 20.
     | 30.00th=[   43], 40.00th=[   44], 50.00th=[   44], 60. |	     | 30.00th=[44800], 40.00th=[45312], 50.00th=[45312], 60.
     | 70.00th=[   53], 80.00th=[   58], 90.00th=[   64], 95. |	     | 70.00th=[45312], 80.00th=[45824], 90.00th=[45824], 95.
     | 99.00th=[   95], 99.50th=[  103], 99.90th=[  127], 99. |	     | 99.00th=[49408], 99.50th=[54528], 99.90th=[64256], 99.
     | 99.99th=[  149]					      |	     | 99.99th=[70144]
Type=read,Size=4K,NumJobs=1,depth=64				Type=read,Size=4K,NumJobs=1,depth=64
   read: IOPS=353k, BW=1381MiB/s (1448MB/s)(13.5GiB/10001msec |	   read: IOPS=338k, BW=1319MiB/s (1383MB/s)(12.9GiB/10001msec
    slat (nsec): min=1636, max=31675, avg=2040.73, stdev=224. |	    slat (nsec): min=1621, max=52632, avg=2145.29, stdev=209.
    clat (usec): min=44, max=244, avg=178.71, stdev= 2.85     |	    clat (usec): min=36, max=236, avg=187.02, stdev= 2.05
     lat (usec): min=46, max=246, avg=180.79, stdev= 2.87     |	     lat (usec): min=38, max=238, avg=189.20, stdev= 2.06
     |  1.00th=[  176],  5.00th=[  176], 10.00th=[  178], 20. |	     |  1.00th=[  184],  5.00th=[  184], 10.00th=[  186], 20.
     | 30.00th=[  178], 40.00th=[  178], 50.00th=[  180], 60. |	     | 30.00th=[  186], 40.00th=[  188], 50.00th=[  188], 60.
     | 70.00th=[  180], 80.00th=[  180], 90.00th=[  182], 95. |	     | 70.00th=[  188], 80.00th=[  188], 90.00th=[  190], 95.
     | 99.00th=[  190], 99.50th=[  194], 99.90th=[  202], 99. |	     | 99.00th=[  192], 99.50th=[  196], 99.90th=[  200], 99.
     | 99.99th=[  210]					      |	     | 99.99th=[  217]
Type=read,Size=4K,NumJobs=16,depth=1				Type=read,Size=4K,NumJobs=16,depth=1
   read: IOPS=878k, BW=3430MiB/s (3597MB/s)(33.5GiB/10001msec |	   read: IOPS=869k, BW=3396MiB/s (3561MB/s)(33.2GiB/10002msec
    slat (nsec): min=1947, max=53618, avg=2438.66, stdev=659. |	    slat (nsec): min=1890, max=70947, avg=2497.58, stdev=813.
    clat (nsec): min=593, max=420818, avg=15328.26, stdev=153 |	    clat (nsec): min=590, max=79209, avg=15456.14, stdev=1968
     lat (usec): min=13, max=423, avg=17.82, stdev= 1.67      |	     lat (usec): min=12, max=110, avg=18.01, stdev= 2.18
     |  1.00th=[12736],  5.00th=[14144], 10.00th=[14272], 20. |	     |  1.00th=[12352],  5.00th=[13888], 10.00th=[14144], 20.
     | 30.00th=[14656], 40.00th=[14912], 50.00th=[15040], 60. |	     | 30.00th=[14528], 40.00th=[14784], 50.00th=[15040], 60.
     | 70.00th=[15424], 80.00th=[15936], 90.00th=[16768], 95. |	     | 70.00th=[15552], 80.00th=[16192], 90.00th=[17280], 95.
     | 99.00th=[20352], 99.50th=[22912], 99.90th=[27264], 99. |	     | 99.00th=[23936], 99.50th=[25728], 99.90th=[33536], 99.
     | 99.99th=[70144]					      |	     | 99.99th=[36096]
Type=read,Size=4K,NumJobs=16,depth=16				Type=read,Size=4K,NumJobs=16,depth=16
   read: IOPS=2677k, BW=10.2GiB/s (10.0GB/s)(102GiB/10001msec |	   read: IOPS=1843k, BW=7201MiB/s (7551MB/s)(70.3GiB/10002mse
    slat (nsec): min=1634, max=169095, avg=2282.01, stdev=978 |	    slat (nsec): min=1582, max=104435, avg=3170.92, stdev=170
    clat (usec): min=23, max=416, avg=92.99, stdev=14.19      |	    clat (usec): min=40, max=554, avg=135.18, stdev=32.57
     lat (usec): min=29, max=419, avg=95.32, stdev=14.22      |	     lat (usec): min=42, max=557, avg=138.42, stdev=32.59
     |  1.00th=[   75],  5.00th=[   78], 10.00th=[   80], 20. |	     |  1.00th=[   68],  5.00th=[   92], 10.00th=[  104], 20.
     | 30.00th=[   87], 40.00th=[   90], 50.00th=[   94], 60. |	     | 30.00th=[  125], 40.00th=[  130], 50.00th=[  135], 60.
     | 70.00th=[   98], 80.00th=[   99], 90.00th=[  101], 95. |	     | 70.00th=[  141], 80.00th=[  147], 90.00th=[  161], 95.
     | 99.00th=[  169], 99.50th=[  174], 99.90th=[  192], 99. |	     | 99.00th=[  262], 99.50th=[  306], 99.90th=[  392], 99.
     | 99.99th=[  273]					      |	     | 99.99th=[  461]
Type=read,Size=4K,NumJobs=16,depth=64				Type=read,Size=4K,NumJobs=16,depth=64
   read: IOPS=2673k, BW=10.2GiB/s (10.9GB/s)(102GiB/10001msec |	   read: IOPS=1917k, BW=7487MiB/s (7850MB/s)(73.1GiB/10001mse
    slat (nsec): min=1620, max=148710, avg=2532.78, stdev=332 |	    slat (nsec): min=1566, max=653412, avg=3347.14, stdev=385
    clat (usec): min=19, max=1229, avg=380.20, stdev=82.66    |	    clat (usec): min=43, max=1445, avg=530.39, stdev=127.84
     lat (usec): min=23, max=1232, avg=382.77, stdev=83.63    |	     lat (usec): min=45, max=1619, avg=533.81, stdev=128.49
     |  1.00th=[  318],  5.00th=[  330], 10.00th=[  343], 20. |	     |  1.00th=[  188],  5.00th=[  351], 10.00th=[  420], 20.
     | 30.00th=[  351], 40.00th=[  355], 50.00th=[  363], 60. |	     | 30.00th=[  486], 40.00th=[  502], 50.00th=[  519], 60.
     | 70.00th=[  371], 80.00th=[  383], 90.00th=[  388], 95. |	     | 70.00th=[  553], 80.00th=[  578], 90.00th=[  644], 95.
     | 99.00th=[  725], 99.50th=[  742], 99.90th=[  750], 99. |	     | 99.00th=[ 1012], 99.50th=[ 1037], 99.90th=[ 1090], 99.
     | 99.99th=[  857]					      |	     | 99.99th=[ 1205]
Type=read,Size=4K,NumJobs=56,depth=1				Type=read,Size=4K,NumJobs=56,depth=1
   read: IOPS=1606k, BW=6273MiB/s (6578MB/s)(61.3GiB/10002mse |	   read: IOPS=1571k, BW=6138MiB/s (6436MB/s)(59.9GiB/10001mse
    slat (nsec): min=1955, max=745694, avg=2980.95, stdev=139 |	    slat (nsec): min=1693, max=505053, avg=2771.65, stdev=110
    clat (nsec): min=633, max=1012.2k, avg=31216.71, stdev=89 |	    clat (nsec): min=479, max=1348.1k, avg=32311.24, stdev=36
     lat (usec): min=13, max=1015, avg=34.28, stdev= 9.05     |	     lat (usec): min=15, max=1350, avg=35.15, stdev= 3.81
     |  1.00th=[   22],  5.00th=[   27], 10.00th=[   29], 20. |	     |  1.00th=[26752],  5.00th=[28800], 10.00th=[29568], 20.
     | 30.00th=[   30], 40.00th=[   31], 50.00th=[   31], 60. |	     | 30.00th=[30848], 40.00th=[31360], 50.00th=[31872], 60.
     | 70.00th=[   32], 80.00th=[   33], 90.00th=[   35], 95. |	     | 70.00th=[33024], 80.00th=[33536], 90.00th=[35584], 95.
     | 99.00th=[   43], 99.50th=[   46], 99.90th=[   70], 99. |	     | 99.00th=[42240], 99.50th=[43776], 99.90th=[49920], 99.
     | 99.99th=[  474]					      |	     | 99.99th=[59136]
Type=read,Size=4K,NumJobs=56,depth=16				Type=read,Size=4K,NumJobs=56,depth=16
   read: IOPS=2591k, BW=9.88GiB/s (10.6GB/s)(98.9GiB/10002mse |	   read: IOPS=2009k, BW=7846MiB/s (8227MB/s)(76.6GiB/10001mse
    slat (nsec): min=1655, max=626114, avg=3005.44, stdev=179 |	    slat (nsec): min=1566, max=379585, avg=3204.59, stdev=230
    clat (usec): min=25, max=1308, avg=342.16, stdev=55.44    |	    clat (usec): min=44, max=1761, avg=442.22, stdev=244.27
     lat (usec): min=27, max=1311, avg=345.24, stdev=55.45    |	     lat (usec): min=46, max=1764, avg=445.50, stdev=244.24
     |  1.00th=[  151],  5.00th=[  192], 10.00th=[  318], 20. |	     |  1.00th=[  161],  5.00th=[  194], 10.00th=[  215], 20.
     | 30.00th=[  330], 40.00th=[  338], 50.00th=[  343], 60. |	     | 30.00th=[  273], 40.00th=[  306], 50.00th=[  351], 60.
     | 70.00th=[  359], 80.00th=[  367], 90.00th=[  388], 95. |	     | 70.00th=[  506], 80.00th=[  644], 90.00th=[  832], 95.
     | 99.00th=[  478], 99.50th=[  506], 99.90th=[  635], 99. |	     | 99.00th=[ 1156], 99.50th=[ 1221], 99.90th=[ 1336], 99.
     | 99.99th=[  955]					      |	     | 99.99th=[ 1450]
Type=read,Size=4K,NumJobs=56,depth=64				Type=read,Size=4K,NumJobs=56,depth=64
   read: IOPS=2652k, BW=10.1GiB/s (10.9GB/s)(101GiB/10003msec |	   read: IOPS=1801k, BW=7036MiB/s (7378MB/s)(68.7GiB/10002mse
    slat (nsec): min=1638, max=1266.7k, avg=14402.06, stdev=5 |	    slat (nsec): min=1586, max=2410.2k, avg=20136.27, stdev=8
    clat (usec): min=23, max=7119, avg=1336.05, stdev=712.32  |	    clat (usec): min=58, max=8215, avg=1968.75, stdev=947.22
     lat (usec): min=30, max=7224, avg=1350.51, stdev=725.25  |	     lat (usec): min=60, max=8632, avg=1988.96, stdev=962.84
     |  1.00th=[  570],  5.00th=[  644], 10.00th=[  652], 20. |	     |  1.00th=[  449],  5.00th=[  734], 10.00th=[  865], 20.
     | 30.00th=[  685], 40.00th=[ 1221], 50.00th=[ 1303], 60. |	     | 30.00th=[ 1418], 40.00th=[ 1729], 50.00th=[ 1909], 60.
     | 70.00th=[ 1401], 80.00th=[ 1926], 90.00th=[ 2409], 95. |	     | 70.00th=[ 2212], 80.00th=[ 2507], 90.00th=[ 3326], 95.
     | 99.00th=[ 3359], 99.50th=[ 3589], 99.90th=[ 4047], 99. |	     | 99.00th=[ 4817], 99.50th=[ 5080], 99.90th=[ 5735], 99.
     | 99.99th=[ 5080]					      |	     | 99.99th=[ 6587]
Type=read,Size=64K,NumJobs=1,depth=1				Type=read,Size=64K,NumJobs=1,depth=1
   read: IOPS=29.6k, BW=1849MiB/s (1939MB/s)(18.1GiB/10001mse |	   read: IOPS=29.8k, BW=1864MiB/s (1955MB/s)(18.2GiB/10001mse
    slat (nsec): min=4727, max=45982, avg=6068.77, stdev=2849 |	    slat (nsec): min=4768, max=95484, avg=6119.74, stdev=2779
    clat (usec): min=3, max=110, avg=27.22, stdev= 7.91	      |	    clat (nsec): min=1555, max=75556, avg=26881.73, stdev=781
     lat (usec): min=25, max=122, avg=33.34, stdev= 9.17      |	     lat (usec): min=25, max=102, avg=33.05, stdev= 9.07
     |  1.00th=[21888],  5.00th=[21888], 10.00th=[21888], 20. |	     |  1.00th=[21632],  5.00th=[21632], 10.00th=[21632], 20.
     | 30.00th=[22144], 40.00th=[22400], 50.00th=[22400], 60. |	     | 30.00th=[21888], 40.00th=[22144], 50.00th=[22144], 60.
     | 70.00th=[28288], 80.00th=[35072], 90.00th=[41216], 95. |	     | 70.00th=[28032], 80.00th=[34048], 90.00th=[40704], 95.
     | 99.00th=[47872], 99.50th=[50432], 99.90th=[63744], 99. |	     | 99.00th=[47360], 99.50th=[49920], 99.90th=[59648], 99.
     | 99.99th=[78336]					      |	     | 99.99th=[70144]
Type=read,Size=64K,NumJobs=1,depth=16				Type=read,Size=64K,NumJobs=1,depth=16
   read: IOPS=168k, BW=10.3GiB/s (11.0GB/s)(103GiB/10001msec) |	   read: IOPS=172k, BW=10.5GiB/s (11.3GB/s)(105GiB/10001msec)
    slat (nsec): min=3795, max=48145, avg=4685.87, stdev=473. |	    slat (nsec): min=3756, max=65673, avg=4930.28, stdev=334.
    clat (usec): min=78, max=259, avg=90.18, stdev= 7.63      |	    clat (usec): min=83, max=243, avg=87.86, stdev= 2.04
     lat (usec): min=83, max=300, avg=94.90, stdev= 7.60      |	     lat (usec): min=87, max=310, avg=92.83, stdev= 2.05
     |  1.00th=[   82],  5.00th=[   82], 10.00th=[   85], 20. |	     |  1.00th=[   86],  5.00th=[   87], 10.00th=[   87], 20.
     | 30.00th=[   88], 40.00th=[   88], 50.00th=[   89], 60. |	     | 30.00th=[   87], 40.00th=[   88], 50.00th=[   88], 60.
     | 70.00th=[   92], 80.00th=[   93], 90.00th=[   96], 95. |	     | 70.00th=[   88], 80.00th=[   89], 90.00th=[   91], 95.
     | 99.00th=[  127], 99.50th=[  137], 99.90th=[  174], 99. |	     | 99.00th=[   96], 99.50th=[   99], 99.90th=[  108], 99.
     | 99.99th=[  196]					      |	     | 99.99th=[  114]
Type=read,Size=64K,NumJobs=1,depth=64				Type=read,Size=64K,NumJobs=1,depth=64
   read: IOPS=171k, BW=10.4GiB/s (11.2GB/s)(104GiB/10001msec) |	   read: IOPS=168k, BW=10.2GiB/s (10.0GB/s)(102GiB/10001msec)
    slat (usec): min=2, max=384, avg= 3.57, stdev= 1.13	      |	    slat (usec): min=2, max=390, avg= 4.15, stdev= 1.56
    clat (usec): min=50, max=983, avg=370.03, stdev=11.50     |	    clat (usec): min=37, max=947, avg=376.95, stdev=15.70
     lat (usec): min=53, max=1059, avg=373.64, stdev=11.40    |	     lat (usec): min=42, max=988, avg=381.15, stdev=16.25
     |  1.00th=[  351],  5.00th=[  359], 10.00th=[  363], 20. |	     |  1.00th=[  363],  5.00th=[  363], 10.00th=[  367], 20.
     | 30.00th=[  367], 40.00th=[  367], 50.00th=[  371], 60.	     | 30.00th=[  367], 40.00th=[  367], 50.00th=[  371], 60.
     | 70.00th=[  371], 80.00th=[  375], 90.00th=[  379], 95. |	     | 70.00th=[  388], 80.00th=[  392], 90.00th=[  400], 95.
     | 99.00th=[  400], 99.50th=[  408], 99.90th=[  433], 99. |	     | 99.00th=[  412], 99.50th=[  416], 99.90th=[  429], 99.
     | 99.99th=[  644]					      |	     | 99.99th=[  668]
Type=read,Size=64K,NumJobs=16,depth=1				Type=read,Size=64K,NumJobs=16,depth=1
   read: IOPS=171k, BW=10.4GiB/s (11.2GB/s)(104GiB/10001msec) |	   read: IOPS=172k, BW=10.5GiB/s (11.3GB/s)(105GiB/10002msec)
    slat (nsec): min=4077, max=68679, avg=5394.03, stdev=2968 |	    slat (nsec): min=3480, max=74756, avg=5273.21, stdev=2961
    clat (nsec): min=686, max=194443, avg=87548.33, stdev=162 |	    clat (nsec): min=616, max=453786, avg=87271.90, stdev=161
     lat (usec): min=27, max=208, avg=93.00, stdev=16.25      |	     lat (usec): min=26, max=458, avg=92.60, stdev=16.20
     |  1.00th=[   42],  5.00th=[   61], 10.00th=[   71], 20. |	     |  1.00th=[   37],  5.00th=[   57], 10.00th=[   68], 20.
     | 30.00th=[   81], 40.00th=[   84], 50.00th=[   87], 60. |	     | 30.00th=[   83], 40.00th=[   87], 50.00th=[   90], 60.
     | 70.00th=[   94], 80.00th=[   99], 90.00th=[  111], 95. |	     | 70.00th=[   95], 80.00th=[   98], 90.00th=[  104], 95.
     | 99.00th=[  127], 99.50th=[  133], 99.90th=[  143], 99. |	     | 99.00th=[  126], 99.50th=[  130], 99.90th=[  143], 99.
     | 99.99th=[  161]						     | 99.99th=[  161]
Type=read,Size=64K,NumJobs=16,depth=16				Type=read,Size=64K,NumJobs=16,depth=16
   read: IOPS=175k, BW=10.7GiB/s (11.4GB/s)(107GiB/10002msec) |	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10002msec)
    slat (nsec): min=4066, max=75561, avg=5259.83, stdev=1080 |	    slat (nsec): min=3866, max=57337, avg=5204.47, stdev=1220
    clat (usec): min=118, max=3688, avg=1460.39, stdev=456.30 |	    clat (usec): min=116, max=3985, avg=1453.77, stdev=349.92
     lat (usec): min=132, max=3693, avg=1465.70, stdev=456.22 |	     lat (usec): min=128, max=3990, avg=1459.03, stdev=349.93
     |  1.00th=[ 1172],  5.00th=[ 1172], 10.00th=[ 1172], 20. |	     |  1.00th=[ 1254],  5.00th=[ 1270], 10.00th=[ 1270], 20.
     | 30.00th=[ 1205], 40.00th=[ 1270], 50.00th=[ 1270], 60. |	     | 30.00th=[ 1352], 40.00th=[ 1352], 50.00th=[ 1352], 60.
     | 70.00th=[ 1352], 80.00th=[ 1401], 90.00th=[ 2376], 95. |	     | 70.00th=[ 1369], 80.00th=[ 1401], 90.00th=[ 1467], 95.
     | 99.00th=[ 2606], 99.50th=[ 2737], 99.90th=[ 2769], 99. |	     | 99.00th=[ 2737], 99.50th=[ 2769], 99.90th=[ 2769], 99.
     | 99.99th=[ 2802]						     | 99.99th=[ 2802]
Type=read,Size=64K,NumJobs=16,depth=64				Type=read,Size=64K,NumJobs=16,depth=64
   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10006msec)	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10006msec)
    slat (usec): min=2, max=890, avg= 4.27, stdev= 3.77	      |	    slat (usec): min=2, max=1185, avg= 9.71, stdev=60.22
    clat (usec): min=108, max=11645, avg=5850.95, stdev=186.5 |	    clat (usec): min=116, max=16490, avg=5824.81, stdev=1345.
     lat (usec): min=113, max=12034, avg=5855.28, stdev=186.3 |	     lat (usec): min=133, max=16494, avg=5834.58, stdev=1367.
     |  1.00th=[ 5800],  5.00th=[ 5800], 10.00th=[ 5800], 20. |	     |  1.00th=[ 5407],  5.00th=[ 5407], 10.00th=[ 5473], 20.
     | 30.00th=[ 5866], 40.00th=[ 5866], 50.00th=[ 5866], 60. |	     | 30.00th=[ 5473], 40.00th=[ 5473], 50.00th=[ 5473], 60.
     | 70.00th=[ 5866], 80.00th=[ 5866], 90.00th=[ 5866], 95. |	     | 70.00th=[ 5473], 80.00th=[ 5473], 90.00th=[ 5538], 95.
     | 99.00th=[ 5932], 99.50th=[ 6063], 99.90th=[ 6980], 99. |	     | 99.00th=[10945], 99.50th=[10945], 99.90th=[11076], 99.
     | 99.99th=[11207]					      |	     | 99.99th=[11338]
Type=read,Size=64K,NumJobs=56,depth=1				Type=read,Size=64K,NumJobs=56,depth=1
   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10001msec) |	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10002msec)
    slat (usec): min=4, max=111, avg= 4.93, stdev= 1.11	      |	    slat (nsec): min=3801, max=63985, avg=4896.45, stdev=1489
    clat (usec): min=41, max=531, avg=314.40, stdev=75.66     |	    clat (usec): min=29, max=550, avg=314.45, stdev=76.21
     lat (usec): min=50, max=535, avg=319.39, stdev=75.67     |	     lat (usec): min=35, max=555, avg=319.41, stdev=76.15
     |  1.00th=[  174],  5.00th=[  176], 10.00th=[  178], 20. |	     |  1.00th=[  174],  5.00th=[  178], 10.00th=[  194], 20.
     | 30.00th=[  306], 40.00th=[  306], 50.00th=[  306], 60. |	     | 30.00th=[  302], 40.00th=[  306], 50.00th=[  306], 60.
     | 70.00th=[  310], 80.00th=[  318], 90.00th=[  461], 95. |	     | 70.00th=[  310], 80.00th=[  322], 90.00th=[  457], 95.
     | 99.00th=[  478], 99.50th=[  482], 99.90th=[  490], 99. |	     | 99.00th=[  474], 99.50th=[  482], 99.90th=[  490], 99.
     | 99.99th=[  502]					      |	     | 99.99th=[  506]
Type=read,Size=64K,NumJobs=56,depth=16				Type=read,Size=64K,NumJobs=56,depth=16
   read: IOPS=175k, BW=10.7GiB/s (11.4GB/s)(107GiB/10006msec) |	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10006msec)
    slat (usec): min=4, max=176, avg= 5.48, stdev= 2.61	      |	    slat (usec): min=3, max=251, avg= 5.40, stdev= 2.34
    clat (usec): min=718, max=14322, avg=5122.23, stdev=2028. |	    clat (usec): min=236, max=14486, avg=5106.35, stdev=1241.
     lat (usec): min=744, max=14328, avg=5127.78, stdev=2028. |	     lat (usec): min=265, max=14491, avg=5111.80, stdev=1241.
     |  1.00th=[ 2343],  5.00th=[ 2376], 10.00th=[ 2442], 20. |	     |  1.00th=[ 2442],  5.00th=[ 2540], 10.00th=[ 4883], 20.
     | 30.00th=[ 4752], 40.00th=[ 4752], 50.00th=[ 4948], 60. |	     | 30.00th=[ 4948], 40.00th=[ 5080], 50.00th=[ 5080], 60.
     | 70.00th=[ 5080], 80.00th=[ 5145], 90.00th=[ 9503], 95. |	     | 70.00th=[ 5080], 80.00th=[ 5145], 90.00th=[ 7373], 95.
     | 99.00th=[ 9896], 99.50th=[ 9896], 99.90th=[10028], 99. |	     | 99.00th=[ 9896], 99.50th=[ 9896], 99.90th=[ 9896], 99.
     | 99.99th=[10028]						     | 99.99th=[10028]
Type=read,Size=64K,NumJobs=56,depth=64				Type=read,Size=64K,NumJobs=56,depth=64
   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10017msec) |	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10018msec)
    slat (usec): min=2, max=9087, avg=220.21, stdev=768.63    |	    slat (usec): min=2, max=10184, avg=189.22, stdev=637.78
    clat (usec): min=172, max=64296, avg=20233.61, stdev=8148 |	    clat (usec): min=347, max=61952, avg=20252.15, stdev=7056
     lat (usec): min=186, max=64319, avg=20453.88, stdev=8308 |	     lat (usec): min=351, max=65141, avg=20441.44, stdev=7203
     |  1.00th=[ 9110],  5.00th=[ 9110], 10.00th=[ 9110], 20. |	     |  1.00th=[ 9503],  5.00th=[ 9765], 10.00th=[ 9896], 20.
     | 30.00th=[18220], 40.00th=[18744], 50.00th=[19006], 60. |	     | 30.00th=[19530], 40.00th=[19530], 50.00th=[19792], 60.
     | 70.00th=[27395], 80.00th=[27395], 90.00th=[28705], 95. |	     | 70.00th=[20317], 80.00th=[20579], 90.00th=[29492], 95.
     | 99.00th=[38011], 99.50th=[40109], 99.90th=[44303], 99. |	     | 99.00th=[41157], 99.50th=[58983], 99.90th=[59507], 99.
     | 99.99th=[54789]					      |	     | 99.99th=[61080]
Type=write,Size=512B,NumJobs=1,depth=1				Type=write,Size=512B,NumJobs=1,depth=1
  write: IOPS=57.7k, BW=28.2MiB/s (29.6MB/s)(282MiB/10001msec |	  write: IOPS=56.0k, BW=27.4MiB/s (28.7MB/s)(274MiB/10001msec
    slat (nsec): min=2139, max=32784, avg=2527.92, stdev=1359 |	    slat (nsec): min=2112, max=43957, avg=2561.65, stdev=1423
    clat (nsec): min=624, max=106340, avg=14332.09, stdev=474 |	    clat (nsec): min=633, max=60031, avg=14812.68, stdev=5638
     lat (usec): min=12, max=108, avg=16.90, stdev= 5.16      |	     lat (nsec): min=12369, max=72124, avg=17416.39, stdev=60
     |  1.00th=[12480],  5.00th=[12480], 10.00th=[12608], 20. |	     |  1.00th=[12352],  5.00th=[12352], 10.00th=[12480], 20.
     | 30.00th=[12608], 40.00th=[12736], 50.00th=[12736], 60. |	     | 30.00th=[12480], 40.00th=[12608], 50.00th=[12608], 60.
     | 70.00th=[12864], 80.00th=[13376], 90.00th=[16512], 95. |	     | 70.00th=[12864], 80.00th=[14656], 90.00th=[20096], 95.
     | 99.00th=[34560], 99.50th=[35584], 99.90th=[42240], 99. |	     | 99.00th=[35072], 99.50th=[36608], 99.90th=[47360], 99.
     | 99.99th=[53504]					      |	     | 99.99th=[54016]
Type=write,Size=512B,NumJobs=1,depth=16				Type=write,Size=512B,NumJobs=1,depth=16
  write: IOPS=306k, BW=149MiB/s (156MB/s)(1492MiB/10001msec)  |	  write: IOPS=328k, BW=160MiB/s (168MB/s)(1601MiB/10001msec)
    slat (nsec): min=1650, max=31443, avg=2000.75, stdev=584. |	    slat (nsec): min=1641, max=32941, avg=2199.97, stdev=236.
    clat (usec): min=38, max=193, avg=50.03, stdev=12.10      |	    clat (usec): min=14, max=113, avg=46.29, stdev= 1.27
     lat (usec): min=40, max=195, avg=52.07, stdev=12.12      |	     lat (usec): min=16, max=115, avg=48.52, stdev= 1.28
     |  1.00th=[   41],  5.00th=[   41], 10.00th=[   41], 20. |	     |  1.00th=[43776],  5.00th=[44800], 10.00th=[45312], 20.
     | 30.00th=[   42], 40.00th=[   43], 50.00th=[   45], 60. |	     | 30.00th=[45824], 40.00th=[45824], 50.00th=[46336], 60.
     | 70.00th=[   52], 80.00th=[   59], 90.00th=[   69], 95. |	     | 70.00th=[46336], 80.00th=[46848], 90.00th=[47360], 95.
     | 99.00th=[   93], 99.50th=[  100], 99.90th=[  126], 99. |	     | 99.00th=[50432], 99.50th=[53504], 99.90th=[57600], 99.
     | 99.99th=[  145]					      |	     | 99.99th=[63744]
Type=write,Size=512B,NumJobs=1,depth=64				Type=write,Size=512B,NumJobs=1,depth=64
  write: IOPS=363k, BW=177MiB/s (186MB/s)(1771MiB/10001msec)  |	  write: IOPS=337k, BW=164MiB/s (172MB/s)(1644MiB/10001msec)
    slat (nsec): min=1644, max=41021, avg=1961.04, stdev=238. |	    slat (nsec): min=1642, max=43437, avg=2146.30, stdev=234.
    clat (usec): min=71, max=241, avg=174.18, stdev= 2.48     |	    clat (usec): min=16, max=235, avg=187.62, stdev= 2.46
     lat (usec): min=73, max=243, avg=176.18, stdev= 2.50     |	     lat (usec): min=18, max=237, avg=189.80, stdev= 2.48
     |  1.00th=[  172],  5.00th=[  172], 10.00th=[  172], 20. |	     |  1.00th=[  184],  5.00th=[  184], 10.00th=[  186], 20.
     | 30.00th=[  174], 40.00th=[  174], 50.00th=[  174], 60. |	     | 30.00th=[  186], 40.00th=[  188], 50.00th=[  188], 60.
