Re: [PATCH v3] perf stat: Show percore counts in per CPU output

From: Ravi Bangoria
Date: Thu Feb 13 2020 - 08:21:49 EST


Hi Jin,

On 2/13/20 12:45 PM, Jin Yao wrote:
With this patch, for example,

# perf stat -e cpu/event=cpu-cycles,percore/ -a -A --percore-show-thread -- sleep 1

Performance counter stats for 'system wide':

CPU0 2,453,061 cpu/event=cpu-cycles,percore/
CPU1 1,823,921 cpu/event=cpu-cycles,percore/
CPU2 1,383,166 cpu/event=cpu-cycles,percore/
CPU3 1,102,652 cpu/event=cpu-cycles,percore/
CPU4 2,453,061 cpu/event=cpu-cycles,percore/
CPU5 1,823,921 cpu/event=cpu-cycles,percore/
CPU6 1,383,166 cpu/event=cpu-cycles,percore/
CPU7 1,102,652 cpu/event=cpu-cycles,percore/

We can see counts are duplicated in CPU pairs
(CPU0/CPU4, CPU1/CPU5, CPU2/CPU6, CPU3/CPU7).


I was trying this patch and I am getting bit weird results when any cpu
is offline. Ex,

$ lscpu | grep list
On-line CPU(s) list: 0-4,6,7
Off-line CPU(s) list: 5

$ sudo ./perf stat -e cpu/event=cpu-cycles,percore/ -a -A --percore-show-thread -vv -- sleep 1
...
cpu/event=cpu-cycles,percore/: 0: 23746491 1001189836 1001189836
cpu/event=cpu-cycles,percore/: 1: 19802666 1001291299 1001291299
cpu/event=cpu-cycles,percore/: 2: 24211983 1001394318 1001394318
cpu/event=cpu-cycles,percore/: 3: 54051396 1001516816 1001516816
cpu/event=cpu-cycles,percore/: 4: 6378825 1001064048 1001064048
cpu/event=cpu-cycles,percore/: 5: 21299840 1001166297 1001166297
cpu/event=cpu-cycles,percore/: 6: 13075410 1001274535 1001274535
Performance counter stats for 'system wide':
CPU0 30,125,316 cpu/event=cpu-cycles,percore/
CPU1 19,802,666 cpu/event=cpu-cycles,percore/
CPU2 45,511,823 cpu/event=cpu-cycles,percore/
CPU3 67,126,806 cpu/event=cpu-cycles,percore/
CPU4 30,125,316 cpu/event=cpu-cycles,percore/
CPU7 67,126,806 cpu/event=cpu-cycles,percore/
CPU0 30,125,316 cpu/event=cpu-cycles,percore/
1.001918764 seconds time elapsed

I see proper result without --percore-show-thread:

$ sudo ./perf stat -e cpu/event=cpu-cycles,percore/ -a -A -vv -- sleep 1
...
cpu/event=cpu-cycles,percore/: 0: 11676414 1001190709 1001190709
cpu/event=cpu-cycles,percore/: 1: 39119617 1001291459 1001291459
cpu/event=cpu-cycles,percore/: 2: 41821512 1001391158 1001391158
cpu/event=cpu-cycles,percore/: 3: 46853730 1001492799 1001492799
cpu/event=cpu-cycles,percore/: 4: 14448274 1001095948 1001095948
cpu/event=cpu-cycles,percore/: 5: 42238217 1001191187 1001191187
cpu/event=cpu-cycles,percore/: 6: 33129641 1001292072 1001292072
Performance counter stats for 'system wide':
S0-D0-C0 26,124,688 cpu/event=cpu-cycles,percore/
S0-D0-C1 39,119,617 cpu/event=cpu-cycles,percore/
S0-D0-C2 84,059,729 cpu/event=cpu-cycles,percore/
S0-D0-C3 79,983,371 cpu/event=cpu-cycles,percore/
1.001961563 seconds time elapsed

[...]

+--percore-show-thread::
+The event modifier "percore" has supported to sum up the event counts
+for all hardware threads in a core and show the counts per core.
+
+This option with event modifier "percore" enabled also sums up the event
+counts for all hardware threads in a core but show the sum counts per
+hardware thread. This is essentially a replacement for the any bit and
+convenient for posting process.

s/posting process/post processing/ ? :)

Ravi