[v17,11/15] tools/damon/wss: Implement '--thres' option

Commit Message

SeongJae Park July 6, 2020, 11:53 a.m. UTC

From: SeongJae Park <[e-mail]>

Currently, 'wss' treats every region as working set if accessed at least
once.  Someone who want to know performance important working set only
would want to ignore regions having low access frequency.  '--thres'
option can be used to set the minimal access frequency of the regions to
be classified as the workingset.

Using this, users can plot the reuse histogram.  For example:

    $ damo record $(pidof raytrace)
    $ for t in {1..20}; do ./tools/damon/damo report wss --thres $t | \
    		grep avr | awk -v reuse_time=$(( (21 - t) * 5 )) \
    		'{print reuse_time " " $3}'; done
    100 12838416
    95 2222623
    90 1585480
    85 422890
    80 67040
    75 45218
    70 13242
    65 12896
    60 12136
    55 10872
    50 9648
    45 8136
    40 7052
    35 6304
    30 5736
    25 5404
    20 5305
    15 5187
    10 5069
    5 4873

Above command shows the reuse histogram of parsec3.raytrace.  Remind
that the sampling interval and aggregation interval are 5ms and 100ms
by default and this command used the default values.  So, the above
output means that about 12MB of memory region was reused within 100 ms
while only about 2MB of memory region was reused within 95 ms, then
about 1.5MB within 90ms, and so on, in average.

Signed-off-by: SeongJae Park <[e-mail]>
---
 tools/damon/wss.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Patch

diff --git a/tools/damon/wss.py b/tools/damon/wss.py
index b43065176cfd..d2a1b149e3ea 100644
--- a/tools/damon/wss.py
+++ b/tools/damon/wss.py
@@ -17,6 +17,8 @@  def set_argparser(parser):
     parser.add_argument('--range', '-r', type=int, nargs=3,
             metavar=('<start>', '<stop>', '<step>'),
             help='range of wss percentiles to print')
+    parser.add_argument('--thres', '-t', type=int, metavar='<# accesses>',
+            help='minimal number of accesses for treated as working set')
     parser.add_argument('--sortby', '-s', choices=['time', 'size'],
             help='the metric to be used for the sort of the working set sizes')
     parser.add_argument('--plot', '-p', type=str, metavar='<file>',
@@ -67,7 +69,7 @@  def main(args=None):
             wss = 0
             for p in snapshot:
                 # Ignore regions not accessed
-                if p[1] <= 0:
+                if p[1] < args.thres:
                     continue
                 wss += p[0]
             wss_dist.append(wss)