     | 70.00th=[  176], 80.00th=[  176], 90.00th=[  178], 95. |	     | 70.00th=[  190], 80.00th=[  190], 90.00th=[  192], 95.
     | 99.00th=[  186], 99.50th=[  190], 99.90th=[  192], 99. |	     | 99.00th=[  194], 99.50th=[  196], 99.90th=[  200], 99.
     | 99.99th=[  198]					      |	     | 99.99th=[  204]
Type=write,Size=512B,NumJobs=16,depth=1				Type=write,Size=512B,NumJobs=16,depth=1
  write: IOPS=938k, BW=458MiB/s (480MB/s)(4582MiB/10002msec)  |	  write: IOPS=914k, BW=446MiB/s (468MB/s)(4462MiB/10001msec)
    slat (nsec): min=2061, max=61336, avg=2520.86, stdev=499. |	    slat (nsec): min=1818, max=83376, avg=2566.34, stdev=824.
    clat (nsec): min=595, max=164756, avg=14051.16, stdev=104 |	    clat (nsec): min=465, max=343253, avg=14459.07, stdev=181
     lat (usec): min=11, max=168, avg=16.62, stdev= 1.19      |	     lat (usec): min=11, max=345, avg=17.08, stdev= 2.09
     |  1.00th=[11712],  5.00th=[12864], 10.00th=[13248], 20. |	     |  1.00th=[12224],  5.00th=[12992], 10.00th=[13248], 20.
     | 30.00th=[13632], 40.00th=[13760], 50.00th=[13888], 60. |	     | 30.00th=[13632], 40.00th=[13760], 50.00th=[14016], 60.
     | 70.00th=[14272], 80.00th=[14528], 90.00th=[15040], 95. |	     | 70.00th=[14528], 80.00th=[15040], 90.00th=[16320], 95.
     | 99.00th=[17280], 99.50th=[18048], 99.90th=[23680], 99. |	     | 99.00th=[21888], 99.50th=[23936], 99.90th=[32384], 99.
     | 99.99th=[29568]					      |	     | 99.99th=[35584]
Type=write,Size=512B,NumJobs=16,depth=16			Type=write,Size=512B,NumJobs=16,depth=16
  write: IOPS=4631k, BW=2261MiB/s (2371MB/s)(22.1GiB/10001mse |	  write: IOPS=3218k, BW=1571MiB/s (1648MB/s)(15.3GiB/10001mse
    slat (nsec): min=1666, max=183449, avg=2373.16, stdev=548 |	    slat (nsec): min=1621, max=154997, avg=3548.60, stdev=323
    clat (usec): min=18, max=440, avg=52.52, stdev= 9.27      |	    clat (usec): min=14, max=473, avg=75.44, stdev=65.86
     lat (usec): min=20, max=444, avg=54.94, stdev= 9.57      |	     lat (usec): min=17, max=489, avg=79.05, stdev=68.82
     |  1.00th=[   43],  5.00th=[   44], 10.00th=[   44], 20. |	     |  1.00th=[   45],  5.00th=[   45], 10.00th=[   46], 20.
     | 30.00th=[   48], 40.00th=[   51], 50.00th=[   52], 60. |	     | 30.00th=[   47], 40.00th=[   48], 50.00th=[   52], 60.
     | 70.00th=[   56], 80.00th=[   59], 90.00th=[   62], 95. |	     | 70.00th=[   61], 80.00th=[   64], 90.00th=[  178], 95.
     | 99.00th=[   74], 99.50th=[   79], 99.90th=[   89], 99. |	     | 99.00th=[  326], 99.50th=[  334], 99.90th=[  351], 99.
     | 99.99th=[  355]					      |	     | 99.99th=[  367]
Type=write,Size=512B,NumJobs=16,depth=64			Type=write,Size=512B,NumJobs=16,depth=64
  write: IOPS=5021k, BW=2452MiB/s (2571MB/s)(23.9GiB/10002mse |	  write: IOPS=3973k, BW=1940MiB/s (2034MB/s)(18.0GiB/10002mse
    slat (nsec): min=1670, max=80008, avg=2293.73, stdev=923. |	    slat (nsec): min=1627, max=171980, avg=2926.70, stdev=209
    clat (usec): min=24, max=1715, avg=201.23, stdev=28.44    |	    clat (usec): min=10, max=1606, avg=254.29, stdev=176.36
     lat (usec): min=26, max=1733, avg=203.56, stdev=28.76    |	     lat (usec): min=16, max=1615, avg=257.27, stdev=178.32
     |  1.00th=[  176],  5.00th=[  178], 10.00th=[  180], 20. |	     |  1.00th=[  184],  5.00th=[  186], 10.00th=[  188], 20.
     | 30.00th=[  182], 40.00th=[  184], 50.00th=[  208], 60. |	     | 30.00th=[  194], 40.00th=[  202], 50.00th=[  237], 60.
     | 70.00th=[  212], 80.00th=[  215], 90.00th=[  219], 95. |	     | 70.00th=[  255], 80.00th=[  260], 90.00th=[  293], 95.
     | 99.00th=[  293], 99.50th=[  297], 99.90th=[  314], 99. |	     | 99.00th=[ 1385], 99.50th=[ 1418], 99.90th=[ 1467], 99.
     | 99.99th=[  898]					      |	     | 99.99th=[ 1500]
Type=write,Size=512B,NumJobs=56,depth=1				Type=write,Size=512B,NumJobs=56,depth=1
  write: IOPS=2332k, BW=1139MiB/s (1194MB/s)(11.1GiB/10003mse |	  write: IOPS=2369k, BW=1157MiB/s (1213MB/s)(11.3GiB/10002mse
    slat (usec): min=2, max=720, avg= 3.74, stdev= 2.14	      |	    slat (nsec): min=1823, max=501953, avg=3567.43, stdev=201
    clat (nsec): min=624, max=5616.3k, avg=19430.25, stdev=63 |	    clat (nsec): min=449, max=1722.5k, avg=19318.77, stdev=42
     lat (usec): min=10, max=5623, avg=23.28, stdev= 6.86     |	     lat (usec): min=11, max=1724, avg=22.98, stdev= 4.83
     |  1.00th=[   13],  5.00th=[   15], 10.00th=[   16], 20. |	     |  1.00th=[12992],  5.00th=[14656], 10.00th=[15680], 20.
     | 30.00th=[   18], 40.00th=[   19], 50.00th=[   19], 60. |	     | 30.00th=[17536], 40.00th=[18048], 50.00th=[18816], 60.
     | 70.00th=[   21], 80.00th=[   22], 90.00th=[   24], 95. |	     | 70.00th=[20352], 80.00th=[21632], 90.00th=[23424], 95.
     | 99.00th=[   33], 99.50th=[   37], 99.90th=[   47], 99. |	     | 99.00th=[31872], 99.50th=[35584], 99.90th=[41728], 99.
     | 99.99th=[  310]					      |	     | 99.99th=[54016]
Type=write,Size=512B,NumJobs=56,depth=16			Type=write,Size=512B,NumJobs=56,depth=16
  write: IOPS=8099k, BW=3955MiB/s (4147MB/s)(38.6GiB/10002mse |	  write: IOPS=6237k, BW=3046MiB/s (3193MB/s)(29.7GiB/10002mse
    slat (nsec): min=1742, max=11042k, avg=5139.82, stdev=693 |	    slat (nsec): min=1640, max=2381.5k, avg=5462.05, stdev=67
    clat (nsec): min=1973, max=13062k, avg=104665.53, stdev=5 |	    clat (usec): min=5, max=10435, avg=137.19, stdev=82.68
     lat (usec): min=31, max=13064, avg=109.91, stdev=54.98   |	     lat (usec): min=15, max=10447, avg=142.77, stdev=85.58
     |  1.00th=[   66],  5.00th=[   69], 10.00th=[   71], 20. |	     |  1.00th=[   63],  5.00th=[   68], 10.00th=[   70], 20.
     | 30.00th=[   74], 40.00th=[   75], 50.00th=[   76], 60. |	     | 30.00th=[   85], 40.00th=[   96], 50.00th=[  104], 60.
     | 70.00th=[   88], 80.00th=[  167], 90.00th=[  188], 95. |	     | 70.00th=[  149], 80.00th=[  210], 90.00th=[  265], 95.
     | 99.00th=[  215], 99.50th=[  223], 99.90th=[  375], 99. |	     | 99.00th=[  363], 99.50th=[  396], 99.90th=[  553], 99.
     | 99.99th=[  898]					      |	     | 99.99th=[  873]
Type=write,Size=512B,NumJobs=56,depth=64			Type=write,Size=512B,NumJobs=56,depth=64
  write: IOPS=8134k, BW=3972MiB/s (4165MB/s)(38.8GiB/10002mse |	  write: IOPS=7488k, BW=3656MiB/s (3834MB/s)(35.7GiB/10002mse
    slat (nsec): min=1797, max=599123, avg=5146.84, stdev=693 |	    slat (nsec): min=1666, max=7929.3k, avg=5382.53, stdev=12
    clat (usec): min=42, max=1474, avg=434.58, stdev=198.63   |	    clat (usec): min=14, max=12517, avg=472.19, stdev=331.30
     lat (usec): min=49, max=1477, avg=439.83, stdev=201.01   |	     lat (usec): min=17, max=12520, avg=477.70, stdev=334.84
     |  1.00th=[  281],  5.00th=[  289], 10.00th=[  297], 20. |	     |  1.00th=[  265],  5.00th=[  277], 10.00th=[  289], 20.
     | 30.00th=[  310], 40.00th=[  310], 50.00th=[  314], 60. |	     | 30.00th=[  302], 40.00th=[  306], 50.00th=[  314], 60.
     | 70.00th=[  326], 80.00th=[  725], 90.00th=[  758], 95. |	     | 70.00th=[  347], 80.00th=[  562], 90.00th=[ 1057], 95.
     | 99.00th=[  799], 99.50th=[  816], 99.90th=[  840], 99. |	     | 99.00th=[ 1434], 99.50th=[ 1696], 99.90th=[ 2040], 99.
     | 99.99th=[  922]					      |	     | 99.99th=[ 2474]
Type=write,Size=4K,NumJobs=1,depth=1				Type=write,Size=4K,NumJobs=1,depth=1
  write: IOPS=54.1k, BW=211MiB/s (222MB/s)(2115MiB/10001msec) |	  write: IOPS=55.8k, BW=218MiB/s (229MB/s)(2180MiB/10000msec)
    slat (nsec): min=2154, max=32419, avg=2620.30, stdev=1579 |	    slat (nsec): min=2138, max=44613, avg=2552.61, stdev=1404
    clat (nsec): min=639, max=171519, avg=15378.78, stdev=558 |	    clat (nsec): min=627, max=624863, avg=14910.38, stdev=517
     lat (usec): min=13, max=173, avg=18.04, stdev= 6.03      |	     lat (usec): min=13, max=627, avg=17.50, stdev= 5.55
     |  1.00th=[12864],  5.00th=[12992], 10.00th=[12992], 20. |	     |  1.00th=[12736],  5.00th=[12864], 10.00th=[12864], 20.
     | 30.00th=[13120], 40.00th=[13120], 50.00th=[13248], 60. |	     | 30.00th=[12992], 40.00th=[12992], 50.00th=[12992], 60.
     | 70.00th=[13376], 80.00th=[15424], 90.00th=[20864], 95. |	     | 70.00th=[13248], 80.00th=[13632], 90.00th=[20352], 95.
     | 99.00th=[36096], 99.50th=[38144], 99.90th=[48384], 99. |	     | 99.00th=[35584], 99.50th=[37120], 99.90th=[44800], 99.
     | 99.99th=[69120]					      |	     | 99.99th=[53504]
Type=write,Size=4K,NumJobs=1,depth=16				Type=write,Size=4K,NumJobs=1,depth=16
  write: IOPS=285k, BW=1112MiB/s (1166MB/s)(10.9GiB/10001msec |	  write: IOPS=325k, BW=1271MiB/s (1332MB/s)(12.4GiB/10001msec
    slat (nsec): min=1784, max=29113, avg=2129.85, stdev=618. |	    slat (nsec): min=1697, max=32986, avg=2232.67, stdev=205.
    clat (usec): min=40, max=179, avg=53.75, stdev=12.39      |	    clat (nsec): min=14620, max=89670, avg=46638.22, stdev=11
     lat (usec): min=42, max=182, avg=55.91, stdev=12.40      |	     lat (nsec): min=16814, max=91833, avg=48907.48, stdev=11
     |  1.00th=[   43],  5.00th=[   44], 10.00th=[   44], 20. |	     |  1.00th=[44800],  5.00th=[45312], 10.00th=[45824], 20.
     | 30.00th=[   45], 40.00th=[   46], 50.00th=[   50], 60. |	     | 30.00th=[46336], 40.00th=[46336], 50.00th=[46336], 60.
     | 70.00th=[   58], 80.00th=[   64], 90.00th=[   72], 95. |	     | 70.00th=[46848], 80.00th=[46848], 90.00th=[47360], 95.
     | 99.00th=[   96], 99.50th=[  105], 99.90th=[  129], 99. |	     | 99.00th=[50944], 99.50th=[53504], 99.90th=[58112], 99.
     | 99.99th=[  147]					      |	     | 99.99th=[63744]
Type=write,Size=4K,NumJobs=1,depth=64				Type=write,Size=4K,NumJobs=1,depth=64
  write: IOPS=340k, BW=1327MiB/s (1391MB/s)(12.0GiB/10001msec |	  write: IOPS=322k, BW=1259MiB/s (1320MB/s)(12.3GiB/10001msec
    slat (nsec): min=1701, max=34828, avg=2128.31, stdev=233. |	    slat (nsec): min=1697, max=33655, avg=2273.54, stdev=208.
    clat (usec): min=52, max=233, avg=185.96, stdev= 1.94     |	    clat (usec): min=18, max=238, avg=195.95, stdev= 2.07
     lat (usec): min=54, max=235, avg=188.13, stdev= 1.95     |	     lat (usec): min=20, max=241, avg=198.26, stdev= 2.08
     |  1.00th=[  182],  5.00th=[  184], 10.00th=[  184], 20. |	     |  1.00th=[  192],  5.00th=[  194], 10.00th=[  194], 20.
     | 30.00th=[  186], 40.00th=[  186], 50.00th=[  186], 60. |	     | 30.00th=[  196], 40.00th=[  196], 50.00th=[  196], 60.
     | 70.00th=[  188], 80.00th=[  188], 90.00th=[  188], 95. |	     | 70.00th=[  198], 80.00th=[  198], 90.00th=[  198], 95.
     | 99.00th=[  192], 99.50th=[  192], 99.90th=[  196], 99. |	     | 99.00th=[  202], 99.50th=[  204], 99.90th=[  208], 99.
     | 99.99th=[  202]					      |	     | 99.99th=[  212]
Type=write,Size=4K,NumJobs=16,depth=1				Type=write,Size=4K,NumJobs=16,depth=1
  write: IOPS=883k, BW=3449MiB/s (3616MB/s)(33.7GiB/10002msec |	  write: IOPS=855k, BW=3339MiB/s (3502MB/s)(32.6GiB/10001msec
    slat (nsec): min=2093, max=58113, avg=2489.62, stdev=384. |	    slat (nsec): min=1832, max=65363, avg=2579.38, stdev=826.
    clat (nsec): min=601, max=198998, avg=15168.57, stdev=132 |	    clat (nsec): min=472, max=134403, avg=15646.36, stdev=196
     lat (usec): min=12, max=201, avg=17.70, stdev= 1.39      |	     lat (usec): min=12, max=136, avg=18.28, stdev= 2.21
     |  1.00th=[12608],  5.00th=[13760], 10.00th=[14144], 20. |	     |  1.00th=[13120],  5.00th=[13760], 10.00th=[14016], 20.
     | 30.00th=[14656], 40.00th=[14784], 50.00th=[15040], 60. |	     | 30.00th=[14656], 40.00th=[14912], 50.00th=[15168], 60.
     | 70.00th=[15424], 80.00th=[15808], 90.00th=[16320], 95. |	     | 70.00th=[15936], 80.00th=[16512], 90.00th=[17536], 95.
     | 99.00th=[18816], 99.50th=[20096], 99.90th=[25984], 99. |	     | 99.00th=[23680], 99.50th=[25728], 99.90th=[33536], 99.
     | 99.99th=[44800]					      |	     | 99.99th=[37120]
Type=write,Size=4K,NumJobs=16,depth=16				Type=write,Size=4K,NumJobs=16,depth=16
  write: IOPS=2677k, BW=10.2GiB/s (10.0GB/s)(102GiB/10001msec |	  write: IOPS=1826k, BW=7133MiB/s (7480MB/s)(69.7GiB/10001mse
    slat (nsec): min=1663, max=57295, avg=2380.16, stdev=1242 |	    slat (nsec): min=1639, max=113071, avg=3312.71, stdev=236
    clat (usec): min=37, max=345, avg=92.86, stdev=12.80      |	    clat (usec): min=40, max=454, avg=136.35, stdev=35.22
     lat (usec): min=41, max=351, avg=95.28, stdev=12.74      |	     lat (usec): min=43, max=456, avg=139.72, stdev=35.40
     |  1.00th=[   69],  5.00th=[   75], 10.00th=[   78], 20. |	     |  1.00th=[   72],  5.00th=[   91], 10.00th=[  101], 20.
     | 30.00th=[   86], 40.00th=[   92], 50.00th=[   96], 60. |	     | 30.00th=[  121], 40.00th=[  127], 50.00th=[  133], 60.
     | 70.00th=[   99], 80.00th=[  101], 90.00th=[  104], 95. |	     | 70.00th=[  143], 80.00th=[  149], 90.00th=[  182], 95.
     | 99.00th=[  133], 99.50th=[  151], 99.90th=[  178], 99. |	     | 99.00th=[  258], 99.50th=[  273], 99.90th=[  314], 99.
     | 99.99th=[  212]					      |	     | 99.99th=[  367]
Type=write,Size=4K,NumJobs=16,depth=64				Type=write,Size=4K,NumJobs=16,depth=64
  write: IOPS=2691k, BW=10.3GiB/s (11.0GB/s)(103GiB/10001msec |	  write: IOPS=1875k, BW=7325MiB/s (7680MB/s)(71.5GiB/10001mse
    slat (nsec): min=1672, max=299471, avg=2451.55, stdev=255 |	    slat (nsec): min=1638, max=627057, avg=3232.91, stdev=324
    clat (usec): min=70, max=965, avg=377.69, stdev=68.69     |	    clat (usec): min=49, max=1357, avg=542.32, stdev=162.71
     lat (usec): min=77, max=980, avg=380.18, stdev=69.11     |	     lat (usec): min=52, max=1384, avg=545.61, stdev=162.80
     |  1.00th=[  223],  5.00th=[  285], 10.00th=[  314], 20. |	     |  1.00th=[  202],  5.00th=[  277], 10.00th=[  330], 20.
     | 30.00th=[  351], 40.00th=[  363], 50.00th=[  375], 60. |	     | 30.00th=[  453], 40.00th=[  498], 50.00th=[  537], 60.
     | 70.00th=[  396], 80.00th=[  412], 90.00th=[  433], 95. |	     | 70.00th=[  635], 80.00th=[  685], 90.00th=[  758], 95.
     | 99.00th=[  693], 99.50th=[  717], 99.90th=[  758], 99. |	     | 99.00th=[  922], 99.50th=[  955], 99.90th=[ 1029], 99.
     | 99.99th=[  807]					      |	     | 99.99th=[ 1123]
Type=write,Size=4K,NumJobs=56,depth=1				Type=write,Size=4K,NumJobs=56,depth=1
  write: IOPS=1417k, BW=5534MiB/s (5803MB/s)(54.1GiB/10002mse |	  write: IOPS=1416k, BW=5533MiB/s (5802MB/s)(54.0GiB/10002mse
    slat (usec): min=2, max=722, avg= 3.06, stdev= 1.31	      |	    slat (nsec): min=1846, max=533231, avg=2881.71, stdev=111
    clat (nsec): min=647, max=796809, avg=35776.52, stdev=697 |	    clat (nsec): min=498, max=4453.2k, avg=36057.87, stdev=38
     lat (usec): min=14, max=799, avg=38.92, stdev= 7.08      |	     lat (usec): min=15, max=4456, avg=39.01, stdev= 4.01
     |  1.00th=[   28],  5.00th=[   32], 10.00th=[   33], 20. |	     |  1.00th=[29312],  5.00th=[32128], 10.00th=[33024], 20.
     | 30.00th=[   35], 40.00th=[   35], 50.00th=[   36], 60. |	     | 30.00th=[34560], 40.00th=[35072], 50.00th=[35584], 60.
     | 70.00th=[   37], 80.00th=[   38], 90.00th=[   40], 95. |	     | 70.00th=[37120], 80.00th=[37632], 90.00th=[38656], 95.
     | 99.00th=[   50], 99.50th=[   53], 99.90th=[   69], 99. |	     | 99.00th=[49408], 99.50th=[52480], 99.90th=[56576], 99.
     | 99.99th=[  367]					      |	     | 99.99th=[64768]
Type=write,Size=4K,NumJobs=56,depth=16				Type=write,Size=4K,NumJobs=56,depth=16
  write: IOPS=2470k, BW=9649MiB/s (10.1GB/s)(94.3GiB/10002mse |	  write: IOPS=1616k, BW=6314MiB/s (6621MB/s)(61.7GiB/10001mse
    slat (nsec): min=1714, max=353380, avg=3021.76, stdev=221 |	    slat (nsec): min=1649, max=738008, avg=3245.81, stdev=163
    clat (usec): min=44, max=1490, avg=359.06, stdev=129.14   |	    clat (usec): min=44, max=5046, avg=550.36, stdev=143.07
     lat (usec): min=47, max=1493, avg=362.15, stdev=129.13   |	     lat (usec): min=46, max=5049, avg=553.68, stdev=143.03
     |  1.00th=[  212],  5.00th=[  235], 10.00th=[  247], 20. |	     |  1.00th=[  215],  5.00th=[  343], 10.00th=[  404], 20.
     | 30.00th=[  277], 40.00th=[  285], 50.00th=[  302], 60. |	     | 30.00th=[  490], 40.00th=[  510], 50.00th=[  529], 60.
     | 70.00th=[  392], 80.00th=[  478], 90.00th=[  570], 95. |	     | 70.00th=[  586], 80.00th=[  660], 90.00th=[  758], 95.
     | 99.00th=[  717], 99.50th=[  750], 99.90th=[  840], 99. |	     | 99.00th=[  938], 99.50th=[  996], 99.90th=[ 1106], 99.
     | 99.99th=[ 1004]					      |	     | 99.99th=[ 1221]
Type=write,Size=4K,NumJobs=56,depth=64				Type=write,Size=4K,NumJobs=56,depth=64
  write: IOPS=2093k, BW=8177MiB/s (8574MB/s)(79.9GiB/10003mse |	  write: IOPS=1625k, BW=6348MiB/s (6656MB/s)(62.0GiB/10003mse
    slat (nsec): min=1735, max=4573.4k, avg=18118.68, stdev=6 |	    slat (nsec): min=1656, max=1933.3k, avg=21620.58, stdev=7
    clat (usec): min=40, max=8076, avg=1693.04, stdev=804.74  |	    clat (usec): min=66, max=8663, avg=2182.95, stdev=879.90
     lat (usec): min=44, max=8079, avg=1711.22, stdev=817.33  |	     lat (usec): min=69, max=8669, avg=2204.64, stdev=895.51
     |  1.00th=[  375],  5.00th=[  644], 10.00th=[  766], 20. |	     |  1.00th=[  693],  5.00th=[  971], 10.00th=[ 1074], 20.
     | 30.00th=[ 1205], 40.00th=[ 1516], 50.00th=[ 1647], 60. |	     | 30.00th=[ 1942], 40.00th=[ 2114], 50.00th=[ 2212], 60.
     | 70.00th=[ 1909], 80.00th=[ 2278], 90.00th=[ 2835], 95. |	     | 70.00th=[ 2343], 80.00th=[ 2507], 90.00th=[ 3392], 95.
     | 99.00th=[ 4015], 99.50th=[ 4293], 99.90th=[ 4883], 99. |	     | 99.00th=[ 4883], 99.50th=[ 5211], 99.90th=[ 5800], 99.
     | 99.99th=[ 5735]					      |	     | 99.99th=[ 6587]
Type=write,Size=64K,NumJobs=1,depth=1				Type=write,Size=64K,NumJobs=1,depth=1
  write: IOPS=23.3k, BW=1455MiB/s (1526MB/s)(14.2GiB/10001mse |	  write: IOPS=23.3k, BW=1454MiB/s (1525MB/s)(14.2GiB/10001mse
    slat (usec): min=4, max=109, avg= 6.71, stdev= 3.24	      |	    slat (nsec): min=4849, max=39702, avg=6880.02, stdev=3240
    clat (usec): min=14, max=124, avg=35.67, stdev= 9.52      |	    clat (usec): min=17, max=133, avg=35.53, stdev= 9.39
     lat (usec): min=33, max=159, avg=42.42, stdev=10.65      |	     lat (usec): min=33, max=146, avg=42.46, stdev=10.52
     |  1.00th=[29312],  5.00th=[29312], 10.00th=[29312], 20. |	     |  1.00th=[29312],  5.00th=[29312], 10.00th=[29568], 20.
     | 30.00th=[29568], 40.00th=[29824], 50.00th=[30080], 60. |	     | 30.00th=[29568], 40.00th=[29824], 50.00th=[29824], 60.
     | 70.00th=[36608], 80.00th=[44800], 90.00th=[49920], 95. |	     | 70.00th=[36608], 80.00th=[43776], 90.00th=[49408], 95.
     | 99.00th=[68096], 99.50th=[72192], 99.90th=[86528], 99. |	     | 99.00th=[68096], 99.50th=[72192], 99.90th=[85504], 99.
     | 99.99th=[91648]					      |	     | 99.99th=[90624]
Type=write,Size=64K,NumJobs=1,depth=16				Type=write,Size=64K,NumJobs=1,depth=16
  write: IOPS=141k, BW=8811MiB/s (9239MB/s)(86.1GiB/10001msec |	  write: IOPS=146k, BW=9109MiB/s (9552MB/s)(88.0GiB/10001msec
    slat (nsec): min=3960, max=44206, avg=5332.37, stdev=776. |	    slat (nsec): min=3787, max=44724, avg=5539.44, stdev=622.
    clat (usec): min=83, max=307, avg=107.82, stdev=13.00     |	    clat (usec): min=76, max=228, avg=103.90, stdev=11.08
     lat (usec): min=88, max=312, avg=113.19, stdev=13.00     |	     lat (usec): min=82, max=233, avg=109.48, stdev=11.10
     |  1.00th=[   92],  5.00th=[   96], 10.00th=[   98], 20. |	     |  1.00th=[   90],  5.00th=[   93], 10.00th=[   95], 20.
     | 30.00th=[  102], 40.00th=[  103], 50.00th=[  105], 60. |	     | 30.00th=[   98], 40.00th=[   99], 50.00th=[  101], 60.
     | 70.00th=[  110], 80.00th=[  113], 90.00th=[  121], 95. |	     | 70.00th=[  105], 80.00th=[  109], 90.00th=[  120], 95.
     | 99.00th=[  163], 99.50th=[  174], 99.90th=[  204], 99. |	     | 99.00th=[  141], 99.50th=[  149], 99.90th=[  182], 99.
     | 99.99th=[  241]					      |	     | 99.99th=[  200]
Type=write,Size=64K,NumJobs=1,depth=64				Type=write,Size=64K,NumJobs=1,depth=64
  write: IOPS=144k, BW=8971MiB/s (9407MB/s)(87.6GiB/10001msec |	  write: IOPS=139k, BW=8708MiB/s (9131MB/s)(85.0GiB/10001msec
    slat (nsec): min=2732, max=48101, avg=4533.01, stdev=1509 |	    slat (nsec): min=2676, max=64177, avg=5388.79, stdev=1847
    clat (usec): min=162, max=1015, avg=440.98, stdev=74.73   |	    clat (usec): min=39, max=894, avg=453.54, stdev=56.76
     lat (usec): min=174, max=1022, avg=445.55, stdev=74.68   |	     lat (usec): min=45, max=898, avg=458.97, stdev=57.24
     |  1.00th=[  322],  5.00th=[  326], 10.00th=[  338], 20. |	     |  1.00th=[  338],  5.00th=[  355], 10.00th=[  367], 20.
     | 30.00th=[  400], 40.00th=[  412], 50.00th=[  437], 60. |	     | 30.00th=[  408], 40.00th=[  457], 50.00th=[  478], 60.
     | 70.00th=[  482], 80.00th=[  502], 90.00th=[  537], 95. |	     | 70.00th=[  494], 80.00th=[  498], 90.00th=[  506], 95.
     | 99.00th=[  635], 99.50th=[  693], 99.90th=[  799], 99. |	     | 99.00th=[  545], 99.50th=[  562], 99.90th=[  734], 99.
     | 99.99th=[  898]					      |	     | 99.99th=[  848]
Type=write,Size=64K,NumJobs=16,depth=1				Type=write,Size=64K,NumJobs=16,depth=1
  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10001msec)	  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10001msec)
    slat (nsec): min=4432, max=41303, avg=5462.66, stdev=782. |	    slat (nsec): min=4383, max=73755, avg=5607.60, stdev=1278
    clat (usec): min=23, max=214, avg=85.24, stdev= 5.53      |	    clat (usec): min=22, max=424, avg=85.24, stdev= 8.07
     lat (usec): min=37, max=220, avg=90.75, stdev= 5.58      |	     lat (usec): min=38, max=430, avg=90.90, stdev= 8.07
     |  1.00th=[   68],  5.00th=[   80], 10.00th=[   82], 20. |	     |  1.00th=[   68],  5.00th=[   76], 10.00th=[   79], 20.
     | 30.00th=[   84], 40.00th=[   85], 50.00th=[   86], 60. |	     | 30.00th=[   82], 40.00th=[   83], 50.00th=[   84], 60.
     | 70.00th=[   87], 80.00th=[   88], 90.00th=[   90], 95. |	     | 70.00th=[   87], 80.00th=[   91], 90.00th=[   98], 95.
     | 99.00th=[  103], 99.50th=[  110], 99.90th=[  123], 99. |	     | 99.00th=[  106], 99.50th=[  112], 99.90th=[  124], 99.
     | 99.99th=[  137]						     | 99.99th=[  137]
Type=write,Size=64K,NumJobs=16,depth=16				Type=write,Size=64K,NumJobs=16,depth=16
  write: IOPS=176k, BW=10.7GiB/s (11.5GB/s)(107GiB/10002msec)	  write: IOPS=176k, BW=10.7GiB/s (11.5GB/s)(107GiB/10002msec)
    slat (nsec): min=4024, max=70542, avg=6729.45, stdev=1425 |	    slat (nsec): min=3611, max=70931, avg=6507.12, stdev=1389
    clat (usec): min=103, max=4145, avg=1447.24, stdev=446.02 |	    clat (usec): min=109, max=3915, avg=1451.13, stdev=409.41
     lat (usec): min=120, max=4152, avg=1454.01, stdev=446.19 |	     lat (usec): min=129, max=3920, avg=1457.68, stdev=409.50
     |  1.00th=[  474],  5.00th=[ 1237], 10.00th=[ 1254], 20. |	     |  1.00th=[ 1254],  5.00th=[ 1270], 10.00th=[ 1270], 20.
     | 30.00th=[ 1270], 40.00th=[ 1270], 50.00th=[ 1270], 60. |	     | 30.00th=[ 1270], 40.00th=[ 1270], 50.00th=[ 1287], 60.
     | 70.00th=[ 1369], 80.00th=[ 1450], 90.00th=[ 2540], 95. |	     | 70.00th=[ 1352], 80.00th=[ 1369], 90.00th=[ 2540], 95.
     | 99.00th=[ 2671], 99.50th=[ 2704], 99.90th=[ 2933], 99. |	     | 99.00th=[ 2737], 99.50th=[ 2737], 99.90th=[ 2769], 99.
     | 99.99th=[ 3916]					      |	     | 99.99th=[ 2769]
Type=write,Size=64K,NumJobs=16,depth=64				Type=write,Size=64K,NumJobs=16,depth=64
  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10004msec)	  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10004msec)
    slat (usec): min=2, max=6195, avg=13.26, stdev=111.30     |	    slat (nsec): min=3042, max=96073, avg=6866.24, stdev=2170
    clat (usec): min=112, max=16931, avg=5822.91, stdev=1645. |	    clat (usec): min=124, max=10103, avg=5850.40, stdev=798.1
     lat (usec): min=118, max=17733, avg=5836.21, stdev=1665. |	     lat (usec): min=131, max=10109, avg=5857.31, stdev=798.1
     |  1.00th=[ 4015],  5.00th=[ 4293], 10.00th=[ 4424], 20. |	     |  1.00th=[ 4752],  5.00th=[ 4883], 10.00th=[ 4948], 20.
     | 30.00th=[ 4883], 40.00th=[ 5080], 50.00th=[ 5276], 60. |	     | 30.00th=[ 5276], 40.00th=[ 5407], 50.00th=[ 5669], 60.
     | 70.00th=[ 6063], 80.00th=[ 6587], 90.00th=[ 7570], 95. |	     | 70.00th=[ 6259], 80.00th=[ 6587], 90.00th=[ 7046], 95.
     | 99.00th=[11600], 99.50th=[12387], 99.90th=[14091], 99. |	     | 99.00th=[ 7767], 99.50th=[ 7898], 99.90th=[ 8455], 99.
     | 99.99th=[15795]					      |	     | 99.99th=[ 8848]
Type=write,Size=64K,NumJobs=56,depth=1				Type=write,Size=64K,NumJobs=56,depth=1
  write: IOPS=176k, BW=10.7GiB/s (11.5GB/s)(107GiB/10001msec) |	  write: IOPS=175k, BW=10.7GiB/s (11.4GB/s)(107GiB/10002msec)
    slat (nsec): min=4406, max=47421, avg=6031.08, stdev=1752 |	    slat (nsec): min=4029, max=63780, avg=5771.34, stdev=1537
    clat (usec): min=49, max=915, avg=312.16, stdev=47.66     |	    clat (usec): min=32, max=979, avg=314.28, stdev=57.25
     lat (usec): min=57, max=924, avg=318.25, stdev=47.66     |	     lat (usec): min=40, max=985, avg=320.11, stdev=57.29
     |  1.00th=[  194],  5.00th=[  281], 10.00th=[  289], 20. |	     |  1.00th=[  186],  5.00th=[  293], 10.00th=[  297], 20.
     | 30.00th=[  302], 40.00th=[  306], 50.00th=[  306], 60. |	     | 30.00th=[  297], 40.00th=[  302], 50.00th=[  302], 60.
     | 70.00th=[  310], 80.00th=[  310], 90.00th=[  326], 95. |	     | 70.00th=[  310], 80.00th=[  310], 90.00th=[  326], 95.
     | 99.00th=[  469], 99.50th=[  482], 99.90th=[  611], 99. |	     | 99.00th=[  603], 99.50th=[  611], 99.90th=[  627], 99.
     | 99.99th=[  644]					      |	     | 99.99th=[  750]
Type=write,Size=64K,NumJobs=56,depth=16				Type=write,Size=64K,NumJobs=56,depth=16
  write: IOPS=176k, BW=10.7GiB/s (11.5GB/s)(107GiB/10005msec) |	  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10005msec)
    slat (usec): min=3, max=102, avg= 7.88, stdev= 2.26	      |	    slat (nsec): min=3542, max=89310, avg=7585.26, stdev=2460
    clat (usec): min=157, max=13620, avg=5092.13, stdev=1280. |	    clat (usec): min=190, max=14367, avg=5116.17, stdev=1396.
     lat (usec): min=183, max=13626, avg=5100.08, stdev=1280. |	     lat (usec): min=198, max=14375, avg=5123.82, stdev=1396.
     |  1.00th=[ 2442],  5.00th=[ 2540], 10.00th=[ 2540], 20. |	     |  1.00th=[ 2540],  5.00th=[ 2540], 10.00th=[ 2573], 20.
     | 30.00th=[ 5080], 40.00th=[ 5080], 50.00th=[ 5080], 60. |	     | 30.00th=[ 5080], 40.00th=[ 5080], 50.00th=[ 5145], 60.
     | 70.00th=[ 5145], 80.00th=[ 5538], 90.00th=[ 6652], 95. |	     | 70.00th=[ 5145], 80.00th=[ 5669], 90.00th=[ 6849], 95.
     | 99.00th=[ 8586], 99.50th=[ 9241], 99.90th=[10552], 99. |	     | 99.00th=[ 8979], 99.50th=[ 9634], 99.90th=[11207], 99.
     | 99.99th=[12256]					      |	     | 99.99th=[12780]
Type=write,Size=64K,NumJobs=56,depth=64				Type=write,Size=64K,NumJobs=56,depth=64
  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10013msec) |	  write: IOPS=174k, BW=10.6GiB/s (11.4GB/s)(107GiB/10011msec)
    slat (usec): min=2, max=36257, avg=176.13, stdev=892.90   |	    slat (usec): min=3, max=32839, avg=176.71, stdev=925.53
    clat (usec): min=184, max=83223, avg=20322.48, stdev=6000 |	    clat (usec): min=189, max=132076, avg=20366.61, stdev=728
     lat (usec): min=188, max=83231, avg=20498.69, stdev=6090 |	     lat (usec): min=218, max=132086, avg=20543.39, stdev=740
     |  1.00th=[ 7832],  5.00th=[10028], 10.00th=[13173], 20. |	     |  1.00th=[ 7898],  5.00th=[ 9110], 10.00th=[10945], 20.
     | 30.00th=[18220], 40.00th=[19792], 50.00th=[20317], 60. |	     | 30.00th=[17957], 40.00th=[19530], 50.00th=[20317], 60.
     | 70.00th=[21103], 80.00th=[22938], 90.00th=[26870], 95. |	     | 70.00th=[21103], 80.00th=[23462], 90.00th=[28705], 95.
     | 99.00th=[40633], 99.50th=[45351], 99.90th=[54264], 99. |	     | 99.00th=[45876], 99.50th=[50594], 99.90th=[63177], 99.
     | 99.99th=[67634]					      |	     | 99.99th=[84411]
With DIM on initiator side					Without DIM
Type=read,Size=512B,NumJobs=1,depth=1				Type=read,Size=512B,NumJobs=1,depth=1
   read: IOPS=46.7k, BW=22.8MiB/s (23.9MB/s)(228MiB/10001msec |	   read: IOPS=47.0k, BW=22.0MiB/s (24.1MB/s)(230MiB/10001msec
    slat (nsec): min=2008, max=43561, avg=3175.68, stdev=2353 |	    slat (nsec): min=1981, max=53196, avg=3048.29, stdev=2391
    clat (nsec): min=456, max=103914, avg=17680.68, stdev=638 |	    clat (nsec): min=442, max=107156, avg=17658.37, stdev=647
     lat (usec): min=13, max=106, avg=20.93, stdev= 7.10      |	     lat (usec): min=13, max=109, avg=20.78, stdev= 7.18
     |  1.00th=[12864],  5.00th=[12992], 10.00th=[14016], 20. |	     |  1.00th=[13632],  5.00th=[13760], 10.00th=[13888], 20.
     | 30.00th=[14144], 40.00th=[14272], 50.00th=[14400], 60. |	     | 30.00th=[14144], 40.00th=[14144], 50.00th=[14400], 60.
     | 70.00th=[17280], 80.00th=[19072], 90.00th=[28032], 95. |	     | 70.00th=[17024], 80.00th=[19072], 90.00th=[27776], 95.
     | 99.00th=[40192], 99.50th=[44800], 99.90th=[50944], 99. |	     | 99.00th=[40704], 99.50th=[45824], 99.90th=[51456], 99.
     | 99.99th=[60160]					      |	     | 99.99th=[58624]
Type=read,Size=512B,NumJobs=1,depth=16				Type=read,Size=512B,NumJobs=1,depth=16
   read: IOPS=281k, BW=137MiB/s (144MB/s)(1371MiB/10001msec)  |	   read: IOPS=334k, BW=163MiB/s (171MB/s)(1629MiB/10001msec)
    slat (nsec): min=1558, max=56900, avg=2362.92, stdev=2075 |	    slat (nsec): min=1564, max=28780, avg=2162.48, stdev=588.
    clat (usec): min=32, max=158, avg=54.17, stdev= 9.58      |	    clat (usec): min=17, max=118, avg=45.50, stdev= 6.30
     lat (usec): min=34, max=160, avg=56.59, stdev= 9.72      |	     lat (usec): min=20, max=120, avg=47.70, stdev= 6.64
     |  1.00th=[   40],  5.00th=[   41], 10.00th=[   43], 20. |	     |  1.00th=[41728],  5.00th=[42752], 10.00th=[42752], 20.
     | 30.00th=[   47], 40.00th=[   53], 50.00th=[   56], 60. |	     | 30.00th=[43264], 40.00th=[43776], 50.00th=[43776], 60.
     | 70.00th=[   60], 80.00th=[   62], 90.00th=[   65], 95. |	     | 70.00th=[44288], 80.00th=[44288], 90.00th=[46336], 95.
     | 99.00th=[   80], 99.50th=[   85], 99.90th=[   99], 99. |	     | 99.00th=[72192], 99.50th=[73216], 99.90th=[77312], 99.
     | 99.99th=[  133]					      |	     | 99.99th=[86528]
Type=read,Size=512B,NumJobs=1,depth=64				Type=read,Size=512B,NumJobs=1,depth=64
   read: IOPS=377k, BW=184MiB/s (193MB/s)(1840MiB/10001msec)  |	   read: IOPS=344k, BW=168MiB/s (176MB/s)(1682MiB/10001msec)
    slat (nsec): min=1591, max=41737, avg=1885.11, stdev=245. |	    slat (nsec): min=1607, max=30650, avg=2093.56, stdev=233.
    clat (usec): min=29, max=222, avg=167.69, stdev= 1.31     |	    clat (usec): min=19, max=238, avg=183.42, stdev= 2.22
     lat (usec): min=32, max=224, avg=169.61, stdev= 1.33     |	     lat (usec): min=21, max=241, avg=185.55, stdev= 2.23
     |  1.00th=[  165],  5.00th=[  167], 10.00th=[  167], 20. |	     |  1.00th=[  180],  5.00th=[  182], 10.00th=[  182], 20.
     | 30.00th=[  167], 40.00th=[  167], 50.00th=[  167], 60. |	     | 30.00th=[  182], 40.00th=[  184], 50.00th=[  184], 60.
     | 70.00th=[  169], 80.00th=[  169], 90.00th=[  169], 95. |	     | 70.00th=[  184], 80.00th=[  186], 90.00th=[  186], 95.
     | 99.00th=[  172], 99.50th=[  172], 99.90th=[  178], 99. |	     | 99.00th=[  190], 99.50th=[  192], 99.90th=[  196], 99.
     | 99.99th=[  196]					      |	     | 99.99th=[  210]
Type=read,Size=512B,NumJobs=16,depth=1				Type=read,Size=512B,NumJobs=16,depth=1
   read: IOPS=906k, BW=442MiB/s (464MB/s)(4423MiB/10001msec)  |	   read: IOPS=907k, BW=443MiB/s (465MB/s)(4431MiB/10001msec)
    slat (nsec): min=1919, max=71597, avg=2515.26, stdev=515. |	    slat (nsec): min=1925, max=69082, avg=2505.79, stdev=869.
    clat (nsec): min=678, max=1006.2k, avg=14689.14, stdev=14 |	    clat (nsec): min=476, max=112521, avg=14672.56, stdev=182
     lat (usec): min=12, max=1008, avg=17.26, stdev= 1.66     |	     lat (usec): min=12, max=131, avg=17.23, stdev= 2.08
     |  1.00th=[11712],  5.00th=[12736], 10.00th=[13376], 20. |	     |  1.00th=[11584],  5.00th=[13248], 10.00th=[13376], 20.
     | 30.00th=[14144], 40.00th=[14400], 50.00th=[14656], 60. |	     | 30.00th=[13888], 40.00th=[14016], 50.00th=[14272], 60.
     | 70.00th=[15168], 80.00th=[15424], 90.00th=[15936], 95. |	     | 70.00th=[14784], 80.00th=[15296], 90.00th=[16512], 95.
     | 99.00th=[18048], 99.50th=[18816], 99.90th=[23424], 99. |	     | 99.00th=[22400], 99.50th=[24448], 99.90th=[32384], 99.
     | 99.99th=[62208]					      |	     | 99.99th=[35584]
Type=read,Size=512B,NumJobs=16,depth=16				Type=read,Size=512B,NumJobs=16,depth=16
   read: IOPS=4317k, BW=2108MiB/s (2211MB/s)(20.6GiB/10001mse |	   read: IOPS=4055k, BW=1980MiB/s (2076MB/s)(19.3GiB/10003mse
    slat (nsec): min=1572, max=75553, avg=2642.25, stdev=1686 |	    slat (nsec): min=1572, max=239521, avg=2836.25, stdev=282
    clat (usec): min=21, max=1043, avg=56.25, stdev=12.46     |	    clat (usec): min=9, max=616, avg=59.85, stdev=55.40
     lat (usec): min=23, max=1046, avg=58.94, stdev=13.08     |	     lat (usec): min=15, max=630, avg=62.74, stdev=57.98
     |  1.00th=[   42],  5.00th=[   43], 10.00th=[   44], 20. |	     |  1.00th=[   43],  5.00th=[   44], 10.00th=[   44], 20.
     | 30.00th=[   48], 40.00th=[   51], 50.00th=[   53], 60. |	     | 30.00th=[   45], 40.00th=[   46], 50.00th=[   46], 60.
     | 70.00th=[   62], 80.00th=[   66], 90.00th=[   72], 95. |	     | 70.00th=[   48], 80.00th=[   52], 90.00th=[   60], 95.
     | 99.00th=[   87], 99.50th=[  112], 99.90th=[  124], 99. |	     | 99.00th=[  326], 99.50th=[  343], 99.90th=[  367], 99.
     | 99.99th=[  289]					      |	     | 99.99th=[  519]
Type=read,Size=512B,NumJobs=16,depth=64				Type=read,Size=512B,NumJobs=16,depth=64
   read: IOPS=4484k, BW=2189MiB/s (2296MB/s)(21.4GiB/10001mse |	   read: IOPS=3451k, BW=1685MiB/s (1767MB/s)(16.5GiB/10001mse
    slat (nsec): min=1566, max=71087, avg=2603.51, stdev=2276 |	    slat (nsec): min=1569, max=204726, avg=3322.74, stdev=274
    clat (usec): min=19, max=1784, avg=225.36, stdev=53.73    |	    clat (usec): min=5, max=1543, avg=292.82, stdev=220.59
     lat (usec): min=26, max=1807, avg=228.01, stdev=54.38    |	     lat (usec): min=14, max=1561, avg=296.20, stdev=223.03
     |  1.00th=[  174],  5.00th=[  174], 10.00th=[  176], 20. |	     |  1.00th=[  176],  5.00th=[  178], 10.00th=[  180], 20.
     | 30.00th=[  182], 40.00th=[  184], 50.00th=[  231], 60. |	     | 30.00th=[  188], 40.00th=[  227], 50.00th=[  243], 60.
     | 70.00th=[  247], 80.00th=[  269], 90.00th=[  302], 95. |	     | 70.00th=[  249], 80.00th=[  285], 90.00th=[  685], 95.
     | 99.00th=[  375], 99.50th=[  388], 99.90th=[  490], 99. |	     | 99.00th=[ 1270], 99.50th=[ 1319], 99.90th=[ 1385], 99.
     | 99.99th=[ 1631]					      |	     | 99.99th=[ 1450]
Type=read,Size=512B,NumJobs=56,depth=1				Type=read,Size=512B,NumJobs=56,depth=1
   read: IOPS=2301k, BW=1123MiB/s (1178MB/s)(10.0GiB/10002mse |	   read: IOPS=2351k, BW=1148MiB/s (1204MB/s)(11.2GiB/10002mse
    slat (nsec): min=1723, max=958519, avg=3130.18, stdev=300 |	    slat (nsec): min=1744, max=502891, avg=3525.21, stdev=199
    clat (nsec): min=456, max=987494, avg=20564.41, stdev=170 |	    clat (nsec): min=456, max=781278, avg=19563.02, stdev=418
     lat (usec): min=12, max=989, avg=23.77, stdev=17.43      |	     lat (usec): min=11, max=785, avg=23.18, stdev= 4.79
     |  1.00th=[   14],  5.00th=[   16], 10.00th=[   17], 20. |	     |  1.00th=[13120],  5.00th=[14912], 10.00th=[15936], 20.
     | 30.00th=[   19], 40.00th=[   19], 50.00th=[   20], 60. |	     | 30.00th=[17792], 40.00th=[18304], 50.00th=[19072], 60.
     | 70.00th=[   21], 80.00th=[   22], 90.00th=[   24], 95. |	     | 70.00th=[20608], 80.00th=[21632], 90.00th=[23680], 95.
     | 99.00th=[   37], 99.50th=[   45], 99.90th=[  367], 99. |	     | 99.00th=[31872], 99.50th=[35072], 99.90th=[41728], 99.
     | 99.99th=[  545]					      |	     | 99.99th=[56064]
Type=read,Size=512B,NumJobs=56,depth=16				Type=read,Size=512B,NumJobs=56,depth=16
   read: IOPS=7830k, BW=3823MiB/s (4009MB/s)(37.3GiB/10002mse |	   read: IOPS=7234k, BW=3532MiB/s (3704MB/s)(34.5GiB/10002mse
    slat (nsec): min=1591, max=2609.7M, avg=5126.50, stdev=10 |	    slat (nsec): min=1588, max=13073k, avg=5296.03, stdev=602
    clat (usec): min=16, max=3527.6k, avg=108.61, stdev=6148. |	    clat (usec): min=9, max=13311, avg=117.66, stdev=77.29
     lat (usec): min=25, max=3527.6k, avg=113.83, stdev=6235. |	     lat (usec): min=18, max=13322, avg=123.07, stdev=80.61
     |  1.00th=[   47],  5.00th=[   54], 10.00th=[   63], 20. |	     |  1.00th=[   63],  5.00th=[   67], 10.00th=[   69], 20.
     | 30.00th=[   74], 40.00th=[   77], 50.00th=[   80], 60. |	     | 30.00th=[   72], 40.00th=[   76], 50.00th=[   81], 60.
     | 70.00th=[   97], 80.00th=[  100], 90.00th=[  105], 95. |	     | 70.00th=[  108], 80.00th=[  184], 90.00th=[  247], 95.
     | 99.00th=[  457], 99.50th=[  519], 99.90th=[  611], 99. |	     | 99.00th=[  338], 99.50th=[  383], 99.90th=[  494], 99.
     | 99.99th=[  889]					      |	     | 99.99th=[  594]
Type=read,Size=512B,NumJobs=56,depth=64				Type=read,Size=512B,NumJobs=56,depth=64
   read: IOPS=8456k, BW=4129MiB/s (4330MB/s)(40.3GiB/10002mse |	   read: IOPS=7769k, BW=3793MiB/s (3978MB/s)(37.1GiB/10002mse
    slat (nsec): min=1607, max=2432.8k, avg=4995.25, stdev=87 |	    slat (nsec): min=1583, max=3785.4k, avg=5319.20, stdev=81
    clat (usec): min=28, max=3427, avg=418.02, stdev=272.37   |	    clat (usec): min=2, max=5212, avg=455.02, stdev=334.71
     lat (usec): min=34, max=3431, avg=423.12, stdev=275.63   |	     lat (usec): min=17, max=5214, avg=460.45, stdev=338.64
     |  1.00th=[  281],  5.00th=[  293], 10.00th=[  302], 20. |	     |  1.00th=[  265],  5.00th=[  273], 10.00th=[  281], 20.
     | 30.00th=[  314], 40.00th=[  322], 50.00th=[  383], 60. |	     | 30.00th=[  293], 40.00th=[  297], 50.00th=[  302], 60.
     | 70.00th=[  408], 80.00th=[  412], 90.00th=[  429], 95. |	     | 70.00th=[  310], 80.00th=[  437], 90.00th=[ 1074], 95.
     | 99.00th=[ 1909], 99.50th=[ 1991], 99.90th=[ 2180], 99. |	     | 99.00th=[ 1401], 99.50th=[ 1991], 99.90th=[ 2245], 99.
     | 99.99th=[ 2442]						     | 99.99th=[ 2442]
Type=read,Size=4K,NumJobs=1,depth=1				Type=read,Size=4K,NumJobs=1,depth=1
   read: IOPS=43.6k, BW=170MiB/s (178MB/s)(1701MiB/10001msec) |	   read: IOPS=54.4k, BW=213MiB/s (223MB/s)(2126MiB/10001msec)
    slat (nsec): min=2041, max=41259, avg=3359.33, stdev=2676 |	    slat (nsec): min=2044, max=28781, avg=2556.03, stdev=1441
    clat (nsec): min=449, max=95097, avg=18989.95, stdev=6853 |	    clat (nsec): min=621, max=70566, avg=15359.54, stdev=5250
     lat (nsec): min=14883, max=99275, avg=22427.07, stdev=76 |	     lat (nsec): min=13302, max=73823, avg=17966.25, stdev=56
     |  1.00th=[14400],  5.00th=[14656], 10.00th=[14656], 20. |	     |  1.00th=[13120],  5.00th=[13248], 10.00th=[13248], 20.
     | 30.00th=[14912], 40.00th=[15040], 50.00th=[15296], 60. |	     | 30.00th=[13376], 40.00th=[13376], 50.00th=[13376], 60.
     | 70.00th=[19072], 80.00th=[22144], 90.00th=[30592], 95. |	     | 70.00th=[13632], 80.00th=[14144], 90.00th=[19328], 95.
     | 99.00th=[43776], 99.50th=[46848], 99.90th=[52480], 99. |	     | 99.00th=[35584], 99.50th=[36608], 99.90th=[45312], 99.
     | 99.99th=[63744]					      |	     | 99.99th=[53504]
Type=read,Size=4K,NumJobs=1,depth=16				Type=read,Size=4K,NumJobs=1,depth=16
   read: IOPS=345k, BW=1346MiB/s (1411MB/s)(13.1GiB/10001msec |	   read: IOPS=334k, BW=1306MiB/s (1370MB/s)(12.8GiB/10001msec
    slat (nsec): min=1584, max=67990, avg=2073.10, stdev=588. |	    slat (nsec): min=1696, max=45538, avg=2168.76, stdev=208.
    clat (usec): min=32, max=154, avg=44.06, stdev= 4.57      |	    clat (usec): min=18, max=148, avg=45.39, stdev= 1.47
     lat (usec): min=35, max=156, avg=46.17, stdev= 4.62      |	     lat (usec): min=20, max=192, avg=47.59, stdev= 1.48
     |  1.00th=[   42],  5.00th=[   42], 10.00th=[   43], 20. |	     |  1.00th=[43776],  5.00th=[44288], 10.00th=[44288], 20.
     | 30.00th=[   43], 40.00th=[   43], 50.00th=[   43], 60. |	     | 30.00th=[44800], 40.00th=[45312], 50.00th=[45312], 60.
     | 70.00th=[   44], 80.00th=[   44], 90.00th=[   45], 95. |	     | 70.00th=[45312], 80.00th=[45824], 90.00th=[45824], 95.
     | 99.00th=[   67], 99.50th=[   74], 99.90th=[   86], 99. |	     | 99.00th=[49408], 99.50th=[54528], 99.90th=[64256], 99.
     | 99.99th=[  117]					      |	     | 99.99th=[70144]
Type=read,Size=4K,NumJobs=1,depth=64				Type=read,Size=4K,NumJobs=1,depth=64
   read: IOPS=257k, BW=1004MiB/s (1053MB/s)(9.80GiB/10001msec |	   read: IOPS=338k, BW=1319MiB/s (1383MB/s)(12.9GiB/10001msec
    slat (nsec): min=1562, max=50661, avg=2614.87, stdev=2515 |	    slat (nsec): min=1621, max=52632, avg=2145.29, stdev=209.
    clat (usec): min=49, max=325, avg=245.85, stdev=11.72     |	    clat (usec): min=36, max=236, avg=187.02, stdev= 2.05
     lat (usec): min=50, max=328, avg=248.54, stdev=11.76     |	     lat (usec): min=38, max=238, avg=189.20, stdev= 2.06
     |  1.00th=[  219],  5.00th=[  227], 10.00th=[  231], 20. |	     |  1.00th=[  184],  5.00th=[  184], 10.00th=[  186], 20.
     | 30.00th=[  241], 40.00th=[  243], 50.00th=[  245], 60. |	     | 30.00th=[  186], 40.00th=[  188], 50.00th=[  188], 60.
     | 70.00th=[  253], 80.00th=[  255], 90.00th=[  262], 95. |	     | 70.00th=[  188], 80.00th=[  188], 90.00th=[  190], 95.
     | 99.00th=[  273], 99.50th=[  273], 99.90th=[  281], 99. |	     | 99.00th=[  192], 99.50th=[  196], 99.90th=[  200], 99.
     | 99.99th=[  285]					      |	     | 99.99th=[  217]
Type=read,Size=4K,NumJobs=16,depth=1				Type=read,Size=4K,NumJobs=16,depth=1
   read: IOPS=849k, BW=3318MiB/s (3479MB/s)(32.4GiB/10002msec |	   read: IOPS=869k, BW=3396MiB/s (3561MB/s)(33.2GiB/10002msec
    slat (nsec): min=1894, max=65881, avg=2560.26, stdev=657. |	    slat (nsec): min=1890, max=70947, avg=2497.58, stdev=813.
    clat (nsec): min=516, max=708489, avg=15808.69, stdev=196 |	    clat (nsec): min=590, max=79209, avg=15456.14, stdev=1968
     lat (usec): min=13, max=710, avg=18.42, stdev= 2.19      |	     lat (usec): min=12, max=110, avg=18.01, stdev= 2.18
     |  1.00th=[13120],  5.00th=[14016], 10.00th=[14272], 20. |	     |  1.00th=[12352],  5.00th=[13888], 10.00th=[14144], 20.
     | 30.00th=[15040], 40.00th=[15424], 50.00th=[15680], 60. |	     | 30.00th=[14528], 40.00th=[14784], 50.00th=[15040], 60.
     | 70.00th=[16320], 80.00th=[16768], 90.00th=[17280], 95. |	     | 70.00th=[15552], 80.00th=[16192], 90.00th=[17280], 95.
     | 99.00th=[19840], 99.50th=[21120], 99.90th=[34560], 99. |	     | 99.00th=[23936], 99.50th=[25728], 99.90th=[33536], 99.
     | 99.99th=[76288]					      |	     | 99.99th=[36096]
Type=read,Size=4K,NumJobs=16,depth=16				Type=read,Size=4K,NumJobs=16,depth=16
   read: IOPS=2593k, BW=9.89GiB/s (10.6GB/s)(98.9GiB/10002mse |	   read: IOPS=1843k, BW=7201MiB/s (7551MB/s)(70.3GiB/10002mse
    slat (nsec): min=1568, max=163092, avg=2599.50, stdev=145 |	    slat (nsec): min=1582, max=104435, avg=3170.92, stdev=170
    clat (usec): min=34, max=4287.1k, avg=95.73, stdev=9113.1 |	    clat (usec): min=40, max=554, avg=135.18, stdev=32.57
     lat (usec): min=36, max=4287.1k, avg=98.38, stdev=9113.1 |	     lat (usec): min=42, max=557, avg=138.42, stdev=32.59
     |  1.00th=[   43],  5.00th=[   53], 10.00th=[   57], 20. |	     |  1.00th=[   68],  5.00th=[   92], 10.00th=[  104], 20.
     | 30.00th=[   68], 40.00th=[   71], 50.00th=[   73], 60. |	     | 30.00th=[  125], 40.00th=[  130], 50.00th=[  135], 60.
     | 70.00th=[   80], 80.00th=[   88], 90.00th=[   96], 95. |	     | 70.00th=[  141], 80.00th=[  147], 90.00th=[  161], 95.
     | 99.00th=[  104], 99.50th=[  106], 99.90th=[  126], 99. |	     | 99.00th=[  262], 99.50th=[  306], 99.90th=[  392], 99.
     | 99.99th=[  176]					      |	     | 99.99th=[  461]
Type=read,Size=4K,NumJobs=16,depth=64				Type=read,Size=4K,NumJobs=16,depth=64
   read: IOPS=2646k, BW=10.1GiB/s (10.8GB/s)(101GiB/10001msec |	   read: IOPS=1917k, BW=7487MiB/s (7850MB/s)(73.1GiB/10001mse
    slat (nsec): min=1566, max=100447, avg=2310.80, stdev=263 |	    slat (nsec): min=1566, max=653412, avg=3347.14, stdev=385
    clat (usec): min=21, max=1440, avg=384.23, stdev=61.42    |	    clat (usec): min=43, max=1445, avg=530.39, stdev=127.84
     lat (usec): min=24, max=1443, avg=386.58, stdev=62.18    |	     lat (usec): min=45, max=1619, avg=533.81, stdev=128.49
     |  1.00th=[  330],  5.00th=[  347], 10.00th=[  347], 20. |	     |  1.00th=[  188],  5.00th=[  351], 10.00th=[  420], 20.
     | 30.00th=[  363], 40.00th=[  371], 50.00th=[  379], 60. |	     | 30.00th=[  486], 40.00th=[  502], 50.00th=[  519], 60.
     | 70.00th=[  388], 80.00th=[  388], 90.00th=[  396], 95. |	     | 70.00th=[  553], 80.00th=[  578], 90.00th=[  644], 95.
     | 99.00th=[  717], 99.50th=[  725], 99.90th=[  750], 99. |	     | 99.00th=[ 1012], 99.50th=[ 1037], 99.90th=[ 1090], 99.
     | 99.99th=[  766]					      |	     | 99.99th=[ 1205]
Type=read,Size=4K,NumJobs=56,depth=1				Type=read,Size=4K,NumJobs=56,depth=1
   read: IOPS=1514k, BW=5912MiB/s (6200MB/s)(57.7GiB/10002mse |	   read: IOPS=1571k, BW=6138MiB/s (6436MB/s)(59.9GiB/10001mse
    slat (nsec): min=1722, max=911960, avg=2705.80, stdev=141 |	    slat (nsec): min=1693, max=505053, avg=2771.65, stdev=110
    clat (nsec): min=466, max=4359.5k, avg=33747.42, stdev=23 |	    clat (nsec): min=479, max=1348.1k, avg=32311.24, stdev=36
     lat (usec): min=13, max=4362, avg=36.52, stdev=23.37     |	     lat (usec): min=15, max=1350, avg=35.15, stdev= 3.81
     |  1.00th=[   20],  5.00th=[   24], 10.00th=[   27], 20. |	     |  1.00th=[26752],  5.00th=[28800], 10.00th=[29568], 20.
     | 30.00th=[   32], 40.00th=[   33], 50.00th=[   34], 60. |	     | 30.00th=[30848], 40.00th=[31360], 50.00th=[31872], 60.
     | 70.00th=[   35], 80.00th=[   36], 90.00th=[   37], 95. |	     | 70.00th=[33024], 80.00th=[33536], 90.00th=[35584], 95.
     | 99.00th=[   48], 99.50th=[  143], 99.90th=[  433], 99. |	     | 99.00th=[42240], 99.50th=[43776], 99.90th=[49920], 99.
     | 99.99th=[  594]					      |	     | 99.99th=[59136]
Type=read,Size=4K,NumJobs=56,depth=16				Type=read,Size=4K,NumJobs=56,depth=16
   read: IOPS=2381k, BW=9299MiB/s (9751MB/s)(97.7GiB/10756mse |	   read: IOPS=2009k, BW=7846MiB/s (8227MB/s)(76.6GiB/10001mse
    slat (nsec): min=1573, max=1580.7M, avg=2676.88, stdev=44 |	    slat (nsec): min=1566, max=379585, avg=3204.59, stdev=230
    clat (usec): min=17, max=4303.4k, avg=355.77, stdev=27892 |	    clat (usec): min=44, max=1761, avg=442.22, stdev=244.27
     lat (usec): min=20, max=4303.4k, avg=358.50, stdev=27901 |	     lat (usec): min=46, max=1764, avg=445.50, stdev=244.24
     |  1.00th=[   44],  5.00th=[   55], 10.00th=[   62], 20. |	     |  1.00th=[  161],  5.00th=[  194], 10.00th=[  215], 20.
     | 30.00th=[   72], 40.00th=[   75], 50.00th=[   81], 60. |	     | 30.00th=[  273], 40.00th=[  306], 50.00th=[  351], 60.
     | 70.00th=[  109], 80.00th=[  135], 90.00th=[  167], 95. |	     | 70.00th=[  506], 80.00th=[  644], 90.00th=[  832], 95.
     | 99.00th=[  285], 99.50th=[  400], 99.90th=[  506], 99. |	     | 99.00th=[ 1156], 99.50th=[ 1221], 99.90th=[ 1336], 99.
     | 99.99th=[ 1172]					      |	     | 99.99th=[ 1450]
Type=read,Size=4K,NumJobs=56,depth=64				Type=read,Size=4K,NumJobs=56,depth=64
   read: IOPS=2496k, BW=9749MiB/s (10.2GB/s)(95.2GiB/10003mse |	   read: IOPS=1801k, BW=7036MiB/s (7378MB/s)(68.7GiB/10002mse
    slat (nsec): min=1568, max=2721.3M, avg=14509.11, stdev=1 |	    slat (nsec): min=1586, max=2410.2k, avg=20136.27, stdev=8
    clat (usec): min=35, max=2725.6k, avg=1420.77, stdev=1481 |	    clat (usec): min=58, max=8215, avg=1968.75, stdev=947.22
     lat (usec): min=42, max=2725.9k, avg=1435.33, stdev=1491 |	     lat (usec): min=60, max=8632, avg=1988.96, stdev=962.84
     |  1.00th=[  570],  5.00th=[  611], 10.00th=[  676], 20. |	     |  1.00th=[  449],  5.00th=[  734], 10.00th=[  865], 20.
     | 30.00th=[  742], 40.00th=[ 1221], 50.00th=[ 1369], 60. |	     | 30.00th=[ 1418], 40.00th=[ 1729], 50.00th=[ 1909], 60.
     | 70.00th=[ 1434], 80.00th=[ 1500], 90.00th=[ 2147], 95. |	     | 70.00th=[ 2212], 80.00th=[ 2507], 90.00th=[ 3326], 95.
     | 99.00th=[ 3163], 99.50th=[ 3261], 99.90th=[ 3654], 99. |	     | 99.00th=[ 4817], 99.50th=[ 5080], 99.90th=[ 5735], 99.
     | 99.99th=[ 4686]					      |	     | 99.99th=[ 6587]
Type=read,Size=64K,NumJobs=1,depth=1				Type=read,Size=64K,NumJobs=1,depth=1
   read: IOPS=29.7k, BW=1854MiB/s (1944MB/s)(18.1GiB/10001mse |	   read: IOPS=29.8k, BW=1864MiB/s (1955MB/s)(18.2GiB/10001mse
    slat (nsec): min=4746, max=53468, avg=6046.72, stdev=2806 |	    slat (nsec): min=4768, max=95484, avg=6119.74, stdev=2779
    clat (usec): min=3, max=132, avg=27.14, stdev= 7.87	      |	    clat (nsec): min=1555, max=75556, avg=26881.73, stdev=781
     lat (usec): min=25, max=137, avg=33.23, stdev= 9.11      |	     lat (usec): min=25, max=102, avg=33.05, stdev= 9.07
     |  1.00th=[21888],  5.00th=[21888], 10.00th=[21888], 20. |	     |  1.00th=[21632],  5.00th=[21632], 10.00th=[21632], 20.
     | 30.00th=[22144], 40.00th=[22144], 50.00th=[22400], 60. |	     | 30.00th=[21888], 40.00th=[22144], 50.00th=[22144], 60.
     | 70.00th=[28288], 80.00th=[35072], 90.00th=[41216], 95. |	     | 70.00th=[28032], 80.00th=[34048], 90.00th=[40704], 95.
     | 99.00th=[47360], 99.50th=[49920], 99.90th=[61696], 99. |	     | 99.00th=[47360], 99.50th=[49920], 99.90th=[59648], 99.
     | 99.99th=[70144]						     | 99.99th=[70144]
Type=read,Size=64K,NumJobs=1,depth=16				Type=read,Size=64K,NumJobs=1,depth=16
   read: IOPS=172k, BW=10.5GiB/s (11.3GB/s)(105GiB/10001msec)	   read: IOPS=172k, BW=10.5GiB/s (11.3GB/s)(105GiB/10001msec)
    slat (nsec): min=3300, max=48724, avg=4702.06, stdev=397. |	    slat (nsec): min=3756, max=65673, avg=4930.28, stdev=334.
    clat (usec): min=59, max=241, avg=88.09, stdev= 4.24      |	    clat (usec): min=83, max=243, avg=87.86, stdev= 2.04
     lat (usec): min=65, max=277, avg=92.83, stdev= 4.22      |	     lat (usec): min=87, max=310, avg=92.83, stdev= 2.05
     |  1.00th=[   82],  5.00th=[   83], 10.00th=[   83], 20. |	     |  1.00th=[   86],  5.00th=[   87], 10.00th=[   87], 20.
     | 30.00th=[   87], 40.00th=[   88], 50.00th=[   88], 60. |	     | 30.00th=[   87], 40.00th=[   88], 50.00th=[   88], 60.
     | 70.00th=[   91], 80.00th=[   92], 90.00th=[   94], 95. |	     | 70.00th=[   88], 80.00th=[   89], 90.00th=[   91], 95.
     | 99.00th=[   98], 99.50th=[   99], 99.90th=[  111], 99. |	     | 99.00th=[   96], 99.50th=[   99], 99.90th=[  108], 99.
     | 99.99th=[  145]					      |	     | 99.99th=[  114]
Type=read,Size=64K,NumJobs=1,depth=64				Type=read,Size=64K,NumJobs=1,depth=64
   read: IOPS=172k, BW=10.5GiB/s (11.2GB/s)(105GiB/10001msec) |	   read: IOPS=168k, BW=10.2GiB/s (10.0GB/s)(102GiB/10001msec)
    slat (usec): min=2, max=420, avg= 4.14, stdev= 3.46	      |	    slat (usec): min=2, max=390, avg= 4.15, stdev= 1.56
    clat (usec): min=299, max=1058, avg=368.25, stdev=18.60   |	    clat (usec): min=37, max=947, avg=376.95, stdev=15.70
     lat (usec): min=306, max=1099, avg=372.44, stdev=18.47   |	     lat (usec): min=42, max=988, avg=381.15, stdev=16.25
     |  1.00th=[  326],  5.00th=[  343], 10.00th=[  351], 20. |	     |  1.00th=[  363],  5.00th=[  363], 10.00th=[  367], 20.
     | 30.00th=[  359], 40.00th=[  367], 50.00th=[  367], 60. |	     | 30.00th=[  367], 40.00th=[  367], 50.00th=[  371], 60.
     | 70.00th=[  375], 80.00th=[  379], 90.00th=[  388], 95. |	     | 70.00th=[  388], 80.00th=[  392], 90.00th=[  400], 95.
     | 99.00th=[  433], 99.50th=[  445], 99.90th=[  465], 99. |	     | 99.00th=[  412], 99.50th=[  416], 99.90th=[  429], 99.
     | 99.99th=[  758]					      |	     | 99.99th=[  668]
Type=read,Size=64K,NumJobs=16,depth=1				Type=read,Size=64K,NumJobs=16,depth=1
   read: IOPS=150k, BW=9390MiB/s (9847MB/s)(105GiB/11428msec) |	   read: IOPS=172k, BW=10.5GiB/s (11.3GB/s)(105GiB/10002msec)
    slat (nsec): min=3750, max=58669, avg=5120.99, stdev=2896 |	    slat (nsec): min=3480, max=74756, avg=5273.21, stdev=2961
    clat (nsec): min=494, max=2276.2M, avg=88365.74, stdev=17 |	    clat (nsec): min=616, max=453786, avg=87271.90, stdev=161
     lat (usec): min=26, max=2276.2k, avg=93.55, stdev=1737.1 |	     lat (usec): min=26, max=458, avg=92.60, stdev=16.20
     |  1.00th=[   36],  5.00th=[   55], 10.00th=[   68], 20. |	     |  1.00th=[   37],  5.00th=[   57], 10.00th=[   68], 20.
     | 30.00th=[   83], 40.00th=[   87], 50.00th=[   89], 60. |	     | 30.00th=[   83], 40.00th=[   87], 50.00th=[   90], 60.
     | 70.00th=[   95], 80.00th=[   98], 90.00th=[  103], 95. |	     | 70.00th=[   95], 80.00th=[   98], 90.00th=[  104], 95.
     | 99.00th=[  124], 99.50th=[  129], 99.90th=[  141], 99. |	     | 99.00th=[  126], 99.50th=[  130], 99.90th=[  143], 99.
     | 99.99th=[  165]					      |	     | 99.99th=[  161]
Type=read,Size=64K,NumJobs=16,depth=16				Type=read,Size=64K,NumJobs=16,depth=16
   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10003msec) |	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10002msec)
    slat (nsec): min=3284, max=53892, avg=4935.30, stdev=1716 |	    slat (nsec): min=3866, max=57337, avg=5204.47, stdev=1220
    clat (usec): min=102, max=4744, avg=1455.23, stdev=493.14 |	    clat (usec): min=116, max=3985, avg=1453.77, stdev=349.92
     lat (usec): min=116, max=4749, avg=1460.22, stdev=493.10 |	     lat (usec): min=128, max=3990, avg=1459.03, stdev=349.93
     |  1.00th=[ 1172],  5.00th=[ 1172], 10.00th=[ 1188], 20. |	     |  1.00th=[ 1254],  5.00th=[ 1270], 10.00th=[ 1270], 20.
     | 30.00th=[ 1287], 40.00th=[ 1352], 50.00th=[ 1352], 60. |	     | 30.00th=[ 1352], 40.00th=[ 1352], 50.00th=[ 1352], 60.
     | 70.00th=[ 1369], 80.00th=[ 1385], 90.00th=[ 1401], 95. |	     | 70.00th=[ 1369], 80.00th=[ 1401], 90.00th=[ 1467], 95.
     | 99.00th=[ 3818], 99.50th=[ 3851], 99.90th=[ 3851], 99. |	     | 99.00th=[ 2737], 99.50th=[ 2769], 99.90th=[ 2769], 99.
     | 99.99th=[ 3884]					      |	     | 99.99th=[ 2802]
Type=read,Size=64K,NumJobs=16,depth=64				Type=read,Size=64K,NumJobs=16,depth=64
   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10006msec)	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10006msec)
    slat (usec): min=2, max=803, avg=16.08, stdev=83.27	      |	    slat (usec): min=2, max=1185, avg= 9.71, stdev=60.22
    clat (usec): min=109, max=15039, avg=5820.28, stdev=1740. |	    clat (usec): min=116, max=16490, avg=5824.81, stdev=1345.
     lat (usec): min=116, max=15042, avg=5836.41, stdev=1769. |	     lat (usec): min=133, max=16494, avg=5834.58, stdev=1367.
     |  1.00th=[ 4686],  5.00th=[ 4686], 10.00th=[ 4752], 20. |	     |  1.00th=[ 5407],  5.00th=[ 5407], 10.00th=[ 5473], 20.
     | 30.00th=[ 4752], 40.00th=[ 5080], 50.00th=[ 5080], 60. |	     | 30.00th=[ 5473], 40.00th=[ 5473], 50.00th=[ 5473], 60.
     | 70.00th=[ 5473], 80.00th=[ 5866], 90.00th=[ 9503], 95. |	     | 70.00th=[ 5473], 80.00th=[ 5473], 90.00th=[ 5538], 95.
     | 99.00th=[10552], 99.50th=[10945], 99.90th=[10945], 99. |	     | 99.00th=[10945], 99.50th=[10945], 99.90th=[11076], 99.
     | 99.99th=[11076]					      |	     | 99.99th=[11338]
Type=read,Size=64K,NumJobs=56,depth=1				Type=read,Size=64K,NumJobs=56,depth=1
   read: IOPS=160k, BW=9979MiB/s (10.5GB/s)(107GiB/10962msec) |	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10002msec)
    slat (nsec): min=3668, max=66111, avg=4685.24, stdev=1778 |	    slat (nsec): min=3801, max=63985, avg=4896.45, stdev=1489
    clat (usec): min=36, max=3544.7k, avg=317.51, stdev=6531. |	    clat (usec): min=29, max=550, avg=314.45, stdev=76.21
     lat (usec): min=42, max=3544.7k, avg=322.25, stdev=6531. |	     lat (usec): min=35, max=555, avg=319.41, stdev=76.15
     |  1.00th=[  161],  5.00th=[  169], 10.00th=[  174], 20. |	     |  1.00th=[  174],  5.00th=[  178], 10.00th=[  194], 20.
     | 30.00th=[  285], 40.00th=[  293], 50.00th=[  297], 60. |	     | 30.00th=[  302], 40.00th=[  306], 50.00th=[  306], 60.
     | 70.00th=[  306], 80.00th=[  318], 90.00th=[  445], 95. |	     | 70.00th=[  310], 80.00th=[  322], 90.00th=[  457], 95.
     | 99.00th=[  603], 99.50th=[  611], 99.90th=[  627], 99. |	     | 99.00th=[  474], 99.50th=[  482], 99.90th=[  490], 99.
     | 99.99th=[  644]					      |	     | 99.99th=[  506]
Type=read,Size=64K,NumJobs=56,depth=16				Type=read,Size=64K,NumJobs=56,depth=16
   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10007msec) |	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10006msec)
    slat (usec): min=3, max=171, avg= 4.99, stdev= 2.81	      |	    slat (usec): min=3, max=251, avg= 5.40, stdev= 2.34
    clat (usec): min=784, max=14310, avg=5117.02, stdev=1736. |	    clat (usec): min=236, max=14486, avg=5106.35, stdev=1241.
     lat (usec): min=884, max=14315, avg=5122.07, stdev=1736. |	     lat (usec): min=265, max=14491, avg=5111.80, stdev=1241.
     |  1.00th=[ 2409],  5.00th=[ 2442], 10.00th=[ 2474], 20. |	     |  1.00th=[ 2442],  5.00th=[ 2540], 10.00th=[ 4883], 20.
     | 30.00th=[ 4883], 40.00th=[ 4948], 50.00th=[ 4948], 60. |	     | 30.00th=[ 4948], 40.00th=[ 5080], 50.00th=[ 5080], 60.
     | 70.00th=[ 5014], 80.00th=[ 5145], 90.00th=[ 7439], 95. |	     | 70.00th=[ 5080], 80.00th=[ 5145], 90.00th=[ 7373], 95.
     | 99.00th=[ 9896], 99.50th=[ 9896], 99.90th=[10159], 99. |	     | 99.00th=[ 9896], 99.50th=[ 9896], 99.90th=[ 9896], 99.
     | 99.99th=[10290]					      |	     | 99.99th=[10028]
Type=read,Size=64K,NumJobs=56,depth=64				Type=read,Size=64K,NumJobs=56,depth=64
   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10016msec) |	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10018msec)
    slat (usec): min=2, max=3494.5k, avg=194.57, stdev=10422. |	    slat (usec): min=2, max=10184, avg=189.22, stdev=637.78
    clat (usec): min=138, max=3513.0k, avg=20256.38, stdev=11 |	    clat (usec): min=347, max=61952, avg=20252.15, stdev=7056
     lat (usec): min=178, max=3513.0k, avg=20451.00, stdev=11 |	     lat (usec): min=351, max=65141, avg=20441.44, stdev=7203
     |  1.00th=[    5],  5.00th=[    5], 10.00th=[    6], 20. |	     |  1.00th=[ 9503],  5.00th=[ 9765], 10.00th=[ 9896], 20.
     | 30.00th=[   10], 40.00th=[   15], 50.00th=[   19], 60. |	     | 30.00th=[19530], 40.00th=[19530], 50.00th=[19792], 60.
     | 70.00th=[   20], 80.00th=[   20], 90.00th=[   28], 95. |	     | 70.00th=[20317], 80.00th=[20579], 90.00th=[29492], 95.
     | 99.00th=[   40], 99.50th=[   40], 99.90th=[ 3473], 99. |	     | 99.00th=[41157], 99.50th=[58983], 99.90th=[59507], 99.
     | 99.99th=[ 3507]					      |	     | 99.99th=[61080]
Type=write,Size=512B,NumJobs=1,depth=1				Type=write,Size=512B,NumJobs=1,depth=1
  write: IOPS=46.5k, BW=22.7MiB/s (23.8MB/s)(227MiB/10001msec |	  write: IOPS=56.0k, BW=27.4MiB/s (28.7MB/s)(274MiB/10001msec
    slat (nsec): min=2038, max=46160, avg=3144.53, stdev=2438 |	    slat (nsec): min=2112, max=43957, avg=2561.65, stdev=1423
    clat (nsec): min=449, max=133529, avg=17730.81, stdev=670 |	    clat (nsec): min=633, max=60031, avg=14812.68, stdev=5638
     lat (usec): min=13, max=136, avg=20.95, stdev= 7.40      |	     lat (nsec): min=12369, max=72124, avg=17416.39, stdev=60
     |  1.00th=[13504],  5.00th=[13504], 10.00th=[13632], 20. |	     |  1.00th=[12352],  5.00th=[12352], 10.00th=[12480], 20.
     | 30.00th=[13760], 40.00th=[13888], 50.00th=[14016], 60. |	     | 30.00th=[12480], 40.00th=[12608], 50.00th=[12608], 60.
     | 70.00th=[17792], 80.00th=[20608], 90.00th=[28800], 95. |	     | 70.00th=[12864], 80.00th=[14656], 90.00th=[20096], 95.
     | 99.00th=[41728], 99.50th=[45824], 99.90th=[51456], 99. |	     | 99.00th=[35072], 99.50th=[36608], 99.90th=[47360], 99.
     | 99.99th=[64256]					      |	     | 99.99th=[54016]
Type=write,Size=512B,NumJobs=1,depth=16				Type=write,Size=512B,NumJobs=1,depth=16
  write: IOPS=255k, BW=125MiB/s (131MB/s)(1248MiB/10001msec)  |	  write: IOPS=328k, BW=160MiB/s (168MB/s)(1601MiB/10001msec)
    slat (nsec): min=1628, max=69589, avg=2649.87, stdev=2876 |	    slat (nsec): min=1641, max=32941, avg=2199.97, stdev=236.
    clat (usec): min=23, max=182, avg=59.49, stdev= 8.44      |	    clat (usec): min=14, max=113, avg=46.29, stdev= 1.27
     lat (usec): min=25, max=184, avg=62.20, stdev= 8.26      |	     lat (usec): min=16, max=115, avg=48.52, stdev= 1.28
     |  1.00th=[   45],  5.00th=[   46], 10.00th=[   48], 20. |	     |  1.00th=[43776],  5.00th=[44800], 10.00th=[45312], 20.
     | 30.00th=[   58], 40.00th=[   59], 50.00th=[   60], 60. |	     | 30.00th=[45824], 40.00th=[45824], 50.00th=[46336], 60.
     | 70.00th=[   62], 80.00th=[   65], 90.00th=[   70], 95. |	     | 70.00th=[46336], 80.00th=[46848], 90.00th=[47360], 95.
     | 99.00th=[   85], 99.50th=[   90], 99.90th=[  103], 99. |	     | 99.00th=[50432], 99.50th=[53504], 99.90th=[57600], 99.
     | 99.99th=[  139]					      |	     | 99.99th=[63744]
Type=write,Size=512B,NumJobs=1,depth=64				Type=write,Size=512B,NumJobs=1,depth=64
  write: IOPS=262k, BW=128MiB/s (134MB/s)(1277MiB/10001msec)  |	  write: IOPS=337k, BW=164MiB/s (172MB/s)(1644MiB/10001msec)
    slat (nsec): min=1626, max=45520, avg=2718.39, stdev=3134 |	    slat (nsec): min=1642, max=43437, avg=2146.30, stdev=234.
    clat (usec): min=31, max=304, avg=241.50, stdev=10.69     |	    clat (usec): min=16, max=235, avg=187.62, stdev= 2.46
     lat (usec): min=33, max=307, avg=244.27, stdev=10.71     |	     lat (usec): min=18, max=237, avg=189.80, stdev= 2.48
     |  1.00th=[  221],  5.00th=[  225], 10.00th=[  229], 20. |	     |  1.00th=[  184],  5.00th=[  184], 10.00th=[  186], 20.
     | 30.00th=[  235], 40.00th=[  237], 50.00th=[  241], 60. |	     | 30.00th=[  186], 40.00th=[  188], 50.00th=[  188], 60.
     | 70.00th=[  247], 80.00th=[  251], 90.00th=[  258], 95. |	     | 70.00th=[  190], 80.00th=[  190], 90.00th=[  192], 95.
     | 99.00th=[  265], 99.50th=[  265], 99.90th=[  269], 99. |	     | 99.00th=[  194], 99.50th=[  196], 99.90th=[  200], 99.
     | 99.99th=[  277]					      |	     | 99.99th=[  204]
Type=write,Size=512B,NumJobs=16,depth=1				Type=write,Size=512B,NumJobs=16,depth=1
  write: IOPS=886k, BW=433MiB/s (454MB/s)(4327MiB/10002msec)  |	  write: IOPS=914k, BW=446MiB/s (468MB/s)(4462MiB/10001msec)
    slat (nsec): min=1827, max=51604, avg=2633.04, stdev=591. |	    slat (nsec): min=1818, max=83376, avg=2566.34, stdev=824.
    clat (nsec): min=515, max=695999, avg=14905.52, stdev=209 |	    clat (nsec): min=465, max=343253, avg=14459.07, stdev=181
     lat (usec): min=9, max=698, avg=17.60, stdev= 2.26	      |	     lat (usec): min=11, max=345, avg=17.08, stdev= 2.09
     |  1.00th=[12736],  5.00th=[13376], 10.00th=[13632], 20. |	     |  1.00th=[12224],  5.00th=[12992], 10.00th=[13248], 20.
     | 30.00th=[14400], 40.00th=[14528], 50.00th=[14784], 60. |	     | 30.00th=[13632], 40.00th=[13760], 50.00th=[14016], 60.
     | 70.00th=[15168], 80.00th=[15552], 90.00th=[16064], 95. |	     | 70.00th=[14528], 80.00th=[15040], 90.00th=[16320], 95.
     | 99.00th=[18560], 99.50th=[19840], 99.90th=[28544], 99. |	     | 99.00th=[21888], 99.50th=[23936], 99.90th=[32384], 99.
     | 99.99th=[87552]					      |	     | 99.99th=[35584]
Type=write,Size=512B,NumJobs=16,depth=16			Type=write,Size=512B,NumJobs=16,depth=16
  write: IOPS=4155k, BW=2029MiB/s (2127MB/s)(19.8GiB/10002mse |	  write: IOPS=3218k, BW=1571MiB/s (1648MB/s)(15.3GiB/10001mse
    slat (nsec): min=1622, max=171986, avg=2797.12, stdev=184 |	    slat (nsec): min=1621, max=154997, avg=3548.60, stdev=323
    clat (usec): min=18, max=1041, avg=58.37, stdev=10.11     |	    clat (usec): min=14, max=473, avg=75.44, stdev=65.86
     lat (usec): min=20, max=1043, avg=61.22, stdev=10.54     |	     lat (usec): min=17, max=489, avg=79.05, stdev=68.82
     |  1.00th=[   44],  5.00th=[   45], 10.00th=[   47], 20. |	     |  1.00th=[   45],  5.00th=[   45], 10.00th=[   46], 20.
     | 30.00th=[   50], 40.00th=[   56], 50.00th=[   60], 60. |	     | 30.00th=[   47], 40.00th=[   48], 50.00th=[   52], 60.
     | 70.00th=[   65], 80.00th=[   67], 90.00th=[   73], 95. |	     | 70.00th=[   61], 80.00th=[   64], 90.00th=[  178], 95.
     | 99.00th=[   81], 99.50th=[   82], 99.90th=[   86], 99. |	     | 99.00th=[  326], 99.50th=[  334], 99.90th=[  351], 99.
     | 99.99th=[  293]					      |	     | 99.99th=[  367]
Type=write,Size=512B,NumJobs=16,depth=64			Type=write,Size=512B,NumJobs=16,depth=64
  write: IOPS=4304k, BW=2102MiB/s (2204MB/s)(20.5GiB/10001mse |	  write: IOPS=3973k, BW=1940MiB/s (2034MB/s)(18.0GiB/10002mse
    slat (nsec): min=1621, max=167405, avg=2712.72, stdev=195 |	    slat (nsec): min=1627, max=171980, avg=2926.70, stdev=209
    clat (usec): min=23, max=1767, avg=234.76, stdev=51.39    |	    clat (usec): min=10, max=1606, avg=254.29, stdev=176.36
     lat (usec): min=25, max=1788, avg=237.52, stdev=52.04    |	     lat (usec): min=16, max=1615, avg=257.27, stdev=178.32
     |  1.00th=[  180],  5.00th=[  182], 10.00th=[  184], 20. |	     |  1.00th=[  184],  5.00th=[  186], 10.00th=[  188], 20.
     | 30.00th=[  188], 40.00th=[  215], 50.00th=[  237], 60. |	     | 30.00th=[  194], 40.00th=[  202], 50.00th=[  237], 60.
     | 70.00th=[  260], 80.00th=[  265], 90.00th=[  310], 95. |	     | 70.00th=[  255], 80.00th=[  260], 90.00th=[  293], 95.
     | 99.00th=[  359], 99.50th=[  404], 99.90th=[  412], 99. |	     | 99.00th=[ 1385], 99.50th=[ 1418], 99.90th=[ 1467], 99.
     | 99.99th=[ 1565]					      |	     | 99.99th=[ 1500]
Type=write,Size=512B,NumJobs=56,depth=1				Type=write,Size=512B,NumJobs=56,depth=1
  write: IOPS=2325k, BW=1135MiB/s (1191MB/s)(11.1GiB/10001mse |	  write: IOPS=2369k, BW=1157MiB/s (1213MB/s)(11.3GiB/10002mse
    slat (nsec): min=1761, max=894949, avg=3151.04, stdev=221 |	    slat (nsec): min=1823, max=501953, avg=3567.43, stdev=201
    clat (nsec): min=443, max=1452.0k, avg=20283.73, stdev=15 |	    clat (nsec): min=449, max=1722.5k, avg=19318.77, stdev=42
     lat (usec): min=12, max=1454, avg=23.51, stdev=16.01     |	     lat (usec): min=11, max=1724, avg=22.98, stdev= 4.83
     |  1.00th=[   14],  5.00th=[   16], 10.00th=[   17], 20. |	     |  1.00th=[12992],  5.00th=[14656], 10.00th=[15680], 20.
     | 30.00th=[   19], 40.00th=[   19], 50.00th=[   20], 60. |	     | 30.00th=[17536], 40.00th=[18048], 50.00th=[18816], 60.
     | 70.00th=[   21], 80.00th=[   22], 90.00th=[   24], 95. |	     | 70.00th=[20352], 80.00th=[21632], 90.00th=[23424], 95.
     | 99.00th=[   34], 99.50th=[   40], 99.90th=[  347], 99. |	     | 99.00th=[31872], 99.50th=[35584], 99.90th=[41728], 99.
     | 99.99th=[  519]					      |	     | 99.99th=[54016]
Type=write,Size=512B,NumJobs=56,depth=16			Type=write,Size=512B,NumJobs=56,depth=16
  write: IOPS=7222k, BW=3526MiB/s (3698MB/s)(34.4GiB/10001mse |	  write: IOPS=6237k, BW=3046MiB/s (3193MB/s)(29.7GiB/10002mse
    slat (nsec): min=1630, max=325931, avg=4224.26, stdev=656 |	    slat (nsec): min=1640, max=2381.5k, avg=5462.05, stdev=67
    clat (usec): min=15, max=5606, avg=119.11, stdev=48.84    |	    clat (usec): min=5, max=10435, avg=137.19, stdev=82.68
     lat (usec): min=26, max=5610, avg=123.43, stdev=49.91    |	     lat (usec): min=15, max=10447, avg=142.77, stdev=85.58
     |  1.00th=[   68],  5.00th=[   74], 10.00th=[   78], 20. |	     |  1.00th=[   63],  5.00th=[   68], 10.00th=[   70], 20.
     | 30.00th=[   93], 40.00th=[   99], 50.00th=[  104], 60. |	     | 30.00th=[   85], 40.00th=[   96], 50.00th=[  104], 60.
     | 70.00th=[  126], 80.00th=[  145], 90.00th=[  176], 95. |	     | 70.00th=[  149], 80.00th=[  210], 90.00th=[  265], 95.
     | 99.00th=[  318], 99.50th=[  351], 99.90th=[  420], 99. |	     | 99.00th=[  363], 99.50th=[  396], 99.90th=[  553], 99.
     | 99.99th=[  603]					      |	     | 99.99th=[  873]
Type=write,Size=512B,NumJobs=56,depth=64			Type=write,Size=512B,NumJobs=56,depth=64
  write: IOPS=8407k, BW=4105MiB/s (4304MB/s)(40.1GiB/10002mse |	  write: IOPS=7488k, BW=3656MiB/s (3834MB/s)(35.7GiB/10002mse
    slat (nsec): min=1642, max=22197k, avg=4988.96, stdev=171 |	    slat (nsec): min=1666, max=7929.3k, avg=5382.53, stdev=12
    clat (usec): min=21, max=22624, avg=420.46, stdev=251.29  |	    clat (usec): min=14, max=12517, avg=472.19, stdev=331.30
     lat (usec): min=25, max=22628, avg=425.55, stdev=253.99  |	     lat (usec): min=17, max=12520, avg=477.70, stdev=334.84
     |  1.00th=[  277],  5.00th=[  293], 10.00th=[  302], 20. |	     |  1.00th=[  265],  5.00th=[  277], 10.00th=[  289], 20.
     | 30.00th=[  322], 40.00th=[  338], 50.00th=[  359], 60. |	     | 30.00th=[  302], 40.00th=[  306], 50.00th=[  314], 60.
     | 70.00th=[  404], 80.00th=[  441], 90.00th=[  482], 95. |	     | 70.00th=[  347], 80.00th=[  562], 90.00th=[ 1057], 95.
     | 99.00th=[ 1729], 99.50th=[ 1860], 99.90th=[ 2024], 99. |	     | 99.00th=[ 1434], 99.50th=[ 1696], 99.90th=[ 2040], 99.
     | 99.99th=[ 2606]					      |	     | 99.99th=[ 2474]
Type=write,Size=4K,NumJobs=1,depth=1				Type=write,Size=4K,NumJobs=1,depth=1
  write: IOPS=44.4k, BW=173MiB/s (182MB/s)(1735MiB/10001msec) |	  write: IOPS=55.8k, BW=218MiB/s (229MB/s)(2180MiB/10000msec)
    slat (nsec): min=2009, max=36370, avg=3363.25, stdev=2555 |	    slat (nsec): min=2138, max=44613, avg=2552.61, stdev=1404
    clat (nsec): min=453, max=106416, avg=18501.98, stdev=678 |	    clat (nsec): min=627, max=624863, avg=14910.38, stdev=517
     lat (usec): min=14, max=109, avg=21.94, stdev= 7.54      |	     lat (usec): min=13, max=627, avg=17.50, stdev= 5.55
     |  1.00th=[12864],  5.00th=[14016], 10.00th=[14144], 20. |	     |  1.00th=[12736],  5.00th=[12864], 10.00th=[12864], 20.
     | 30.00th=[14272], 40.00th=[14400], 50.00th=[14656], 60. |	     | 30.00th=[12992], 40.00th=[12992], 50.00th=[12992], 60.
     | 70.00th=[18816], 80.00th=[21632], 90.00th=[30080], 95. |	     | 70.00th=[13248], 80.00th=[13632], 90.00th=[20352], 95.
     | 99.00th=[42240], 99.50th=[46336], 99.90th=[51456], 99. |	     | 99.00th=[35584], 99.50th=[37120], 99.90th=[44800], 99.
     | 99.99th=[64768]					      |	     | 99.99th=[53504]
Type=write,Size=4K,NumJobs=1,depth=16				Type=write,Size=4K,NumJobs=1,depth=16
  write: IOPS=245k, BW=956MiB/s (1002MB/s)(9556MiB/10001msec) |	  write: IOPS=325k, BW=1271MiB/s (1332MB/s)(12.4GiB/10001msec
    slat (nsec): min=1646, max=69111, avg=2966.78, stdev=2405 |	    slat (nsec): min=1697, max=32986, avg=2232.67, stdev=205.
    clat (usec): min=28, max=146, avg=61.95, stdev= 5.61      |	    clat (nsec): min=14620, max=89670, avg=46638.22, stdev=11
     lat (usec): min=31, max=148, avg=64.97, stdev= 5.64      |	     lat (nsec): min=16814, max=91833, avg=48907.48, stdev=11
     |  1.00th=[   51],  5.00th=[   56], 10.00th=[   57], 20. |	     |  1.00th=[44800],  5.00th=[45312], 10.00th=[45824], 20.
     | 30.00th=[   59], 40.00th=[   60], 50.00th=[   62], 60. |	     | 30.00th=[46336], 40.00th=[46336], 50.00th=[46336], 60.
     | 70.00th=[   65], 80.00th=[   66], 90.00th=[   69], 95. |	     | 70.00th=[46848], 80.00th=[46848], 90.00th=[47360], 95.
     | 99.00th=[   81], 99.50th=[   83], 99.90th=[   90], 99. |	     | 99.00th=[50944], 99.50th=[53504], 99.90th=[58112], 99.
     | 99.99th=[  108]					      |	     | 99.99th=[63744]
Type=write,Size=4K,NumJobs=1,depth=64				Type=write,Size=4K,NumJobs=1,depth=64
  write: IOPS=350k, BW=1367MiB/s (1434MB/s)(13.4GiB/10001msec |	  write: IOPS=322k, BW=1259MiB/s (1320MB/s)(12.3GiB/10001msec
    slat (nsec): min=1669, max=39786, avg=2055.37, stdev=197. |	    slat (nsec): min=1697, max=33655, avg=2273.54, stdev=208.
    clat (usec): min=21, max=222, avg=180.47, stdev= 1.67     |	    clat (usec): min=18, max=238, avg=195.95, stdev= 2.07
     lat (usec): min=23, max=224, avg=182.56, stdev= 1.69     |	     lat (usec): min=20, max=241, avg=198.26, stdev= 2.08
     |  1.00th=[  178],  5.00th=[  178], 10.00th=[  180], 20. |	     |  1.00th=[  192],  5.00th=[  194], 10.00th=[  194], 20.
     | 30.00th=[  180], 40.00th=[  180], 50.00th=[  182], 60. |	     | 30.00th=[  196], 40.00th=[  196], 50.00th=[  196], 60.
     | 70.00th=[  182], 80.00th=[  182], 90.00th=[  184], 95. |	     | 70.00th=[  198], 80.00th=[  198], 90.00th=[  198], 95.
     | 99.00th=[  186], 99.50th=[  186], 99.90th=[  190], 99. |	     | 99.00th=[  202], 99.50th=[  204], 99.90th=[  208], 99.
     | 99.99th=[  194]					      |	     | 99.99th=[  212]
Type=write,Size=4K,NumJobs=16,depth=1				Type=write,Size=4K,NumJobs=16,depth=1
  write: IOPS=851k, BW=3325MiB/s (3486MB/s)(32.5GiB/10001msec |	  write: IOPS=855k, BW=3339MiB/s (3502MB/s)(32.6GiB/10001msec
    slat (nsec): min=1844, max=67596, avg=2580.48, stdev=494. |	    slat (nsec): min=1832, max=65363, avg=2579.38, stdev=826.
    clat (nsec): min=670, max=351867, avg=15732.23, stdev=195 |	    clat (nsec): min=472, max=134403, avg=15646.36, stdev=196
     lat (usec): min=12, max=354, avg=18.37, stdev= 2.08      |	     lat (usec): min=12, max=136, avg=18.28, stdev= 2.21
     |  1.00th=[12608],  5.00th=[14016], 10.00th=[14400], 20. |	     |  1.00th=[13120],  5.00th=[13760], 10.00th=[14016], 20.
     | 30.00th=[15168], 40.00th=[15296], 50.00th=[15552], 60. |	     | 30.00th=[14656], 40.00th=[14912], 50.00th=[15168], 60.
     | 70.00th=[16064], 80.00th=[16512], 90.00th=[17024], 95. |	     | 70.00th=[15936], 80.00th=[16512], 90.00th=[17536], 95.
     | 99.00th=[20096], 99.50th=[22144], 99.90th=[34048], 99. |	     | 99.00th=[23680], 99.50th=[25728], 99.90th=[33536], 99.
     | 99.99th=[89600]					      |	     | 99.99th=[37120]
Type=write,Size=4K,NumJobs=16,depth=16				Type=write,Size=4K,NumJobs=16,depth=16
  write: IOPS=1885k, BW=7364MiB/s (7722MB/s)(71.9GiB/10001mse |	  write: IOPS=1826k, BW=7133MiB/s (7480MB/s)(69.7GiB/10001mse
    slat (nsec): min=1629, max=135927, avg=2178.04, stdev=783 |	    slat (nsec): min=1639, max=113071, avg=3312.71, stdev=236
    clat (usec): min=40, max=387, avg=133.25, stdev=31.90     |	    clat (usec): min=40, max=454, avg=136.35, stdev=35.22
     lat (usec): min=44, max=391, avg=135.47, stdev=31.86     |	     lat (usec): min=43, max=456, avg=139.72, stdev=35.40
     |  1.00th=[   89],  5.00th=[   99], 10.00th=[  103], 20. |	     |  1.00th=[   72],  5.00th=[   91], 10.00th=[  101], 20.
     | 30.00th=[  116], 40.00th=[  125], 50.00th=[  131], 60. |	     | 30.00th=[  121], 40.00th=[  127], 50.00th=[  133], 60.
     | 70.00th=[  139], 80.00th=[  143], 90.00th=[  176], 95. |	     | 70.00th=[  143], 80.00th=[  149], 90.00th=[  182], 95.
     | 99.00th=[  253], 99.50th=[  269], 99.90th=[  289], 99. |	     | 99.00th=[  258], 99.50th=[  273], 99.90th=[  314], 99.
     | 99.99th=[  310]					      |	     | 99.99th=[  367]
Type=write,Size=4K,NumJobs=16,depth=64				Type=write,Size=4K,NumJobs=16,depth=64
  write: IOPS=481k, BW=1880MiB/s (1971MB/s)(23.6GiB/12877msec |	  write: IOPS=1875k, BW=7325MiB/s (7680MB/s)(71.5GiB/10001mse
    slat (nsec): min=1635, max=3356.0M, avg=3206.62, stdev=13 |	    slat (nsec): min=1638, max=627057, avg=3232.91, stdev=324
    clat (usec): min=27, max=7636.9k, avg=1985.08, stdev=8553 |	    clat (usec): min=49, max=1357, avg=542.32, stdev=162.71
     lat (usec): min=29, max=7636.9k, avg=1988.34, stdev=8555 |	     lat (usec): min=52, max=1384, avg=545.61, stdev=162.80
     |  1.00th=[    178],  5.00th=[    180], 10.00th=[    180 |	     |  1.00th=[  202],  5.00th=[  277], 10.00th=[  330], 20.
     | 20.00th=[    182], 30.00th=[    192], 40.00th=[    233 |	     | 30.00th=[  453], 40.00th=[  498], 50.00th=[  537], 60.
     | 50.00th=[    235], 60.00th=[    239], 70.00th=[    247 |	     | 70.00th=[  635], 80.00th=[  685], 90.00th=[  758], 95.
     | 80.00th=[    253], 90.00th=[    262], 95.00th=[    269 |	     | 99.00th=[  922], 99.50th=[  955], 99.90th=[ 1029], 99.
     | 99.00th=[    330], 99.50th=[    338], 99.90th=[    506 |	     | 99.99th=[ 1123]
     | 99.95th=[    832], 99.99th=[4328522]		      <
Type=write,Size=4K,NumJobs=56,depth=1				Type=write,Size=4K,NumJobs=56,depth=1
  write: IOPS=1456k, BW=5686MiB/s (5962MB/s)(55.5GiB/10002mse |	  write: IOPS=1416k, BW=5533MiB/s (5802MB/s)(54.0GiB/10002mse
    slat (nsec): min=1774, max=827924, avg=2804.02, stdev=137 |	    slat (nsec): min=1846, max=533231, avg=2881.71, stdev=111
    clat (nsec): min=462, max=3982.0k, avg=35108.12, stdev=23 |	    clat (nsec): min=498, max=4453.2k, avg=36057.87, stdev=38
     lat (usec): min=13, max=3987, avg=37.98, stdev=23.37     |	     lat (usec): min=15, max=4456, avg=39.01, stdev= 4.01
     |  1.00th=[   21],  5.00th=[   27], 10.00th=[   29], 20. |	     |  1.00th=[29312],  5.00th=[32128], 10.00th=[33024], 20.
     | 30.00th=[   33], 40.00th=[   34], 50.00th=[   35], 60. |	     | 30.00th=[34560], 40.00th=[35072], 50.00th=[35584], 60.
     | 70.00th=[   36], 80.00th=[   37], 90.00th=[   38], 95. |	     | 70.00th=[37120], 80.00th=[37632], 90.00th=[38656], 95.
     | 99.00th=[   52], 99.50th=[  141], 99.90th=[  437], 99. |	     | 99.00th=[49408], 99.50th=[52480], 99.90th=[56576], 99.
     | 99.99th=[  603]					      |	     | 99.99th=[64768]
Type=write,Size=4K,NumJobs=56,depth=16				Type=write,Size=4K,NumJobs=56,depth=16
  write: IOPS=1662k, BW=6493MiB/s (6808MB/s)(75.8GiB/11961mse |	  write: IOPS=1616k, BW=6314MiB/s (6621MB/s)(61.7GiB/10001mse
    slat (nsec): min=1629, max=3178.7M, avg=3532.32, stdev=16 |	    slat (nsec): min=1649, max=738008, avg=3245.81, stdev=163
    clat (usec): min=20, max=4303.6k, avg=500.29, stdev=35989 |	    clat (usec): min=44, max=5046, avg=550.36, stdev=143.07
     lat (usec): min=27, max=4303.6k, avg=503.87, stdev=36029 |	     lat (usec): min=46, max=5049, avg=553.68, stdev=143.03
     |  1.00th=[     60],  5.00th=[     68], 10.00th=[     75 |	     |  1.00th=[  215],  5.00th=[  343], 10.00th=[  404], 20.
     | 20.00th=[     89], 30.00th=[     97], 40.00th=[    101 |	     | 30.00th=[  490], 40.00th=[  510], 50.00th=[  529], 60.
     | 50.00th=[    109], 60.00th=[    125], 70.00th=[    139 |	     | 70.00th=[  586], 80.00th=[  660], 90.00th=[  758], 95.
     | 80.00th=[    153], 90.00th=[    182], 95.00th=[    215 |	     | 99.00th=[  938], 99.50th=[  996], 99.90th=[ 1106], 99.
     | 99.00th=[    297], 99.50th=[    326], 99.90th=[    392 |	     | 99.99th=[ 1221]
     | 99.95th=[    429], 99.99th=[2298479]		      <
Type=write,Size=4K,NumJobs=56,depth=64				Type=write,Size=4K,NumJobs=56,depth=64
  write: IOPS=699k, BW=2730MiB/s (2862MB/s)(33.6GiB/12594msec |	  write: IOPS=1625k, BW=6348MiB/s (6656MB/s)(62.0GiB/10003mse
    slat (nsec): min=1628, max=4304.3M, avg=40740.65, stdev=1 |	    slat (nsec): min=1656, max=1933.3k, avg=21620.58, stdev=7
    clat (usec): min=27, max=12592k, avg=4999.61, stdev=14711 |	    clat (usec): min=66, max=8663, avg=2182.95, stdev=879.90
     lat (usec): min=43, max=12592k, avg=5040.40, stdev=14795 |	     lat (usec): min=69, max=8669, avg=2204.64, stdev=895.51
     |  1.00th=[    178],  5.00th=[    180], 10.00th=[    182 |	     |  1.00th=[  693],  5.00th=[  971], 10.00th=[ 1074], 20.
     | 20.00th=[    184], 30.00th=[    229], 40.00th=[    245 |	     | 30.00th=[ 1942], 40.00th=[ 2114], 50.00th=[ 2212], 60.
     | 50.00th=[    247], 60.00th=[    251], 70.00th=[    255 |	     | 70.00th=[ 2343], 80.00th=[ 2507], 90.00th=[ 3392], 95.
     | 80.00th=[    260], 90.00th=[    273], 95.00th=[    334 |	     | 99.00th=[ 4883], 99.50th=[ 5211], 99.90th=[ 5800], 99.
     | 99.00th=[    383], 99.50th=[    388], 99.90th=[3003122 |	     | 99.99th=[ 6587]
     | 99.95th=[4328522], 99.99th=[5402264]		      <
Type=write,Size=64K,NumJobs=1,depth=1				Type=write,Size=64K,NumJobs=1,depth=1
  write: IOPS=23.3k, BW=1458MiB/s (1529MB/s)(14.2GiB/10001mse |	  write: IOPS=23.3k, BW=1454MiB/s (1525MB/s)(14.2GiB/10001mse
    slat (nsec): min=4814, max=41610, avg=6697.07, stdev=3203 |	    slat (nsec): min=4849, max=39702, avg=6880.02, stdev=3240
    clat (usec): min=3, max=130, avg=35.61, stdev= 9.49	      |	    clat (usec): min=17, max=133, avg=35.53, stdev= 9.39
     lat (usec): min=33, max=136, avg=42.35, stdev=10.60      |	     lat (usec): min=33, max=146, avg=42.46, stdev=10.52
     |  1.00th=[29312],  5.00th=[29312], 10.00th=[29312], 20. |	     |  1.00th=[29312],  5.00th=[29312], 10.00th=[29568], 20.
     | 30.00th=[29568], 40.00th=[29824], 50.00th=[30080], 60. |	     | 30.00th=[29568], 40.00th=[29824], 50.00th=[29824], 60.
     | 70.00th=[37120], 80.00th=[43776], 90.00th=[49920], 95. |	     | 70.00th=[36608], 80.00th=[43776], 90.00th=[49408], 95.
     | 99.00th=[69120], 99.50th=[72192], 99.90th=[86528], 99. |	     | 99.00th=[68096], 99.50th=[72192], 99.90th=[85504], 99.
     | 99.99th=[90624]						     | 99.99th=[90624]
Type=write,Size=64K,NumJobs=1,depth=16				Type=write,Size=64K,NumJobs=1,depth=16
  write: IOPS=140k, BW=8738MiB/s (9162MB/s)(85.3GiB/10001msec |	  write: IOPS=146k, BW=9109MiB/s (9552MB/s)(88.0GiB/10001msec
    slat (nsec): min=3279, max=79959, avg=5945.90, stdev=3397 |	    slat (nsec): min=3787, max=44724, avg=5539.44, stdev=622.
    clat (usec): min=63, max=272, avg=108.05, stdev=10.23     |	    clat (usec): min=76, max=228, avg=103.90, stdev=11.08
     lat (usec): min=67, max=276, avg=114.06, stdev=10.46     |	     lat (usec): min=82, max=233, avg=109.48, stdev=11.10
     |  1.00th=[   91],  5.00th=[   97], 10.00th=[   99], 20. |	     |  1.00th=[   90],  5.00th=[   93], 10.00th=[   95], 20.
     | 30.00th=[  102], 40.00th=[  103], 50.00th=[  105], 60. |	     | 30.00th=[   98], 40.00th=[   99], 50.00th=[  101], 60.
     | 70.00th=[  112], 80.00th=[  116], 90.00th=[  123], 95. |	     | 70.00th=[  105], 80.00th=[  109], 90.00th=[  120], 95.
     | 99.00th=[  139], 99.50th=[  145], 99.90th=[  169], 99. |	     | 99.00th=[  141], 99.50th=[  149], 99.90th=[  182], 99.
     | 99.99th=[  206]					      |	     | 99.99th=[  200]
Type=write,Size=64K,NumJobs=1,depth=64				Type=write,Size=64K,NumJobs=1,depth=64
  write: IOPS=145k, BW=9085MiB/s (9526MB/s)(88.7GiB/10001msec |	  write: IOPS=139k, BW=8708MiB/s (9131MB/s)(85.0GiB/10001msec
    slat (usec): min=2, max=164, avg= 5.45, stdev= 3.82	      |	    slat (nsec): min=2676, max=64177, avg=5388.79, stdev=1847
    clat (usec): min=235, max=952, avg=434.38, stdev=42.96    |	    clat (usec): min=39, max=894, avg=453.54, stdev=56.76
     lat (usec): min=239, max=957, avg=439.89, stdev=42.76    |	     lat (usec): min=45, max=898, avg=458.97, stdev=57.24
     |  1.00th=[  371],  5.00th=[  383], 10.00th=[  392], 20. |	     |  1.00th=[  338],  5.00th=[  355], 10.00th=[  367], 20.
     | 30.00th=[  408], 40.00th=[  416], 50.00th=[  424], 60. |	     | 30.00th=[  408], 40.00th=[  457], 50.00th=[  478], 60.
     | 70.00th=[  445], 80.00th=[  461], 90.00th=[  494], 95. |	     | 70.00th=[  494], 80.00th=[  498], 90.00th=[  506], 95.
     | 99.00th=[  570], 99.50th=[  586], 99.90th=[  652], 99. |	     | 99.00th=[  545], 99.50th=[  562], 99.90th=[  734], 99.
     | 99.99th=[  881]					      |	     | 99.99th=[  848]
Type=write,Size=64K,NumJobs=16,depth=1				Type=write,Size=64K,NumJobs=16,depth=1
  write: IOPS=170k, BW=10.4GiB/s (11.2GB/s)(104GiB/10001msec) |	  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10001msec)
    slat (nsec): min=3811, max=82769, avg=5460.88, stdev=1987 |	    slat (nsec): min=4383, max=73755, avg=5607.60, stdev=1278
    clat (usec): min=13, max=490, avg=87.97, stdev=11.39      |	    clat (usec): min=22, max=424, avg=85.24, stdev= 8.07
     lat (usec): min=37, max=494, avg=93.49, stdev=11.57      |	     lat (usec): min=38, max=430, avg=90.90, stdev= 8.07
     |  1.00th=[   57],  5.00th=[   72], 10.00th=[   77], 20. |	     |  1.00th=[   68],  5.00th=[   76], 10.00th=[   79], 20.
     | 30.00th=[   84], 40.00th=[   86], 50.00th=[   88], 60. |	     | 30.00th=[   82], 40.00th=[   83], 50.00th=[   84], 60.
     | 70.00th=[   92], 80.00th=[   96], 90.00th=[  102], 95. |	     | 70.00th=[   87], 80.00th=[   91], 90.00th=[   98], 95.
     | 99.00th=[  120], 99.50th=[  126], 99.90th=[  141], 99. |	     | 99.00th=[  106], 99.50th=[  112], 99.90th=[  124], 99.
     | 99.99th=[  190]					      |	     | 99.99th=[  137]
Type=write,Size=64K,NumJobs=16,depth=16				Type=write,Size=64K,NumJobs=16,depth=16
  write: IOPS=172k, BW=10.5GiB/s (11.3GB/s)(105GiB/10002msec) |	  write: IOPS=176k, BW=10.7GiB/s (11.5GB/s)(107GiB/10002msec)
    slat (nsec): min=3415, max=60314, avg=6662.15, stdev=1972 |	    slat (nsec): min=3611, max=70931, avg=6507.12, stdev=1389
    clat (usec): min=94, max=3871, avg=1480.76, stdev=360.52  |	    clat (usec): min=109, max=3915, avg=1451.13, stdev=409.41
     lat (usec): min=111, max=3880, avg=1487.48, stdev=360.80 |	     lat (usec): min=129, max=3920, avg=1457.68, stdev=409.50
     |  1.00th=[ 1270],  5.00th=[ 1287], 10.00th=[ 1287], 20. |	     |  1.00th=[ 1254],  5.00th=[ 1270], 10.00th=[ 1270], 20.
     | 30.00th=[ 1369], 40.00th=[ 1369], 50.00th=[ 1385], 60. |	     | 30.00th=[ 1270], 40.00th=[ 1270], 50.00th=[ 1287], 60.
     | 70.00th=[ 1418], 80.00th=[ 1467], 90.00th=[ 1500], 95. |	     | 70.00th=[ 1352], 80.00th=[ 1369], 90.00th=[ 2540], 95.
     | 99.00th=[ 2802], 99.50th=[ 2802], 99.90th=[ 2835], 99. |	     | 99.00th=[ 2737], 99.50th=[ 2737], 99.90th=[ 2769], 99.
     | 99.99th=[ 2900]					      |	     | 99.99th=[ 2769]
Type=write,Size=64K,NumJobs=16,depth=64				Type=write,Size=64K,NumJobs=16,depth=64
  write: IOPS=172k, BW=10.5GiB/s (11.3GB/s)(105GiB/10005msec) |	  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10004msec)
    slat (usec): min=2, max=5927, avg=15.49, stdev=123.06     |	    slat (nsec): min=3042, max=96073, avg=6866.24, stdev=2170
    clat (usec): min=114, max=16669, avg=5934.82, stdev=1800. |	    clat (usec): min=124, max=10103, avg=5850.40, stdev=798.1
     lat (usec): min=122, max=17554, avg=5950.36, stdev=1825. |	     lat (usec): min=131, max=10109, avg=5857.31, stdev=798.1
     |  1.00th=[ 4293],  5.00th=[ 4424], 10.00th=[ 4555], 20. |	     |  1.00th=[ 4752],  5.00th=[ 4883], 10.00th=[ 4948], 20.
     | 30.00th=[ 4948], 40.00th=[ 5145], 50.00th=[ 5342], 60. |	     | 30.00th=[ 5276], 40.00th=[ 5407], 50.00th=[ 5669], 60.
     | 70.00th=[ 5932], 80.00th=[ 6390], 90.00th=[ 9110], 95. |	     | 70.00th=[ 6259], 80.00th=[ 6587], 90.00th=[ 7046], 95.
     | 99.00th=[11994], 99.50th=[12518], 99.90th=[13566], 99. |	     | 99.00th=[ 7767], 99.50th=[ 7898], 99.90th=[ 8455], 99.
     | 99.99th=[15401]					      |	     | 99.99th=[ 8848]
Type=write,Size=64K,NumJobs=56,depth=1				Type=write,Size=64K,NumJobs=56,depth=1
  write: IOPS=171k, BW=10.5GiB/s (11.2GB/s)(105GiB/10002msec) |	  write: IOPS=175k, BW=10.7GiB/s (11.4GB/s)(107GiB/10002msec)
    slat (nsec): min=3796, max=64949, avg=5590.94, stdev=2033 |	    slat (nsec): min=4029, max=63780, avg=5771.34, stdev=1537
    clat (usec): min=47, max=940, avg=320.48, stdev=68.32     |	    clat (usec): min=32, max=979, avg=314.28, stdev=57.25
     lat (usec): min=53, max=947, avg=326.12, stdev=68.40     |	     lat (usec): min=40, max=985, avg=320.11, stdev=57.29
     |  1.00th=[  190],  5.00th=[  202], 10.00th=[  302], 20. |	     |  1.00th=[  186],  5.00th=[  293], 10.00th=[  297], 20.
     | 30.00th=[  306], 40.00th=[  310], 50.00th=[  314], 60. |	     | 30.00th=[  297], 40.00th=[  302], 50.00th=[  302], 60.
     | 70.00th=[  322], 80.00th=[  326], 90.00th=[  334], 95. |	     | 70.00th=[  310], 80.00th=[  310], 90.00th=[  326], 95.
     | 99.00th=[  635], 99.50th=[  644], 99.90th=[  660], 99. |	     | 99.00th=[  603], 99.50th=[  611], 99.90th=[  627], 99.
     | 99.99th=[  783]					      |	     | 99.99th=[  750]
Type=write,Size=64K,NumJobs=56,depth=16				Type=write,Size=64K,NumJobs=56,depth=16
  write: IOPS=172k, BW=10.5GiB/s (11.3GB/s)(105GiB/10005msec) |	  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10005msec)
    slat (nsec): min=3155, max=79675, avg=7689.97, stdev=2581 |	    slat (nsec): min=3542, max=89310, avg=7585.26, stdev=2460
    clat (usec): min=151, max=13705, avg=5207.44, stdev=985.7 |	    clat (usec): min=190, max=14367, avg=5116.17, stdev=1396.
     lat (usec): min=175, max=13712, avg=5215.19, stdev=985.8 |	     lat (usec): min=198, max=14375, avg=5123.82, stdev=1396.
     |  1.00th=[ 2573],  5.00th=[ 2606], 10.00th=[ 5014], 20. |	     |  1.00th=[ 2540],  5.00th=[ 2540], 10.00th=[ 2573], 20.
     | 30.00th=[ 5145], 40.00th=[ 5211], 50.00th=[ 5211], 60. |	     | 30.00th=[ 5080], 40.00th=[ 5080], 50.00th=[ 5145], 60.
     | 70.00th=[ 5211], 80.00th=[ 5276], 90.00th=[ 5866], 95. |	     | 70.00th=[ 5145], 80.00th=[ 5669], 90.00th=[ 6849], 95.
     | 99.00th=[ 8455], 99.50th=[ 8848], 99.90th=[10028], 99. |	     | 99.00th=[ 8979], 99.50th=[ 9634], 99.90th=[11207], 99.
     | 99.99th=[11600]					      |	     | 99.99th=[12780]
Type=write,Size=64K,NumJobs=56,depth=64				Type=write,Size=64K,NumJobs=56,depth=64
  write: IOPS=172k, BW=10.5GiB/s (11.3GB/s)(105GiB/10014msec) |	  write: IOPS=174k, BW=10.6GiB/s (11.4GB/s)(107GiB/10011msec)
    slat (usec): min=2, max=45005, avg=184.96, stdev=907.62   |	    slat (usec): min=3, max=32839, avg=176.71, stdev=925.53
    clat (usec): min=195, max=128349, avg=20590.58, stdev=799 |	    clat (usec): min=189, max=132076, avg=20366.61, stdev=728
     lat (usec): min=229, max=128358, avg=20775.62, stdev=814 |	     lat (usec): min=218, max=132086, avg=20543.39, stdev=740
     |  1.00th=[ 8455],  5.00th=[ 9503], 10.00th=[10552], 20. |	     |  1.00th=[ 7898],  5.00th=[ 9110], 10.00th=[10945], 20.
     | 30.00th=[19006], 40.00th=[20055], 50.00th=[20317], 60. |	     | 30.00th=[17957], 40.00th=[19530], 50.00th=[20317], 60.
     | 70.00th=[20841], 80.00th=[22152], 90.00th=[26608], 95. |	     | 70.00th=[21103], 80.00th=[23462], 90.00th=[28705], 95.
     | 99.00th=[52691], 99.50th=[58459], 99.90th=[70779], 99. |	     | 99.00th=[45876], 99.50th=[50594], 99.90th=[63177], 99.
     | 99.99th=[94897]					      |	     | 99.99th=[84411]
Target with DIM							Without DIM
Type=read,Size=512B,NumJobs=1,depth=1				Type=read,Size=512B,NumJobs=1,depth=1
   read: IOPS=57.6k, BW=28.1MiB/s (29.5MB/s)(281MiB/10001msec |	   read: IOPS=47.0k, BW=22.0MiB/s (24.1MB/s)(230MiB/10001msec
    slat (nsec): min=2050, max=44230, avg=2445.95, stdev=1144 |	    slat (nsec): min=1981, max=53196, avg=3048.29, stdev=2391
    clat (nsec): min=838, max=98356, avg=14486.18, stdev=4396 |	    clat (nsec): min=442, max=107156, avg=17658.37, stdev=647
     lat (usec): min=12, max=116, avg=16.97, stdev= 4.73      |	     lat (usec): min=13, max=109, avg=20.78, stdev= 7.18
     |  1.00th=[12864],  5.00th=[12864], 10.00th=[12992], 20. |	     |  1.00th=[13632],  5.00th=[13760], 10.00th=[13888], 20.
     | 30.00th=[13120], 40.00th=[13120], 50.00th=[13120], 60. |	     | 30.00th=[14144], 40.00th=[14144], 50.00th=[14400], 60.
     | 70.00th=[13376], 80.00th=[13504], 90.00th=[15552], 95. |	     | 70.00th=[17024], 80.00th=[19072], 90.00th=[27776], 95.
     | 99.00th=[35072], 99.50th=[36096], 99.90th=[40704], 99. |	     | 99.00th=[40704], 99.50th=[45824], 99.90th=[51456], 99.
     | 99.99th=[51968]					      |	     | 99.99th=[58624]
Type=read,Size=512B,NumJobs=1,depth=16				Type=read,Size=512B,NumJobs=1,depth=16
   read: IOPS=348k, BW=170MiB/s (178MB/s)(1697MiB/10001msec)  |	   read: IOPS=334k, BW=163MiB/s (171MB/s)(1629MiB/10001msec)
    slat (nsec): min=1594, max=45285, avg=1933.31, stdev=410. |	    slat (nsec): min=1564, max=28780, avg=2162.48, stdev=588.
    clat (usec): min=36, max=162, avg=43.80, stdev= 8.09      |	    clat (usec): min=17, max=118, avg=45.50, stdev= 6.30
     lat (usec): min=37, max=164, avg=45.77, stdev= 8.11      |	     lat (usec): min=20, max=120, avg=47.70, stdev= 6.64
     |  1.00th=[   40],  5.00th=[   40], 10.00th=[   40], 20. |	     |  1.00th=[41728],  5.00th=[42752], 10.00th=[42752], 20.
     | 30.00th=[   41], 40.00th=[   41], 50.00th=[   42], 60. |	     | 30.00th=[43264], 40.00th=[43776], 50.00th=[43776], 60.
     | 70.00th=[   43], 80.00th=[   44], 90.00th=[   52], 95. |	     | 70.00th=[44288], 80.00th=[44288], 90.00th=[46336], 95.
     | 99.00th=[   81], 99.50th=[   90], 99.90th=[  112], 99. |	     | 99.00th=[72192], 99.50th=[73216], 99.90th=[77312], 99.
     | 99.99th=[  139]					      |	     | 99.99th=[86528]
Type=read,Size=512B,NumJobs=1,depth=64				Type=read,Size=512B,NumJobs=1,depth=64
   read: IOPS=373k, BW=182MiB/s (191MB/s)(1822MiB/10001msec)  |	   read: IOPS=344k, BW=168MiB/s (176MB/s)(1682MiB/10001msec)
    slat (nsec): min=1581, max=43893, avg=1907.45, stdev=238. |	    slat (nsec): min=1607, max=30650, avg=2093.56, stdev=233.
    clat (usec): min=58, max=295, avg=169.31, stdev= 2.74     |	    clat (usec): min=19, max=238, avg=183.42, stdev= 2.22
     lat (usec): min=60, max=297, avg=171.25, stdev= 2.76     |	     lat (usec): min=21, max=241, avg=185.55, stdev= 2.23
     |  1.00th=[  167],  5.00th=[  167], 10.00th=[  167], 20. |	     |  1.00th=[  180],  5.00th=[  182], 10.00th=[  182], 20.
     | 30.00th=[  169], 40.00th=[  169], 50.00th=[  169], 60. |	     | 30.00th=[  182], 40.00th=[  184], 50.00th=[  184], 60.
     | 70.00th=[  169], 80.00th=[  172], 90.00th=[  172], 95. |	     | 70.00th=[  184], 80.00th=[  186], 90.00th=[  186], 95.
     | 99.00th=[  176], 99.50th=[  176], 99.90th=[  210], 99. |	     | 99.00th=[  190], 99.50th=[  192], 99.90th=[  196], 99.
     | 99.99th=[  231]					      |	     | 99.99th=[  210]
Type=read,Size=512B,NumJobs=16,depth=1				Type=read,Size=512B,NumJobs=16,depth=1
   read: IOPS=924k, BW=451MiB/s (473MB/s)(4510MiB/10001msec)  |	   read: IOPS=907k, BW=443MiB/s (465MB/s)(4431MiB/10001msec)
    slat (nsec): min=1865, max=66933, avg=2473.89, stdev=487. |	    slat (nsec): min=1925, max=69082, avg=2505.79, stdev=869.
    clat (nsec): min=598, max=426580, avg=14400.84, stdev=125 |	    clat (nsec): min=476, max=112521, avg=14672.56, stdev=182
     lat (usec): min=11, max=428, avg=16.92, stdev= 1.37      |	     lat (usec): min=12, max=131, avg=17.23, stdev= 2.08
     |  1.00th=[11840],  5.00th=[13248], 10.00th=[13504], 20. |	     |  1.00th=[11584],  5.00th=[13248], 10.00th=[13376], 20.
     | 30.00th=[13888], 40.00th=[14016], 50.00th=[14144], 60. |	     | 30.00th=[13888], 40.00th=[14016], 50.00th=[14272], 60.
     | 70.00th=[14528], 80.00th=[14912], 90.00th=[15552], 95. |	     | 70.00th=[14784], 80.00th=[15296], 90.00th=[16512], 95.
     | 99.00th=[18816], 99.50th=[20864], 99.90th=[24960], 99. |	     | 99.00th=[22400], 99.50th=[24448], 99.90th=[32384], 99.
     | 99.99th=[32128]					      |	     | 99.99th=[35584]
Type=read,Size=512B,NumJobs=16,depth=16				Type=read,Size=512B,NumJobs=16,depth=16
   read: IOPS=4616k, BW=2254MiB/s (2363MB/s)(22.0GiB/10001mse |	   read: IOPS=4055k, BW=1980MiB/s (2076MB/s)(19.3GiB/10003mse
    slat (nsec): min=1571, max=101196, avg=2275.70, stdev=782 |	    slat (nsec): min=1572, max=239521, avg=2836.25, stdev=282
    clat (usec): min=24, max=492, avg=52.83, stdev= 9.68      |	    clat (usec): min=9, max=616, avg=59.85, stdev=55.40
     lat (usec): min=26, max=523, avg=55.15, stdev= 9.91      |	     lat (usec): min=15, max=630, avg=62.74, stdev=57.98
     |  1.00th=[   42],  5.00th=[   43], 10.00th=[   43], 20. |	     |  1.00th=[   43],  5.00th=[   44], 10.00th=[   44], 20.
     | 30.00th=[   47], 40.00th=[   50], 50.00th=[   52], 60. |	     | 30.00th=[   45], 40.00th=[   46], 50.00th=[   46], 60.
     | 70.00th=[   57], 80.00th=[   60], 90.00th=[   65], 95. |	     | 70.00th=[   48], 80.00th=[   52], 90.00th=[   60], 95.
     | 99.00th=[   79], 99.50th=[   82], 99.90th=[   88], 99. |	     | 99.00th=[  326], 99.50th=[  343], 99.90th=[  367], 99.
     | 99.99th=[  306]					      |	     | 99.99th=[  519]
Type=read,Size=512B,NumJobs=16,depth=64				Type=read,Size=512B,NumJobs=16,depth=64
   read: IOPS=5208k, BW=2543MiB/s (2666MB/s)(24.8GiB/10002mse |	   read: IOPS=3451k, BW=1685MiB/s (1767MB/s)(16.5GiB/10001mse
    slat (nsec): min=1575, max=61538, avg=2219.34, stdev=1268 |	    slat (nsec): min=1569, max=204726, avg=3322.74, stdev=274
    clat (usec): min=42, max=378, avg=194.04, stdev=28.59     |	    clat (usec): min=5, max=1543, avg=292.82, stdev=220.59
     lat (usec): min=44, max=380, avg=196.30, stdev=28.94     |	     lat (usec): min=14, max=1561, avg=296.20, stdev=223.03
     |  1.00th=[  172],  5.00th=[  174], 10.00th=[  174], 20. |	     |  1.00th=[  176],  5.00th=[  178], 10.00th=[  180], 20.
     | 30.00th=[  176], 40.00th=[  178], 50.00th=[  178], 60. |	     | 30.00th=[  188], 40.00th=[  227], 50.00th=[  243], 60.
     | 70.00th=[  208], 80.00th=[  212], 90.00th=[  233], 95. |	     | 70.00th=[  249], 80.00th=[  285], 90.00th=[  685], 95.
     | 99.00th=[  306], 99.50th=[  318], 99.90th=[  330], 99. |	     | 99.00th=[ 1270], 99.50th=[ 1319], 99.90th=[ 1385], 99.
     | 99.99th=[  343]					      |	     | 99.99th=[ 1450]
Type=read,Size=512B,NumJobs=56,depth=1				Type=read,Size=512B,NumJobs=56,depth=1
   read: IOPS=2294k, BW=1120MiB/s (1175MB/s)(10.9GiB/10003mse |	   read: IOPS=2351k, BW=1148MiB/s (1204MB/s)(11.2GiB/10002mse
    slat (nsec): min=1826, max=842064, avg=3629.96, stdev=274 |	    slat (nsec): min=1744, max=502891, avg=3525.21, stdev=199
    clat (nsec): min=468, max=3959.9k, avg=20048.74, stdev=10 |	    clat (nsec): min=456, max=781278, avg=19563.02, stdev=418
     lat (usec): min=10, max=3962, avg=23.76, stdev=11.12     |	     lat (usec): min=11, max=785, avg=23.18, stdev= 4.79
     |  1.00th=[   14],  5.00th=[   15], 10.00th=[   16], 20. |	     |  1.00th=[13120],  5.00th=[14912], 10.00th=[15936], 20.
     | 30.00th=[   18], 40.00th=[   19], 50.00th=[   20], 60. |	     | 30.00th=[17792], 40.00th=[18304], 50.00th=[19072], 60.
     | 70.00th=[   21], 80.00th=[   23], 90.00th=[   25], 95. |	     | 70.00th=[20608], 80.00th=[21632], 90.00th=[23680], 95.
     | 99.00th=[   37], 99.50th=[   41], 99.90th=[   61], 99. |	     | 99.00th=[31872], 99.50th=[35072], 99.90th=[41728], 99.
     | 99.99th=[  529]					      |	     | 99.99th=[56064]
Type=read,Size=512B,NumJobs=56,depth=16				Type=read,Size=512B,NumJobs=56,depth=16
   read: IOPS=8192k, BW=4000MiB/s (4194MB/s)(39.1GiB/10001mse |	   read: IOPS=7234k, BW=3532MiB/s (3704MB/s)(34.5GiB/10002mse
    slat (nsec): min=1639, max=7118.6k, avg=5019.37, stdev=74 |	    slat (nsec): min=1588, max=13073k, avg=5296.03, stdev=602
    clat (usec): min=5, max=7261, avg=103.57, stdev=52.37     |	    clat (usec): min=9, max=13311, avg=117.66, stdev=77.29
     lat (usec): min=34, max=7264, avg=108.69, stdev=54.81    |	     lat (usec): min=18, max=13322, avg=123.07, stdev=80.61
     |  1.00th=[   64],  5.00th=[   68], 10.00th=[   69], 20. |	     |  1.00th=[   63],  5.00th=[   67], 10.00th=[   69], 20.
     | 30.00th=[   73], 40.00th=[   74], 50.00th=[   77], 60. |	     | 30.00th=[   72], 40.00th=[   76], 50.00th=[   81], 60.
     | 70.00th=[   94], 80.00th=[  157], 90.00th=[  182], 95. |	     | 70.00th=[  108], 80.00th=[  184], 90.00th=[  247], 95.
     | 99.00th=[  247], 99.50th=[  306], 99.90th=[  412], 99. |	     | 99.00th=[  338], 99.50th=[  383], 99.90th=[  494], 99.
     | 99.99th=[  947]					      |	     | 99.99th=[  594]
Type=read,Size=512B,NumJobs=56,depth=64				Type=read,Size=512B,NumJobs=56,depth=64
   read: IOPS=8312k, BW=4059MiB/s (4256MB/s)(39.6GiB/10002mse |	   read: IOPS=7769k, BW=3793MiB/s (3978MB/s)(37.1GiB/10002mse
    slat (nsec): min=1629, max=669215, avg=5032.55, stdev=769 |	    slat (nsec): min=1583, max=3785.4k, avg=5319.20, stdev=81
    clat (usec): min=43, max=1951, avg=425.25, stdev=208.07   |	    clat (usec): min=2, max=5212, avg=455.02, stdev=334.71
     lat (usec): min=63, max=1968, avg=430.39, stdev=210.56   |	     lat (usec): min=17, max=5214, avg=460.45, stdev=338.64
     |  1.00th=[  273],  5.00th=[  281], 10.00th=[  289], 20. |	     |  1.00th=[  265],  5.00th=[  273], 10.00th=[  281], 20.
     | 30.00th=[  302], 40.00th=[  302], 50.00th=[  306], 60. |	     | 30.00th=[  293], 40.00th=[  297], 50.00th=[  302], 60.
     | 70.00th=[  322], 80.00th=[  709], 90.00th=[  750], 95. |	     | 70.00th=[  310], 80.00th=[  437], 90.00th=[ 1074], 95.
     | 99.00th=[  971], 99.50th=[ 1270], 99.90th=[ 1500], 99. |	     | 99.00th=[ 1401], 99.50th=[ 1991], 99.90th=[ 2245], 99.
     | 99.99th=[ 1663]					      |	     | 99.99th=[ 2442]
Type=read,Size=4K,NumJobs=1,depth=1				Type=read,Size=4K,NumJobs=1,depth=1
   read: IOPS=52.3k, BW=204MiB/s (214MB/s)(2042MiB/10001msec) |	   read: IOPS=54.4k, BW=213MiB/s (223MB/s)(2126MiB/10001msec)
    slat (nsec): min=2073, max=41533, avg=2731.01, stdev=1940 |	    slat (nsec): min=2044, max=28781, avg=2556.03, stdev=1441
    clat (nsec): min=647, max=121164, avg=15919.33, stdev=561 |	    clat (nsec): min=621, max=70566, avg=15359.54, stdev=5250
     lat (usec): min=13, max=123, avg=18.70, stdev= 6.22      |	     lat (nsec): min=13302, max=73823, avg=17966.25, stdev=56
     |  1.00th=[13376],  5.00th=[13504], 10.00th=[13504], 20. |	     |  1.00th=[13120],  5.00th=[13248], 10.00th=[13248], 20.
     | 30.00th=[13632], 40.00th=[13760], 50.00th=[13888], 60. |	     | 30.00th=[13376], 40.00th=[13376], 50.00th=[13376], 60.
     | 70.00th=[14144], 80.00th=[15040], 90.00th=[20096], 95. |	     | 70.00th=[13632], 80.00th=[14144], 90.00th=[19328], 95.
     | 99.00th=[36608], 99.50th=[40192], 99.90th=[50944], 99. |	     | 99.00th=[35584], 99.50th=[36608], 99.90th=[45312], 99.
     | 99.99th=[73216]					      |	     | 99.99th=[53504]
Type=read,Size=4K,NumJobs=1,depth=16				Type=read,Size=4K,NumJobs=1,depth=16
   read: IOPS=321k, BW=1255MiB/s (1316MB/s)(12.3GiB/10001msec |	   read: IOPS=334k, BW=1306MiB/s (1370MB/s)(12.8GiB/10001msec
    slat (nsec): min=1652, max=42407, avg=2045.49, stdev=500. |	    slat (nsec): min=1696, max=45538, avg=2168.76, stdev=208.
    clat (usec): min=34, max=181, avg=47.45, stdev=10.18      |	    clat (usec): min=18, max=148, avg=45.39, stdev= 1.47
     lat (usec): min=36, max=183, avg=49.54, stdev=10.20      |	     lat (usec): min=20, max=192, avg=47.59, stdev= 1.48
     |  1.00th=[   42],  5.00th=[   42], 10.00th=[   42], 20. |	     |  1.00th=[43776],  5.00th=[44288], 10.00th=[44288], 20.
     | 30.00th=[   43], 40.00th=[   43], 50.00th=[   44], 60. |	     | 30.00th=[44800], 40.00th=[45312], 50.00th=[45312], 60.
     | 70.00th=[   45], 80.00th=[   53], 90.00th=[   60], 95. |	     | 70.00th=[45312], 80.00th=[45824], 90.00th=[45824], 95.
     | 99.00th=[   92], 99.50th=[   98], 99.90th=[  122], 99. |	     | 99.00th=[49408], 99.50th=[54528], 99.90th=[64256], 99.
     | 99.99th=[  147]					      |	     | 99.99th=[70144]
Type=read,Size=4K,NumJobs=1,depth=64				Type=read,Size=4K,NumJobs=1,depth=64
   read: IOPS=358k, BW=1398MiB/s (1466MB/s)(13.7GiB/10001msec |	   read: IOPS=338k, BW=1319MiB/s (1383MB/s)(12.9GiB/10001msec
    slat (nsec): min=1586, max=39977, avg=2013.84, stdev=228. |	    slat (nsec): min=1621, max=52632, avg=2145.29, stdev=209.
    clat (usec): min=33, max=277, avg=176.53, stdev= 3.73     |	    clat (usec): min=36, max=236, avg=187.02, stdev= 2.05
     lat (usec): min=34, max=279, avg=178.58, stdev= 3.75     |	     lat (usec): min=38, max=238, avg=189.20, stdev= 2.06
     |  1.00th=[  174],  5.00th=[  174], 10.00th=[  174], 20. |	     |  1.00th=[  184],  5.00th=[  184], 10.00th=[  186], 20.
     | 30.00th=[  176], 40.00th=[  176], 50.00th=[  176], 60. |	     | 30.00th=[  186], 40.00th=[  188], 50.00th=[  188], 60.
     | 70.00th=[  178], 80.00th=[  178], 90.00th=[  180], 95. |	     | 70.00th=[  188], 80.00th=[  188], 90.00th=[  190], 95.
     | 99.00th=[  192], 99.50th=[  204], 99.90th=[  217], 99. |	     | 99.00th=[  192], 99.50th=[  196], 99.90th=[  200], 99.
     | 99.99th=[  260]					      |	     | 99.99th=[  217]
Type=read,Size=4K,NumJobs=16,depth=1				Type=read,Size=4K,NumJobs=16,depth=1
   read: IOPS=876k, BW=3422MiB/s (3588MB/s)(33.4GiB/10001msec |	   read: IOPS=869k, BW=3396MiB/s (3561MB/s)(33.2GiB/10002msec
    slat (nsec): min=2018, max=54502, avg=2490.14, stdev=570. |	    slat (nsec): min=1890, max=70947, avg=2497.58, stdev=813.
    clat (nsec): min=586, max=146434, avg=15319.20, stdev=175 |	    clat (nsec): min=590, max=79209, avg=15456.14, stdev=1968
     lat (usec): min=11, max=148, avg=17.86, stdev= 1.86      |	     lat (usec): min=12, max=110, avg=18.01, stdev= 2.18
     |  1.00th=[12864],  5.00th=[13888], 10.00th=[14144], 20. |	     |  1.00th=[12352],  5.00th=[13888], 10.00th=[14144], 20.
     | 30.00th=[14656], 40.00th=[14784], 50.00th=[14912], 60. |	     | 30.00th=[14528], 40.00th=[14784], 50.00th=[15040], 60.
     | 70.00th=[15424], 80.00th=[15936], 90.00th=[16768], 95. |	     | 70.00th=[15552], 80.00th=[16192], 90.00th=[17280], 95.
     | 99.00th=[22400], 99.50th=[24192], 99.90th=[32640], 99. |	     | 99.00th=[23936], 99.50th=[25728], 99.90th=[33536], 99.
     | 99.99th=[72192]					      |	     | 99.99th=[36096]
Type=read,Size=4K,NumJobs=16,depth=16				Type=read,Size=4K,NumJobs=16,depth=16
   read: IOPS=2665k, BW=10.2GiB/s (10.9GB/s)(102GiB/10001msec |	   read: IOPS=1843k, BW=7201MiB/s (7551MB/s)(70.3GiB/10002mse
    slat (nsec): min=1576, max=160409, avg=2277.16, stdev=105 |	    slat (nsec): min=1582, max=104435, avg=3170.92, stdev=170
    clat (usec): min=21, max=895, avg=93.43, stdev=13.26      |	    clat (usec): min=40, max=554, avg=135.18, stdev=32.57
     lat (usec): min=25, max=897, avg=95.75, stdev=13.22      |	     lat (usec): min=42, max=557, avg=138.42, stdev=32.59
     |  1.00th=[   74],  5.00th=[   78], 10.00th=[   80], 20. |	     |  1.00th=[   68],  5.00th=[   92], 10.00th=[  104], 20.
     | 30.00th=[   88], 40.00th=[   93], 50.00th=[   96], 60. |	     | 30.00th=[  125], 40.00th=[  130], 50.00th=[  135], 60.
     | 70.00th=[   99], 80.00th=[  100], 90.00th=[  102], 95. |	     | 70.00th=[  141], 80.00th=[  147], 90.00th=[  161], 95.
     | 99.00th=[  167], 99.50th=[  174], 99.90th=[  186], 99. |	     | 99.00th=[  262], 99.50th=[  306], 99.90th=[  392], 99.
     | 99.99th=[  258]					      |	     | 99.99th=[  461]
Type=read,Size=4K,NumJobs=16,depth=64				Type=read,Size=4K,NumJobs=16,depth=64
   read: IOPS=2670k, BW=10.2GiB/s (10.9GB/s)(102GiB/10002msec |	   read: IOPS=1917k, BW=7487MiB/s (7850MB/s)(73.1GiB/10001mse
    slat (nsec): min=1559, max=332598, avg=2408.47, stdev=359 |	    slat (nsec): min=1566, max=653412, avg=3347.14, stdev=385
    clat (usec): min=25, max=1243, avg=380.63, stdev=95.54    |	    clat (usec): min=43, max=1445, avg=530.39, stdev=127.84
     lat (usec): min=30, max=1245, avg=383.08, stdev=96.26    |	     lat (usec): min=45, max=1619, avg=533.81, stdev=128.49
     |  1.00th=[  182],  5.00th=[  247], 10.00th=[  302], 20. |	     |  1.00th=[  188],  5.00th=[  351], 10.00th=[  420], 20.
     | 30.00th=[  347], 40.00th=[  359], 50.00th=[  367], 60. |	     | 30.00th=[  486], 40.00th=[  502], 50.00th=[  519], 60.
     | 70.00th=[  383], 80.00th=[  408], 90.00th=[  465], 95. |	     | 70.00th=[  553], 80.00th=[  578], 90.00th=[  644], 95.
     | 99.00th=[  742], 99.50th=[  766], 99.90th=[  816], 99. |	     | 99.00th=[ 1012], 99.50th=[ 1037], 99.90th=[ 1090], 99.
     | 99.99th=[  873]					      |	     | 99.99th=[ 1205]
Type=read,Size=4K,NumJobs=56,depth=1				Type=read,Size=4K,NumJobs=56,depth=1
   read: IOPS=1594k, BW=6227MiB/s (6530MB/s)(60.8GiB/10002mse |	   read: IOPS=1571k, BW=6138MiB/s (6436MB/s)(59.9GiB/10001mse
    slat (nsec): min=1747, max=711701, avg=2984.71, stdev=222 |	    slat (nsec): min=1693, max=505053, avg=2771.65, stdev=110
    clat (nsec): min=468, max=4376.0k, avg=31538.77, stdev=12 |	    clat (nsec): min=479, max=1348.1k, avg=32311.24, stdev=36
     lat (usec): min=13, max=4378, avg=34.59, stdev=13.16     |	     lat (usec): min=15, max=1350, avg=35.15, stdev= 3.81
     |  1.00th=[   22],  5.00th=[   26], 10.00th=[   28], 20. |	     |  1.00th=[26752],  5.00th=[28800], 10.00th=[29568], 20.
     | 30.00th=[   30], 40.00th=[   31], 50.00th=[   31], 60. |	     | 30.00th=[30848], 40.00th=[31360], 50.00th=[31872], 60.
     | 70.00th=[   33], 80.00th=[   34], 90.00th=[   36], 95. |	     | 70.00th=[33024], 80.00th=[33536], 90.00th=[35584], 95.
     | 99.00th=[   45], 99.50th=[   50], 99.90th=[  243], 99. |	     | 99.00th=[42240], 99.50th=[43776], 99.90th=[49920], 99.
     | 99.99th=[  519]					      |	     | 99.99th=[59136]
Type=read,Size=4K,NumJobs=56,depth=16				Type=read,Size=4K,NumJobs=56,depth=16
   read: IOPS=2652k, BW=10.1GiB/s (10.9GB/s)(101GiB/10002msec |	   read: IOPS=2009k, BW=7846MiB/s (8227MB/s)(76.6GiB/10001mse
    slat (nsec): min=1571, max=937333, avg=3004.91, stdev=194 |	    slat (nsec): min=1566, max=379585, avg=3204.59, stdev=230
    clat (usec): min=26, max=1529, avg=334.21, stdev=57.70    |	    clat (usec): min=44, max=1761, avg=442.22, stdev=244.27
     lat (usec): min=28, max=1531, avg=337.28, stdev=57.71    |	     lat (usec): min=46, max=1764, avg=445.50, stdev=244.24
     |  1.00th=[  147],  5.00th=[  210], 10.00th=[  297], 20. |	     |  1.00th=[  161],  5.00th=[  194], 10.00th=[  215], 20.
     | 30.00th=[  322], 40.00th=[  330], 50.00th=[  334], 60. |	     | 30.00th=[  273], 40.00th=[  306], 50.00th=[  351], 60.
     | 70.00th=[  351], 80.00th=[  359], 90.00th=[  388], 95. |	     | 70.00th=[  506], 80.00th=[  644], 90.00th=[  832], 95.
     | 99.00th=[  490], 99.50th=[  529], 99.90th=[  709], 99. |	     | 99.00th=[ 1156], 99.50th=[ 1221], 99.90th=[ 1336], 99.
     | 99.99th=[ 1004]					      |	     | 99.99th=[ 1450]
Type=read,Size=4K,NumJobs=56,depth=64				Type=read,Size=4K,NumJobs=56,depth=64
   read: IOPS=2659k, BW=10.1GiB/s (10.9GB/s)(101GiB/10003msec |	   read: IOPS=1801k, BW=7036MiB/s (7378MB/s)(68.7GiB/10002mse
    slat (nsec): min=1576, max=1678.0k, avg=14535.12, stdev=5 |	    slat (nsec): min=1586, max=2410.2k, avg=20136.27, stdev=8
    clat (usec): min=29, max=7242, avg=1332.35, stdev=730.40  |	    clat (usec): min=58, max=8215, avg=1968.75, stdev=947.22
     lat (usec): min=32, max=7244, avg=1346.94, stdev=743.63  |	     lat (usec): min=60, max=8632, avg=1988.96, stdev=962.84
     |  1.00th=[  515],  5.00th=[  619], 10.00th=[  644], 20. |	     |  1.00th=[  449],  5.00th=[  734], 10.00th=[  865], 20.
     | 30.00th=[  693], 40.00th=[  807], 50.00th=[ 1303], 60. |	     | 30.00th=[ 1418], 40.00th=[ 1729], 50.00th=[ 1909], 60.
     | 70.00th=[ 1418], 80.00th=[ 1975], 90.00th=[ 2442], 95. |	     | 70.00th=[ 2212], 80.00th=[ 2507], 90.00th=[ 3326], 95.
     | 99.00th=[ 3392], 99.50th=[ 3621], 99.90th=[ 4047], 99. |	     | 99.00th=[ 4817], 99.50th=[ 5080], 99.90th=[ 5735], 99.
     | 99.99th=[ 4752]					      |	     | 99.99th=[ 6587]
Type=read,Size=64K,NumJobs=1,depth=1				Type=read,Size=64K,NumJobs=1,depth=1
   read: IOPS=29.7k, BW=1854MiB/s (1944MB/s)(18.1GiB/10001mse |	   read: IOPS=29.8k, BW=1864MiB/s (1955MB/s)(18.2GiB/10001mse
    slat (nsec): min=4662, max=82948, avg=5935.55, stdev=2735 |	    slat (nsec): min=4768, max=95484, avg=6119.74, stdev=2779
    clat (usec): min=3, max=136, avg=27.27, stdev= 7.95	      |	    clat (nsec): min=1555, max=75556, avg=26881.73, stdev=781
     lat (usec): min=25, max=141, avg=33.25, stdev= 9.18      |	     lat (usec): min=25, max=102, avg=33.05, stdev= 9.07
     |  1.00th=[21888],  5.00th=[22144], 10.00th=[22144], 20. |	     |  1.00th=[21632],  5.00th=[21632], 10.00th=[21632], 20.
     | 30.00th=[22400], 40.00th=[22400], 50.00th=[22656], 60. |	     | 30.00th=[21888], 40.00th=[22144], 50.00th=[22144], 60.
     | 70.00th=[28288], 80.00th=[34560], 90.00th=[41216], 95. |	     | 70.00th=[28032], 80.00th=[34048], 90.00th=[40704], 95.
     | 99.00th=[48384], 99.50th=[50944], 99.90th=[66048], 99. |	     | 99.00th=[47360], 99.50th=[49920], 99.90th=[59648], 99.
     | 99.99th=[78336]					      |	     | 99.99th=[70144]
Type=read,Size=64K,NumJobs=1,depth=16				Type=read,Size=64K,NumJobs=1,depth=16
   read: IOPS=168k, BW=10.3GiB/s (11.0GB/s)(103GiB/10001msec) |	   read: IOPS=172k, BW=10.5GiB/s (11.3GB/s)(105GiB/10001msec)
    slat (nsec): min=3463, max=75651, avg=4588.13, stdev=499. |	    slat (nsec): min=3756, max=65673, avg=4930.28, stdev=334.
    clat (usec): min=67, max=271, avg=90.06, stdev= 7.31      |	    clat (usec): min=83, max=243, avg=87.86, stdev= 2.04
     lat (usec): min=72, max=348, avg=94.68, stdev= 7.26      |	     lat (usec): min=87, max=310, avg=92.83, stdev= 2.05
     |  1.00th=[   81],  5.00th=[   84], 10.00th=[   85], 20. |	     |  1.00th=[   86],  5.00th=[   87], 10.00th=[   87], 20.
     | 30.00th=[   87], 40.00th=[   89], 50.00th=[   90], 60. |	     | 30.00th=[   87], 40.00th=[   88], 50.00th=[   88], 60.
     | 70.00th=[   92], 80.00th=[   93], 90.00th=[   96], 95. |	     | 70.00th=[   88], 80.00th=[   89], 90.00th=[   91], 95.
     | 99.00th=[  127], 99.50th=[  135], 99.90th=[  172], 99. |	     | 99.00th=[   96], 99.50th=[   99], 99.90th=[  108], 99.
     | 99.99th=[  194]					      |	     | 99.99th=[  114]
Type=read,Size=64K,NumJobs=1,depth=64				Type=read,Size=64K,NumJobs=1,depth=64
   read: IOPS=171k, BW=10.5GiB/s (11.2GB/s)(105GiB/10001msec) |	   read: IOPS=168k, BW=10.2GiB/s (10.0GB/s)(102GiB/10001msec)
    slat (usec): min=2, max=419, avg= 3.53, stdev= 1.24	      |	    slat (usec): min=2, max=390, avg= 4.15, stdev= 1.56
    clat (usec): min=63, max=973, avg=369.73, stdev=13.21     |	    clat (usec): min=37, max=947, avg=376.95, stdev=15.70
     lat (usec): min=69, max=1053, avg=373.30, stdev=13.09    |	     lat (usec): min=42, max=988, avg=381.15, stdev=16.25
     |  1.00th=[  338],  5.00th=[  359], 10.00th=[  359], 20. |	     |  1.00th=[  363],  5.00th=[  363], 10.00th=[  367], 20.
     | 30.00th=[  367], 40.00th=[  367], 50.00th=[  371], 60.	     | 30.00th=[  367], 40.00th=[  367], 50.00th=[  371], 60.
     | 70.00th=[  371], 80.00th=[  375], 90.00th=[  379], 95. |	     | 70.00th=[  388], 80.00th=[  392], 90.00th=[  400], 95.
     | 99.00th=[  404], 99.50th=[  416], 99.90th=[  461], 99. |	     | 99.00th=[  412], 99.50th=[  416], 99.90th=[  429], 99.
     | 99.99th=[  652]					      |	     | 99.99th=[  668]
Type=read,Size=64K,NumJobs=16,depth=1				Type=read,Size=64K,NumJobs=16,depth=1
   read: IOPS=171k, BW=10.4GiB/s (11.2GB/s)(104GiB/10001msec) |	   read: IOPS=172k, BW=10.5GiB/s (11.3GB/s)(105GiB/10002msec)
    slat (nsec): min=3795, max=63452, avg=5333.62, stdev=2920 |	    slat (nsec): min=3480, max=74756, avg=5273.21, stdev=2961
    clat (nsec): min=626, max=175700, avg=87848.01, stdev=183 |	    clat (nsec): min=616, max=453786, avg=87271.90, stdev=161
     lat (usec): min=26, max=180, avg=93.23, stdev=18.52      |	     lat (usec): min=26, max=458, avg=92.60, stdev=16.20
     |  1.00th=[   31],  5.00th=[   48], 10.00th=[   62], 20. |	     |  1.00th=[   37],  5.00th=[   57], 10.00th=[   68], 20.
     | 30.00th=[   85], 40.00th=[   90], 50.00th=[   93], 60. |	     | 30.00th=[   83], 40.00th=[   87], 50.00th=[   90], 60.
     | 70.00th=[   97], 80.00th=[  100], 90.00th=[  104], 95. |	     | 70.00th=[   95], 80.00th=[   98], 90.00th=[  104], 95.
     | 99.00th=[  125], 99.50th=[  130], 99.90th=[  139], 99. |	     | 99.00th=[  126], 99.50th=[  130], 99.90th=[  143], 99.
     | 99.99th=[  155]					      |	     | 99.99th=[  161]
Type=read,Size=64K,NumJobs=16,depth=16				Type=read,Size=64K,NumJobs=16,depth=16
   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10002msec)	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10002msec)
    slat (nsec): min=3878, max=65832, avg=5321.95, stdev=1310 |	    slat (nsec): min=3866, max=57337, avg=5204.47, stdev=1220
    clat (usec): min=104, max=3983, avg=1458.24, stdev=313.83 |	    clat (usec): min=116, max=3985, avg=1453.77, stdev=349.92
     lat (usec): min=111, max=3988, avg=1463.62, stdev=313.72 |	     lat (usec): min=128, max=3990, avg=1459.03, stdev=349.93
     |  1.00th=[ 1352],  5.00th=[ 1352], 10.00th=[ 1352], 20. |	     |  1.00th=[ 1254],  5.00th=[ 1270], 10.00th=[ 1270], 20.
     | 30.00th=[ 1352], 40.00th=[ 1369], 50.00th=[ 1369], 60. |	     | 30.00th=[ 1352], 40.00th=[ 1352], 50.00th=[ 1352], 60.
     | 70.00th=[ 1385], 80.00th=[ 1450], 90.00th=[ 1467], 95. |	     | 70.00th=[ 1369], 80.00th=[ 1401], 90.00th=[ 1467], 95.
     | 99.00th=[ 2769], 99.50th=[ 2769], 99.90th=[ 2802], 99. |	     | 99.00th=[ 2737], 99.50th=[ 2769], 99.90th=[ 2769], 99.
     | 99.99th=[ 2802]						     | 99.99th=[ 2802]
Type=read,Size=64K,NumJobs=16,depth=64				Type=read,Size=64K,NumJobs=16,depth=64
   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10006msec)	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10006msec)
    slat (usec): min=2, max=1287, avg=13.57, stdev=76.18      |	    slat (usec): min=2, max=1185, avg= 9.71, stdev=60.22
    clat (usec): min=116, max=15982, avg=5828.20, stdev=1647. |	    clat (usec): min=116, max=16490, avg=5824.81, stdev=1345.
     lat (usec): min=127, max=15986, avg=5841.81, stdev=1674. |	     lat (usec): min=133, max=16494, avg=5834.58, stdev=1367.
     |  1.00th=[ 5080],  5.00th=[ 5080], 10.00th=[ 5080], 20. |	     |  1.00th=[ 5407],  5.00th=[ 5407], 10.00th=[ 5473], 20.
     | 30.00th=[ 5080], 40.00th=[ 5145], 50.00th=[ 5145], 60. |	     | 30.00th=[ 5473], 40.00th=[ 5473], 50.00th=[ 5473], 60.
     | 70.00th=[ 5473], 80.00th=[ 5473], 90.00th=[10028], 95. |	     | 70.00th=[ 5473], 80.00th=[ 5473], 90.00th=[ 5538], 95.
     | 99.00th=[10814], 99.50th=[10945], 99.90th=[10945], 99. |	     | 99.00th=[10945], 99.50th=[10945], 99.90th=[11076], 99.
     | 99.99th=[11469]					      |	     | 99.99th=[11338]
Type=read,Size=64K,NumJobs=56,depth=1				Type=read,Size=64K,NumJobs=56,depth=1
   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10002msec)	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10002msec)
    slat (usec): min=3, max=155, avg= 4.93, stdev= 1.32	      |	    slat (nsec): min=3801, max=63985, avg=4896.45, stdev=1489
    clat (usec): min=39, max=624, avg=314.57, stdev=91.62     |	    clat (usec): min=29, max=550, avg=314.45, stdev=76.21
     lat (usec): min=48, max=629, avg=319.55, stdev=91.63     |	     lat (usec): min=35, max=555, avg=319.41, stdev=76.15
     |  1.00th=[  172],  5.00th=[  174], 10.00th=[  176], 20. |	     |  1.00th=[  174],  5.00th=[  178], 10.00th=[  194], 20.
     | 30.00th=[  302], 40.00th=[  302], 50.00th=[  306], 60. |	     | 30.00th=[  302], 40.00th=[  306], 50.00th=[  306], 60.
     | 70.00th=[  310], 80.00th=[  449], 90.00th=[  457], 95. |	     | 70.00th=[  310], 80.00th=[  322], 90.00th=[  457], 95.
     | 99.00th=[  478], 99.50th=[  482], 99.90th=[  494], 99. |	     | 99.00th=[  474], 99.50th=[  482], 99.90th=[  490], 99.
     | 99.99th=[  594]					      |	     | 99.99th=[  506]
Type=read,Size=64K,NumJobs=56,depth=16				Type=read,Size=64K,NumJobs=56,depth=16
   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10006msec)	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10006msec)
    slat (usec): min=3, max=156, avg= 5.40, stdev= 2.21	      |	    slat (usec): min=3, max=251, avg= 5.40, stdev= 2.34
    clat (usec): min=186, max=13998, avg=5112.23, stdev=1678. |	    clat (usec): min=236, max=14486, avg=5106.35, stdev=1241.
     lat (usec): min=201, max=14003, avg=5117.68, stdev=1678. |	     lat (usec): min=265, max=14491, avg=5111.80, stdev=1241.
     |  1.00th=[ 2442],  5.00th=[ 2442], 10.00th=[ 2474], 20. |	     |  1.00th=[ 2442],  5.00th=[ 2540], 10.00th=[ 4883], 20.
     | 30.00th=[ 4883], 40.00th=[ 4948], 50.00th=[ 4948], 60. |	     | 30.00th=[ 4948], 40.00th=[ 5080], 50.00th=[ 5080], 60.
     | 70.00th=[ 5014], 80.00th=[ 7373], 90.00th=[ 7439], 95. |	     | 70.00th=[ 5080], 80.00th=[ 5145], 90.00th=[ 7373], 95.
     | 99.00th=[ 9896], 99.50th=[ 9896], 99.90th=[10290], 99. |	     | 99.00th=[ 9896], 99.50th=[ 9896], 99.90th=[ 9896], 99.
     | 99.99th=[10290]					      |	     | 99.99th=[10028]
Type=read,Size=64K,NumJobs=56,depth=64				Type=read,Size=64K,NumJobs=56,depth=64
   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10016msec) |	   read: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10018msec)
    slat (usec): min=2, max=9793, avg=214.10, stdev=778.24    |	    slat (usec): min=2, max=10184, avg=189.22, stdev=637.78
    clat (usec): min=174, max=61025, avg=20233.29, stdev=9310 |	    clat (usec): min=347, max=61952, avg=20252.15, stdev=7056
     lat (usec): min=178, max=63984, avg=20447.45, stdev=9493 |	     lat (usec): min=351, max=65141, avg=20441.44, stdev=7203
     |  1.00th=[ 9110],  5.00th=[ 9110], 10.00th=[ 9503], 20. |	     |  1.00th=[ 9503],  5.00th=[ 9765], 10.00th=[ 9896], 20.
     | 30.00th=[17957], 40.00th=[18744], 50.00th=[18744], 60. |	     | 30.00th=[19530], 40.00th=[19530], 50.00th=[19792], 60.
     | 70.00th=[23987], 80.00th=[28443], 90.00th=[36439], 95. |	     | 70.00th=[20317], 80.00th=[20579], 90.00th=[29492], 95.
     | 99.00th=[41157], 99.50th=[56886], 99.90th=[56886], 99. |	     | 99.00th=[41157], 99.50th=[58983], 99.90th=[59507], 99.
     | 99.99th=[58459]					      |	     | 99.99th=[61080]
Type=write,Size=512B,NumJobs=1,depth=1				Type=write,Size=512B,NumJobs=1,depth=1
  write: IOPS=57.8k, BW=28.2MiB/s (29.6MB/s)(282MiB/10001msec |	  write: IOPS=56.0k, BW=27.4MiB/s (28.7MB/s)(274MiB/10001msec
    slat (nsec): min=2145, max=77559, avg=2556.80, stdev=1310 |	    slat (nsec): min=2112, max=43957, avg=2561.65, stdev=1423
    clat (nsec): min=626, max=100854, avg=14269.28, stdev=471 |	    clat (nsec): min=633, max=60031, avg=14812.68, stdev=5638
     lat (usec): min=12, max=107, avg=16.87, stdev= 5.13      |	     lat (nsec): min=12369, max=72124, avg=17416.39, stdev=60
     |  1.00th=[12480],  5.00th=[12480], 10.00th=[12480], 20. |	     |  1.00th=[12352],  5.00th=[12352], 10.00th=[12480], 20.
     | 30.00th=[12608], 40.00th=[12608], 50.00th=[12736], 60. |	     | 30.00th=[12480], 40.00th=[12608], 50.00th=[12608], 60.
     | 70.00th=[12864], 80.00th=[13248], 90.00th=[16512], 95. |	     | 70.00th=[12864], 80.00th=[14656], 90.00th=[20096], 95.
     | 99.00th=[34560], 99.50th=[35584], 99.90th=[43264], 99. |	     | 99.00th=[35072], 99.50th=[36608], 99.90th=[47360], 99.
     | 99.99th=[53504]					      |	     | 99.99th=[54016]
Type=write,Size=512B,NumJobs=1,depth=16				Type=write,Size=512B,NumJobs=1,depth=16
  write: IOPS=302k, BW=147MiB/s (154MB/s)(1472MiB/10001msec)  |	  write: IOPS=328k, BW=160MiB/s (168MB/s)(1601MiB/10001msec)
    slat (nsec): min=1637, max=35628, avg=1993.52, stdev=591. |	    slat (nsec): min=1641, max=32941, avg=2199.97, stdev=236.
    clat (usec): min=24, max=194, avg=50.75, stdev=12.31      |	    clat (usec): min=14, max=113, avg=46.29, stdev= 1.27
     lat (usec): min=25, max=198, avg=52.78, stdev=12.32      |	     lat (usec): min=16, max=115, avg=48.52, stdev= 1.28
     |  1.00th=[   41],  5.00th=[   41], 10.00th=[   41], 20. |	     |  1.00th=[43776],  5.00th=[44800], 10.00th=[45312], 20.
     | 30.00th=[   43], 40.00th=[   43], 50.00th=[   47], 60. |	     | 30.00th=[45824], 40.00th=[45824], 50.00th=[46336], 60.
     | 70.00th=[   54], 80.00th=[   59], 90.00th=[   70], 95. |	     | 70.00th=[46336], 80.00th=[46848], 90.00th=[47360], 95.
     | 99.00th=[   93], 99.50th=[  102], 99.90th=[  127], 99. |	     | 99.00th=[50432], 99.50th=[53504], 99.90th=[57600], 99.
     | 99.99th=[  145]					      |	     | 99.99th=[63744]
Type=write,Size=512B,NumJobs=1,depth=64				Type=write,Size=512B,NumJobs=1,depth=64
  write: IOPS=362k, BW=177MiB/s (185MB/s)(1769MiB/10001msec)  |	  write: IOPS=337k, BW=164MiB/s (172MB/s)(1644MiB/10001msec)
    slat (nsec): min=1644, max=36941, avg=1976.26, stdev=244. |	    slat (nsec): min=1642, max=43437, avg=2146.30, stdev=234.
    clat (usec): min=50, max=219, avg=174.38, stdev= 3.50     |	    clat (usec): min=16, max=235, avg=187.62, stdev= 2.46
     lat (usec): min=52, max=221, avg=176.39, stdev= 3.54     |	     lat (usec): min=18, max=237, avg=189.80, stdev= 2.48
     |  1.00th=[  172],  5.00th=[  172], 10.00th=[  172], 20. |	     |  1.00th=[  184],  5.00th=[  184], 10.00th=[  186], 20.
     | 30.00th=[  174], 40.00th=[  174], 50.00th=[  174], 60. |	     | 30.00th=[  186], 40.00th=[  188], 50.00th=[  188], 60.
     | 70.00th=[  176], 80.00th=[  176], 90.00th=[  178], 95. |	     | 70.00th=[  190], 80.00th=[  190], 90.00th=[  192], 95.
     | 99.00th=[  192], 99.50th=[  192], 99.90th=[  194], 99. |	     | 99.00th=[  194], 99.50th=[  196], 99.90th=[  200], 99.
     | 99.99th=[  198]					      |	     | 99.99th=[  204]
Type=write,Size=512B,NumJobs=16,depth=1				Type=write,Size=512B,NumJobs=16,depth=1
  write: IOPS=928k, BW=453MiB/s (475MB/s)(4531MiB/10001msec)  |	  write: IOPS=914k, BW=446MiB/s (468MB/s)(4462MiB/10001msec)
    slat (nsec): min=2034, max=70704, avg=2534.91, stdev=709. |	    slat (nsec): min=1818, max=83376, avg=2566.34, stdev=824.
    clat (nsec): min=588, max=855972, avg=14244.47, stdev=155 |	    clat (nsec): min=465, max=343253, avg=14459.07, stdev=181
     lat (usec): min=12, max=858, avg=16.82, stdev= 1.78      |	     lat (usec): min=11, max=345, avg=17.08, stdev= 2.09
     |  1.00th=[12480],  5.00th=[13120], 10.00th=[13248], 20. |	     |  1.00th=[12224],  5.00th=[12992], 10.00th=[13248], 20.
     | 30.00th=[13632], 40.00th=[13760], 50.00th=[14016], 60.	     | 30.00th=[13632], 40.00th=[13760], 50.00th=[14016], 60.
     | 70.00th=[14272], 80.00th=[14528], 90.00th=[15296], 95. |	     | 70.00th=[14528], 80.00th=[15040], 90.00th=[16320], 95.
     | 99.00th=[20864], 99.50th=[22912], 99.90th=[30848], 99. |	     | 99.00th=[21888], 99.50th=[23936], 99.90th=[32384], 99.
     | 99.99th=[34560]					      |	     | 99.99th=[35584]
Type=write,Size=512B,NumJobs=16,depth=16			Type=write,Size=512B,NumJobs=16,depth=16
  write: IOPS=4532k, BW=2213MiB/s (2320MB/s)(21.6GiB/10001mse |	  write: IOPS=3218k, BW=1571MiB/s (1648MB/s)(15.3GiB/10001mse
    slat (nsec): min=1608, max=191266, avg=2313.43, stdev=780 |	    slat (nsec): min=1621, max=154997, avg=3548.60, stdev=323
    clat (usec): min=13, max=557, avg=53.80, stdev=10.94      |	    clat (usec): min=14, max=473, avg=75.44, stdev=65.86
     lat (usec): min=15, max=574, avg=56.16, stdev=11.19      |	     lat (usec): min=17, max=489, avg=79.05, stdev=68.82
     |  1.00th=[   42],  5.00th=[   43], 10.00th=[   44], 20. |	     |  1.00th=[   45],  5.00th=[   45], 10.00th=[   46], 20.
     | 30.00th=[   49], 40.00th=[   51], 50.00th=[   53], 60. |	     | 30.00th=[   47], 40.00th=[   48], 50.00th=[   52], 60.
     | 70.00th=[   57], 80.00th=[   60], 90.00th=[   65], 95. |	     | 70.00th=[   61], 80.00th=[   64], 90.00th=[  178], 95.
     | 99.00th=[   85], 99.50th=[   90], 99.90th=[  109], 99. |	     | 99.00th=[  326], 99.50th=[  334], 99.90th=[  351], 99.
     | 99.99th=[  396]					      |	     | 99.99th=[  367]
Type=write,Size=512B,NumJobs=16,depth=64			Type=write,Size=512B,NumJobs=16,depth=64
  write: IOPS=5003k, BW=2443MiB/s (2562MB/s)(23.9GiB/10001mse |	  write: IOPS=3973k, BW=1940MiB/s (2034MB/s)(18.0GiB/10002mse
    slat (nsec): min=1613, max=68192, avg=2314.76, stdev=1502 |	    slat (nsec): min=1627, max=171980, avg=2926.70, stdev=209
    clat (usec): min=32, max=588, avg=201.99, stdev=31.05     |	    clat (usec): min=10, max=1606, avg=254.29, stdev=176.36
     lat (usec): min=35, max=590, avg=204.35, stdev=31.42     |	     lat (usec): min=16, max=1615, avg=257.27, stdev=178.32
     |  1.00th=[  176],  5.00th=[  178], 10.00th=[  178], 20. |	     |  1.00th=[  184],  5.00th=[  186], 10.00th=[  188], 20.
     | 30.00th=[  182], 40.00th=[  182], 50.00th=[  186], 60. |	     | 30.00th=[  194], 40.00th=[  202], 50.00th=[  237], 60.
     | 70.00th=[  212], 80.00th=[  217], 90.00th=[  241], 95. |	     | 70.00th=[  255], 80.00th=[  260], 90.00th=[  293], 95.
     | 99.00th=[  310], 99.50th=[  318], 99.90th=[  330], 99. |	     | 99.00th=[ 1385], 99.50th=[ 1418], 99.90th=[ 1467], 99.
     | 99.99th=[  351]					      |	     | 99.99th=[ 1500]
Type=write,Size=512B,NumJobs=56,depth=1				Type=write,Size=512B,NumJobs=56,depth=1
  write: IOPS=2299k, BW=1123MiB/s (1177MB/s)(10.0GiB/10002mse |	  write: IOPS=2369k, BW=1157MiB/s (1213MB/s)(11.3GiB/10002mse
    slat (nsec): min=1875, max=1144.8k, avg=3714.43, stdev=29 |	    slat (nsec): min=1823, max=501953, avg=3567.43, stdev=201
    clat (nsec): min=471, max=1780.1k, avg=19859.06, stdev=11 |	    clat (nsec): min=449, max=1722.5k, avg=19318.77, stdev=42
     lat (usec): min=10, max=1783, avg=23.66, stdev=12.06     |	     lat (usec): min=11, max=1724, avg=22.98, stdev= 4.83
     |  1.00th=[   13],  5.00th=[   15], 10.00th=[   16], 20. |	     |  1.00th=[12992],  5.00th=[14656], 10.00th=[15680], 20.
     | 30.00th=[   18], 40.00th=[   19], 50.00th=[   19], 60. |	     | 30.00th=[17536], 40.00th=[18048], 50.00th=[18816], 60.
     | 70.00th=[   21], 80.00th=[   22], 90.00th=[   25], 95. |	     | 70.00th=[20352], 80.00th=[21632], 90.00th=[23424], 95.
     | 99.00th=[   37], 99.50th=[   41], 99.90th=[   70], 99. |	     | 99.00th=[31872], 99.50th=[35584], 99.90th=[41728], 99.
     | 99.99th=[  553]					      |	     | 99.99th=[54016]
Type=write,Size=512B,NumJobs=56,depth=16			Type=write,Size=512B,NumJobs=56,depth=16
  write: IOPS=8134k, BW=3972MiB/s (4165MB/s)(38.8GiB/10001mse |	  write: IOPS=6237k, BW=3046MiB/s (3193MB/s)(29.7GiB/10002mse
    slat (nsec): min=1697, max=8922.2k, avg=5089.14, stdev=73 |	    slat (nsec): min=1640, max=2381.5k, avg=5462.05, stdev=67
    clat (usec): min=6, max=9049, avg=104.25, stdev=53.53     |	    clat (usec): min=5, max=10435, avg=137.19, stdev=82.68
     lat (usec): min=33, max=9053, avg=109.44, stdev=56.00    |	     lat (usec): min=15, max=10447, avg=142.77, stdev=85.58
     |  1.00th=[   65],  5.00th=[   69], 10.00th=[   70], 20. |	     |  1.00th=[   63],  5.00th=[   68], 10.00th=[   70], 20.
     | 30.00th=[   74], 40.00th=[   75], 50.00th=[   77], 60. |	     | 30.00th=[   85], 40.00th=[   96], 50.00th=[  104], 60.
     | 70.00th=[   92], 80.00th=[  161], 90.00th=[  186], 95. |	     | 70.00th=[  149], 80.00th=[  210], 90.00th=[  265], 95.
     | 99.00th=[  255], 99.50th=[  326], 99.90th=[  424], 99. |	     | 99.00th=[  363], 99.50th=[  396], 99.90th=[  553], 99.
     | 99.99th=[  775]					      |	     | 99.99th=[  873]
Type=write,Size=512B,NumJobs=56,depth=64			Type=write,Size=512B,NumJobs=56,depth=64
  write: IOPS=8214k, BW=4011MiB/s (4205MB/s)(39.2GiB/10002mse |	  write: IOPS=7488k, BW=3656MiB/s (3834MB/s)(35.7GiB/10002mse
    slat (nsec): min=1699, max=946558, avg=5094.80, stdev=734 |	    slat (nsec): min=1666, max=7929.3k, avg=5382.53, stdev=12
    clat (nsec): min=1952, max=1945.9k, avg=430363.82, stdev= |	    clat (usec): min=14, max=12517, avg=472.19, stdev=331.30
     lat (usec): min=43, max=1949, avg=435.56, stdev=208.15   |	     lat (usec): min=17, max=12520, avg=477.70, stdev=334.84
     |  1.00th=[  281],  5.00th=[  289], 10.00th=[  297], 20. |	     |  1.00th=[  265],  5.00th=[  277], 10.00th=[  289], 20.
     | 30.00th=[  306], 40.00th=[  310], 50.00th=[  310], 60. |	     | 30.00th=[  302], 40.00th=[  306], 50.00th=[  314], 60.
     | 70.00th=[  330], 80.00th=[  709], 90.00th=[  750], 95. |	     | 70.00th=[  347], 80.00th=[  562], 90.00th=[ 1057], 95.
     | 99.00th=[  971], 99.50th=[ 1254], 99.90th=[ 1450], 99. |	     | 99.00th=[ 1434], 99.50th=[ 1696], 99.90th=[ 2040], 99.
     | 99.99th=[ 1582]					      |	     | 99.99th=[ 2474]
Type=write,Size=4K,NumJobs=1,depth=1				Type=write,Size=4K,NumJobs=1,depth=1
  write: IOPS=53.8k, BW=210MiB/s (220MB/s)(2100MiB/10001msec) |	  write: IOPS=55.8k, BW=218MiB/s (229MB/s)(2180MiB/10000msec)
    slat (nsec): min=2141, max=79859, avg=2640.37, stdev=1677 |	    slat (nsec): min=2138, max=44613, avg=2552.61, stdev=1404
    clat (nsec): min=616, max=110520, avg=15483.87, stdev=567 |	    clat (nsec): min=627, max=624863, avg=14910.38, stdev=517
     lat (usec): min=13, max=116, avg=18.16, stdev= 6.15      |	     lat (usec): min=13, max=627, avg=17.50, stdev= 5.55
     |  1.00th=[12864],  5.00th=[12992], 10.00th=[12992], 20. |	     |  1.00th=[12736],  5.00th=[12864], 10.00th=[12864], 20.
     | 30.00th=[13120], 40.00th=[13248], 50.00th=[13248], 60. |	     | 30.00th=[12992], 40.00th=[12992], 50.00th=[12992], 60.
     | 70.00th=[13504], 80.00th=[15552], 90.00th=[20864], 95. |	     | 70.00th=[13248], 80.00th=[13632], 90.00th=[20352], 95.
     | 99.00th=[36096], 99.50th=[39168], 99.90th=[50432], 99. |	     | 99.00th=[35584], 99.50th=[37120], 99.90th=[44800], 99.
     | 99.99th=[74240]					      |	     | 99.99th=[53504]
Type=write,Size=4K,NumJobs=1,depth=16				Type=write,Size=4K,NumJobs=1,depth=16
  write: IOPS=280k, BW=1093MiB/s (1146MB/s)(10.7GiB/10001msec |	  write: IOPS=325k, BW=1271MiB/s (1332MB/s)(12.4GiB/10001msec
    slat (nsec): min=1746, max=32381, avg=2137.66, stdev=633. |	    slat (nsec): min=1697, max=32986, avg=2232.67, stdev=205.
    clat (usec): min=41, max=167, avg=54.70, stdev=12.58      |	    clat (nsec): min=14620, max=89670, avg=46638.22, stdev=11
     lat (usec): min=43, max=169, avg=56.87, stdev=12.60      |	     lat (nsec): min=16814, max=91833, avg=48907.48, stdev=11
     |  1.00th=[   43],  5.00th=[   44], 10.00th=[   44], 20. |	     |  1.00th=[44800],  5.00th=[45312], 10.00th=[45824], 20.
     | 30.00th=[   45], 40.00th=[   49], 50.00th=[   52], 60. |	     | 30.00th=[46336], 40.00th=[46336], 50.00th=[46336], 60.
     | 70.00th=[   59], 80.00th=[   64], 90.00th=[   72], 95. |	     | 70.00th=[46848], 80.00th=[46848], 90.00th=[47360], 95.
     | 99.00th=[   96], 99.50th=[  111], 99.90th=[  130], 99. |	     | 99.00th=[50944], 99.50th=[53504], 99.90th=[58112], 99.
     | 99.99th=[  147]					      |	     | 99.99th=[63744]
Type=write,Size=4K,NumJobs=1,depth=64				Type=write,Size=4K,NumJobs=1,depth=64
  write: IOPS=340k, BW=1329MiB/s (1394MB/s)(12.0GiB/10001msec |	  write: IOPS=322k, BW=1259MiB/s (1320MB/s)(12.3GiB/10001msec
    slat (nsec): min=1728, max=34272, avg=2133.44, stdev=216. |	    slat (nsec): min=1697, max=33655, avg=2273.54, stdev=208.
    clat (usec): min=76, max=257, avg=185.64, stdev= 1.60     |	    clat (usec): min=18, max=238, avg=195.95, stdev= 2.07
     lat (usec): min=78, max=259, avg=187.81, stdev= 1.61     |	     lat (usec): min=20, max=241, avg=198.26, stdev= 2.08
     |  1.00th=[  184],  5.00th=[  184], 10.00th=[  184], 20. |	     |  1.00th=[  192],  5.00th=[  194], 10.00th=[  194], 20.
     | 30.00th=[  186], 40.00th=[  186], 50.00th=[  186], 60. |	     | 30.00th=[  196], 40.00th=[  196], 50.00th=[  196], 60.
     | 70.00th=[  186], 80.00th=[  188], 90.00th=[  188], 95. |	     | 70.00th=[  198], 80.00th=[  198], 90.00th=[  198], 95.
     | 99.00th=[  190], 99.50th=[  190], 99.90th=[  194], 99. |	     | 99.00th=[  202], 99.50th=[  204], 99.90th=[  208], 99.
     | 99.99th=[  200]					      |	     | 99.99th=[  212]
Type=write,Size=4K,NumJobs=16,depth=1				Type=write,Size=4K,NumJobs=16,depth=1
  write: IOPS=863k, BW=3371MiB/s (3535MB/s)(32.9GiB/10001msec |	  write: IOPS=855k, BW=3339MiB/s (3502MB/s)(32.6GiB/10001msec
    slat (nsec): min=1980, max=66528, avg=2566.89, stdev=734. |	    slat (nsec): min=1832, max=65363, avg=2579.38, stdev=826.
    clat (nsec): min=583, max=653122, avg=15498.34, stdev=199 |	    clat (nsec): min=472, max=134403, avg=15646.36, stdev=196
     lat (usec): min=12, max=655, avg=18.11, stdev= 2.17      |	     lat (usec): min=12, max=136, avg=18.28, stdev= 2.21
     |  1.00th=[12992],  5.00th=[13888], 10.00th=[14144], 20. |	     |  1.00th=[13120],  5.00th=[13760], 10.00th=[14016], 20.
     | 30.00th=[14656], 40.00th=[14912], 50.00th=[15168], 60. |	     | 30.00th=[14656], 40.00th=[14912], 50.00th=[15168], 60.
     | 70.00th=[15680], 80.00th=[16192], 90.00th=[17280], 95. |	     | 70.00th=[15936], 80.00th=[16512], 90.00th=[17536], 95.
     | 99.00th=[23168], 99.50th=[25216], 99.90th=[33536], 99. |	     | 99.00th=[23680], 99.50th=[25728], 99.90th=[33536], 99.
     | 99.99th=[72192]					      |	     | 99.99th=[37120]
Type=write,Size=4K,NumJobs=16,depth=16				Type=write,Size=4K,NumJobs=16,depth=16
  write: IOPS=2631k, BW=10.0GiB/s (10.8GB/s)(100GiB/10002msec |	  write: IOPS=1826k, BW=7133MiB/s (7480MB/s)(69.7GiB/10001mse
    slat (nsec): min=1629, max=70383, avg=2374.75, stdev=939. |	    slat (nsec): min=1639, max=113071, avg=3312.71, stdev=236
    clat (usec): min=28, max=370, avg=94.53, stdev=12.94      |	    clat (usec): min=40, max=454, avg=136.35, stdev=35.22
     lat (usec): min=31, max=387, avg=96.95, stdev=12.90      |	     lat (usec): min=43, max=456, avg=139.72, stdev=35.40
     |  1.00th=[   72],  5.00th=[   78], 10.00th=[   80], 20. |	     |  1.00th=[   72],  5.00th=[   91], 10.00th=[  101], 20.
     | 30.00th=[   89], 40.00th=[   94], 50.00th=[   97], 60. |	     | 30.00th=[  121], 40.00th=[  127], 50.00th=[  133], 60.
     | 70.00th=[  100], 80.00th=[  102], 90.00th=[  105], 95. |	     | 70.00th=[  143], 80.00th=[  149], 90.00th=[  182], 95.
     | 99.00th=[  141], 99.50th=[  172], 99.90th=[  190], 99. |	     | 99.00th=[  258], 99.50th=[  273], 99.90th=[  314], 99.
     | 99.99th=[  231]					      |	     | 99.99th=[  367]
Type=write,Size=4K,NumJobs=16,depth=64				Type=write,Size=4K,NumJobs=16,depth=64
  write: IOPS=2615k, BW=9.97GiB/s (10.7GB/s)(99.7GiB/10001mse |	  write: IOPS=1875k, BW=7325MiB/s (7680MB/s)(71.5GiB/10001mse
    slat (nsec): min=1633, max=196327, avg=2501.76, stdev=263 |	    slat (nsec): min=1638, max=627057, avg=3232.91, stdev=324
    clat (usec): min=18, max=1044, avg=388.76, stdev=64.99    |	    clat (usec): min=49, max=1357, avg=542.32, stdev=162.71
     lat (usec): min=21, max=1052, avg=391.30, stdev=65.69    |	     lat (usec): min=52, max=1384, avg=545.61, stdev=162.80
     |  1.00th=[  318],  5.00th=[  338], 10.00th=[  347], 20. |	     |  1.00th=[  202],  5.00th=[  277], 10.00th=[  330], 20.
     | 30.00th=[  367], 40.00th=[  375], 50.00th=[  383], 60. |	     | 30.00th=[  453], 40.00th=[  498], 50.00th=[  537], 60.
     | 70.00th=[  392], 80.00th=[  396], 90.00th=[  412], 95. |	     | 70.00th=[  635], 80.00th=[  685], 90.00th=[  758], 95.
     | 99.00th=[  734], 99.50th=[  750], 99.90th=[  775], 99. |	     | 99.00th=[  922], 99.50th=[  955], 99.90th=[ 1029], 99.
     | 99.99th=[  799]					      |	     | 99.99th=[ 1123]
Type=write,Size=4K,NumJobs=56,depth=1				Type=write,Size=4K,NumJobs=56,depth=1
  write: IOPS=1419k, BW=5545MiB/s (5814MB/s)(54.2GiB/10002mse |	  write: IOPS=1416k, BW=5533MiB/s (5802MB/s)(54.0GiB/10002mse
    slat (nsec): min=1813, max=624655, avg=3014.79, stdev=214 |	    slat (nsec): min=1846, max=533231, avg=2881.71, stdev=111
    clat (nsec): min=498, max=1174.4k, avg=35813.70, stdev=12 |	    clat (nsec): min=498, max=4453.2k, avg=36057.87, stdev=38
     lat (usec): min=13, max=1177, avg=38.89, stdev=12.35     |	     lat (usec): min=15, max=4456, avg=39.01, stdev= 4.01
     |  1.00th=[   27],  5.00th=[   31], 10.00th=[   32], 20. |	     |  1.00th=[29312],  5.00th=[32128], 10.00th=[33024], 20.
     | 30.00th=[   35], 40.00th=[   35], 50.00th=[   36], 60. |	     | 30.00th=[34560], 40.00th=[35072], 50.00th=[35584], 60.
     | 70.00th=[   37], 80.00th=[   38], 90.00th=[   40], 95. |	     | 70.00th=[37120], 80.00th=[37632], 90.00th=[38656], 95.
     | 99.00th=[   51], 99.50th=[   56], 99.90th=[  229], 99. |	     | 99.00th=[49408], 99.50th=[52480], 99.90th=[56576], 99.
     | 99.99th=[  498]					      |	     | 99.99th=[64768]
Type=write,Size=4K,NumJobs=56,depth=16				Type=write,Size=4K,NumJobs=56,depth=16
  write: IOPS=2489k, BW=9724MiB/s (10.2GB/s)(94.0GiB/10002mse |	  write: IOPS=1616k, BW=6314MiB/s (6621MB/s)(61.7GiB/10001mse
    slat (nsec): min=1637, max=742083, avg=3046.89, stdev=226 |	    slat (nsec): min=1649, max=738008, avg=3245.81, stdev=163
    clat (usec): min=39, max=1654, avg=356.23, stdev=126.26   |	    clat (usec): min=44, max=5046, avg=550.36, stdev=143.07
     lat (usec): min=44, max=1657, avg=359.34, stdev=126.24   |	     lat (usec): min=46, max=5049, avg=553.68, stdev=143.03
     |  1.00th=[  217],  5.00th=[  237], 10.00th=[  247], 20. |	     |  1.00th=[  215],  5.00th=[  343], 10.00th=[  404], 20.
     | 30.00th=[  277], 40.00th=[  285], 50.00th=[  297], 60. |	     | 30.00th=[  490], 40.00th=[  510], 50.00th=[  529], 60.
     | 70.00th=[  383], 80.00th=[  474], 90.00th=[  562], 95. |	     | 70.00th=[  586], 80.00th=[  660], 90.00th=[  758], 95.
     | 99.00th=[  709], 99.50th=[  742], 99.90th=[  865], 99. |	     | 99.00th=[  938], 99.50th=[  996], 99.90th=[ 1106], 99.
     | 99.99th=[ 1074]					      |	     | 99.99th=[ 1221]
Type=write,Size=4K,NumJobs=56,depth=64				Type=write,Size=4K,NumJobs=56,depth=64
  write: IOPS=2216k, BW=8657MiB/s (9078MB/s)(84.6GiB/10003mse |	  write: IOPS=1625k, BW=6348MiB/s (6656MB/s)(62.0GiB/10003mse
    slat (nsec): min=1658, max=1799.0k, avg=16827.79, stdev=6 |	    slat (nsec): min=1656, max=1933.3k, avg=21620.58, stdev=7
    clat (usec): min=40, max=7314, avg=1599.52, stdev=743.15  |	    clat (usec): min=66, max=8663, avg=2182.95, stdev=879.90
     lat (usec): min=43, max=7318, avg=1616.41, stdev=754.88  |	     lat (usec): min=69, max=8669, avg=2204.64, stdev=895.51
     |  1.00th=[  400],  5.00th=[  644], 10.00th=[  734], 20. |	     |  1.00th=[  693],  5.00th=[  971], 10.00th=[ 1074], 20.
     | 30.00th=[ 1254], 40.00th=[ 1418], 50.00th=[ 1516], 60. |	     | 30.00th=[ 1942], 40.00th=[ 2114], 50.00th=[ 2212], 60.
     | 70.00th=[ 1778], 80.00th=[ 2057], 90.00th=[ 2638], 95. |	     | 70.00th=[ 2343], 80.00th=[ 2507], 90.00th=[ 3392], 95.
     | 99.00th=[ 3884], 99.50th=[ 4178], 99.90th=[ 4817], 99. |	     | 99.00th=[ 4883], 99.50th=[ 5211], 99.90th=[ 5800], 99.
     | 99.99th=[ 5604]					      |	     | 99.99th=[ 6587]
Type=write,Size=64K,NumJobs=1,depth=1				Type=write,Size=64K,NumJobs=1,depth=1
  write: IOPS=23.1k, BW=1444MiB/s (1515MB/s)(14.1GiB/10001mse |	  write: IOPS=23.3k, BW=1454MiB/s (1525MB/s)(14.2GiB/10001mse
    slat (nsec): min=4822, max=98365, avg=6876.23, stdev=3203 |	    slat (nsec): min=4849, max=39702, avg=6880.02, stdev=3240
    clat (usec): min=3, max=121, avg=35.83, stdev= 9.35	      |	    clat (usec): min=17, max=133, avg=35.53, stdev= 9.39
     lat (usec): min=33, max=133, avg=42.75, stdev=10.50      |	     lat (usec): min=33, max=146, avg=42.46, stdev=10.52
     |  1.00th=[29568],  5.00th=[29568], 10.00th=[29568], 20. |	     |  1.00th=[29312],  5.00th=[29312], 10.00th=[29568], 20.
     | 30.00th=[29824], 40.00th=[29824], 50.00th=[30080], 60. |	     | 30.00th=[29568], 40.00th=[29824], 50.00th=[29824], 60.
     | 70.00th=[37120], 80.00th=[44800], 90.00th=[49408], 95. |	     | 70.00th=[36608], 80.00th=[43776], 90.00th=[49408], 95.
     | 99.00th=[68096], 99.50th=[72192], 99.90th=[85504], 99. |	     | 99.00th=[68096], 99.50th=[72192], 99.90th=[85504], 99.
     | 99.99th=[90624]						     | 99.99th=[90624]
Type=write,Size=64K,NumJobs=1,depth=16				Type=write,Size=64K,NumJobs=1,depth=16
  write: IOPS=140k, BW=8743MiB/s (9167MB/s)(85.4GiB/10001msec |	  write: IOPS=146k, BW=9109MiB/s (9552MB/s)(88.0GiB/10001msec
    slat (nsec): min=3861, max=47468, avg=5274.03, stdev=845. |	    slat (nsec): min=3787, max=44724, avg=5539.44, stdev=622.
    clat (usec): min=82, max=297, avg=108.76, stdev=13.98     |	    clat (usec): min=76, max=228, avg=103.90, stdev=11.08
     lat (usec): min=87, max=302, avg=114.07, stdev=13.98     |	     lat (usec): min=82, max=233, avg=109.48, stdev=11.10
     |  1.00th=[   92],  5.00th=[   96], 10.00th=[   98], 20. |	     |  1.00th=[   90],  5.00th=[   93], 10.00th=[   95], 20.
     | 30.00th=[  102], 40.00th=[  104], 50.00th=[  105], 60. |	     | 30.00th=[   98], 40.00th=[   99], 50.00th=[  101], 60.
     | 70.00th=[  111], 80.00th=[  114], 90.00th=[  123], 95. |	     | 70.00th=[  105], 80.00th=[  109], 90.00th=[  120], 95.
     | 99.00th=[  167], 99.50th=[  180], 99.90th=[  206], 99. |	     | 99.00th=[  141], 99.50th=[  149], 99.90th=[  182], 99.
     | 99.99th=[  239]					      |	     | 99.99th=[  200]
Type=write,Size=64K,NumJobs=1,depth=64				Type=write,Size=64K,NumJobs=1,depth=64
  write: IOPS=144k, BW=8977MiB/s (9413MB/s)(87.7GiB/10001msec |	  write: IOPS=139k, BW=8708MiB/s (9131MB/s)(85.0GiB/10001msec
    slat (nsec): min=2697, max=72397, avg=4695.54, stdev=1467 |	    slat (nsec): min=2676, max=64177, avg=5388.79, stdev=1847
    clat (usec): min=125, max=987, avg=440.52, stdev=70.72    |	    clat (usec): min=39, max=894, avg=453.54, stdev=56.76
     lat (usec): min=135, max=993, avg=445.25, stdev=70.70    |	     lat (usec): min=45, max=898, avg=458.97, stdev=57.24
     |  1.00th=[  330],  5.00th=[  338], 10.00th=[  347], 20. |	     |  1.00th=[  338],  5.00th=[  355], 10.00th=[  367], 20.
     | 30.00th=[  396], 40.00th=[  412], 50.00th=[  437], 60. |	     | 30.00th=[  408], 40.00th=[  457], 50.00th=[  478], 60.
     | 70.00th=[  478], 80.00th=[  498], 90.00th=[  529], 95. |	     | 70.00th=[  494], 80.00th=[  498], 90.00th=[  506], 95.
     | 99.00th=[  619], 99.50th=[  668], 99.90th=[  791], 99. |	     | 99.00th=[  545], 99.50th=[  562], 99.90th=[  734], 99.
     | 99.99th=[  889]					      |	     | 99.99th=[  848]
Type=write,Size=64K,NumJobs=16,depth=1				Type=write,Size=64K,NumJobs=16,depth=1
  write: IOPS=174k, BW=10.6GiB/s (11.4GB/s)(106GiB/10001msec) |	  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10001msec)
    slat (nsec): min=4083, max=63053, avg=5530.24, stdev=1175 |	    slat (nsec): min=4383, max=73755, avg=5607.60, stdev=1278
    clat (usec): min=17, max=167, avg=85.80, stdev= 7.67      |	    clat (usec): min=22, max=424, avg=85.24, stdev= 8.07
     lat (usec): min=37, max=172, avg=91.37, stdev= 7.76      |	     lat (usec): min=38, max=430, avg=90.90, stdev= 8.07
     |  1.00th=[   60],  5.00th=[   76], 10.00th=[   80], 20. |	     |  1.00th=[   68],  5.00th=[   76], 10.00th=[   79], 20.
     | 30.00th=[   84], 40.00th=[   85], 50.00th=[   86], 60. |	     | 30.00th=[   82], 40.00th=[   83], 50.00th=[   84], 60.
     | 70.00th=[   88], 80.00th=[   90], 90.00th=[   94], 95. |	     | 70.00th=[   87], 80.00th=[   91], 90.00th=[   98], 95.
     | 99.00th=[  109], 99.50th=[  115], 99.90th=[  128], 99. |	     | 99.00th=[  106], 99.50th=[  112], 99.90th=[  124], 99.
     | 99.99th=[  143]					      |	     | 99.99th=[  137]
Type=write,Size=64K,NumJobs=16,depth=16				Type=write,Size=64K,NumJobs=16,depth=16
  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10002msec) |	  write: IOPS=176k, BW=10.7GiB/s (11.5GB/s)(107GiB/10002msec)
    slat (nsec): min=3477, max=59587, avg=6630.17, stdev=1304 |	    slat (nsec): min=3611, max=70931, avg=6507.12, stdev=1389
    clat (usec): min=125, max=5561, avg=1451.88, stdev=382.58 |	    clat (usec): min=109, max=3915, avg=1451.13, stdev=409.41
     lat (usec): min=136, max=5567, avg=1458.55, stdev=382.73 |	     lat (usec): min=129, max=3920, avg=1457.68, stdev=409.50
     |  1.00th=[ 1254],  5.00th=[ 1270], 10.00th=[ 1270], 20.	     |  1.00th=[ 1254],  5.00th=[ 1270], 10.00th=[ 1270], 20.
     | 30.00th=[ 1352], 40.00th=[ 1352], 50.00th=[ 1352], 60. |	     | 30.00th=[ 1270], 40.00th=[ 1270], 50.00th=[ 1287], 60.
     | 70.00th=[ 1369], 80.00th=[ 1369], 90.00th=[ 1401], 95. |	     | 70.00th=[ 1352], 80.00th=[ 1369], 90.00th=[ 2540], 95.
     | 99.00th=[ 2737], 99.50th=[ 2737], 99.90th=[ 2769], 99.	     | 99.00th=[ 2737], 99.50th=[ 2737], 99.90th=[ 2769], 99.
     | 99.99th=[ 2802]					      |	     | 99.99th=[ 2769]
Type=write,Size=64K,NumJobs=16,depth=64				Type=write,Size=64K,NumJobs=16,depth=64
  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10004msec)	  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10004msec)
    slat (usec): min=2, max=6867, avg=10.05, stdev=84.22      |	    slat (nsec): min=3042, max=96073, avg=6866.24, stdev=2170
    clat (usec): min=114, max=17739, avg=5829.80, stdev=1403. |	    clat (usec): min=124, max=10103, avg=5850.40, stdev=798.1
     lat (usec): min=121, max=17748, avg=5839.90, stdev=1417. |	     lat (usec): min=131, max=10109, avg=5857.31, stdev=798.1
     |  1.00th=[ 4228],  5.00th=[ 4490], 10.00th=[ 4621], 20. |	     |  1.00th=[ 4752],  5.00th=[ 4883], 10.00th=[ 4948], 20.
     | 30.00th=[ 4948], 40.00th=[ 5145], 50.00th=[ 5473], 60. |	     | 30.00th=[ 5276], 40.00th=[ 5407], 50.00th=[ 5669], 60.
     | 70.00th=[ 6194], 80.00th=[ 6718], 90.00th=[ 7242], 95. |	     | 70.00th=[ 6259], 80.00th=[ 6587], 90.00th=[ 7046], 95.
     | 99.00th=[11338], 99.50th=[12125], 99.90th=[14484], 99. |	     | 99.00th=[ 7767], 99.50th=[ 7898], 99.90th=[ 8455], 99.
     | 99.99th=[16188]					      |	     | 99.99th=[ 8848]
Type=write,Size=64K,NumJobs=56,depth=1				Type=write,Size=64K,NumJobs=56,depth=1
  write: IOPS=176k, BW=10.7GiB/s (11.5GB/s)(107GiB/10001msec) |	  write: IOPS=175k, BW=10.7GiB/s (11.4GB/s)(107GiB/10002msec)
    slat (nsec): min=4056, max=63403, avg=6384.23, stdev=1521 |	    slat (nsec): min=4029, max=63780, avg=5771.34, stdev=1537
    clat (usec): min=57, max=4543, avg=312.06, stdev=48.67    |	    clat (usec): min=32, max=979, avg=314.28, stdev=57.25
     lat (usec): min=66, max=4549, avg=318.50, stdev=48.73    |	     lat (usec): min=40, max=985, avg=320.11, stdev=57.29
     |  1.00th=[  188],  5.00th=[  289], 10.00th=[  297], 20. |	     |  1.00th=[  186],  5.00th=[  293], 10.00th=[  297], 20.
     | 30.00th=[  306], 40.00th=[  306], 50.00th=[  310], 60. |	     | 30.00th=[  297], 40.00th=[  302], 50.00th=[  302], 60.
     | 70.00th=[  310], 80.00th=[  314], 90.00th=[  318], 95. |	     | 70.00th=[  310], 80.00th=[  310], 90.00th=[  326], 95.
     | 99.00th=[  478], 99.50th=[  611], 99.90th=[  627], 99. |	     | 99.00th=[  603], 99.50th=[  611], 99.90th=[  627], 99.
     | 99.99th=[  750]						     | 99.99th=[  750]
Type=write,Size=64K,NumJobs=56,depth=16				Type=write,Size=64K,NumJobs=56,depth=16
  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10005msec)	  write: IOPS=175k, BW=10.7GiB/s (11.5GB/s)(107GiB/10005msec)
    slat (usec): min=3, max=197, avg= 7.76, stdev= 2.34	      |	    slat (nsec): min=3542, max=89310, avg=7585.26, stdev=2460
    clat (usec): min=183, max=15544, avg=5102.42, stdev=1298. |	    clat (usec): min=190, max=14367, avg=5116.17, stdev=1396.
     lat (usec): min=193, max=15551, avg=5110.24, stdev=1298. |	     lat (usec): min=198, max=14375, avg=5123.82, stdev=1396.
     |  1.00th=[ 2442],  5.00th=[ 2540], 10.00th=[ 2573], 20. |	     |  1.00th=[ 2540],  5.00th=[ 2540], 10.00th=[ 2573], 20.
     | 30.00th=[ 5080], 40.00th=[ 5080], 50.00th=[ 5080], 60. |	     | 30.00th=[ 5080], 40.00th=[ 5080], 50.00th=[ 5145], 60.
     | 70.00th=[ 5145], 80.00th=[ 5538], 90.00th=[ 6718], 95. |	     | 70.00th=[ 5145], 80.00th=[ 5669], 90.00th=[ 6849], 95.
     | 99.00th=[ 8717], 99.50th=[ 9241], 99.90th=[10552], 99. |	     | 99.00th=[ 8979], 99.50th=[ 9634], 99.90th=[11207], 99.
     | 99.99th=[12125]					      |	     | 99.99th=[12780]
Type=write,Size=64K,NumJobs=56,depth=64				Type=write,Size=64K,NumJobs=56,depth=64
  write: IOPS=175k, BW=10.7GiB/s (11.4GB/s)(107GiB/10013msec) |	  write: IOPS=174k, BW=10.6GiB/s (11.4GB/s)(107GiB/10011msec)
    slat (usec): min=2, max=40799, avg=182.18, stdev=918.40   |	    slat (usec): min=3, max=32839, avg=176.71, stdev=925.53
    clat (usec): min=178, max=107843, avg=20332.76, stdev=698 |	    clat (usec): min=189, max=132076, avg=20366.61, stdev=728
     lat (usec): min=182, max=107849, avg=20515.01, stdev=709 |	     lat (usec): min=218, max=132086, avg=20543.39, stdev=740
     |  1.00th=[ 7832],  5.00th=[ 9372], 10.00th=[11469], 20. |	     |  1.00th=[ 7898],  5.00th=[ 9110], 10.00th=[10945], 20.
     | 30.00th=[18220], 40.00th=[19530], 50.00th=[20317], 60. |	     | 30.00th=[17957], 40.00th=[19530], 50.00th=[20317], 60.
     | 70.00th=[20841], 80.00th=[22938], 90.00th=[27657], 95. |	     | 70.00th=[21103], 80.00th=[23462], 90.00th=[28705], 95.
     | 99.00th=[45876], 99.50th=[51643], 99.90th=[64226], 99. |	     | 99.00th=[45876], 99.50th=[50594], 99.90th=[63177], 99.
     | 99.99th=[84411]						     | 99.99th=[84411]
Leon Romanovsky March 24, 2019, 9:11 a.m. UTC | #14
On Wed, Mar 20, 2019 at 11:34:27AM -0700, Sagi Grimberg wrote:
>
> > It doesn't sound reasonable that we have per-ULP feature (DIM is
> > per-ULP) to be configured globally. Especially given the fact that
> > users can find themselves running different workloads with different
> > requirements on the same system.
> >
> > Currently each ULP has some sort of tool to configure itself and I think
> > that once ULP is converted to use DIM, it should have on/off knob in the
> > tool used by their users.
>
> Its not any different than socket options that can have a global sysctl
> knob that can be overridden by individual socket consumers.

Right, but there is a major difference between socket example and
your proposal.

Combination of socket option with general knob gives you maximum
versatility in order to disable/enable/configure through program
or through some sensible default. In your proposal, you will limit
yourself to some system wide default, without any ability to override
it specifically for your load.

Thanks