Re: [PATCH] better align percpu counter (Was Re: [tip:sched/core]sched: cpuacct: Use bigger percpu counter batch values for stats counters

[KAMEZAWA Hiroyuki]

On Thu, 20 Aug 2009 12:04:03 +0200
Ingo Molnar <mingo@xxxxxxx> wrote:
> * KAMEZAWA Hiroyuki <kamezawa.hiroyu@xxxxxxxxxxxxxx> wrote:
> > with your program
> > before patch.
> > cpuacct off : 414000-416000 ctsw per sec.
> > cpuacct on  : 401000-404000 ctsw per sec.
> > 
> > after patch
> > cpuacct on  : 412000-413000 ctsw per sec.
> > 
> > Maybe I should check cache-miss late ;)
> 
> Btw., in latest upstream you can do that via:
> 
>   cd tools/perf/
>   make -j install
> 
>   perf stat --repeat 5 -- taskset -c 1 ./context_switch
> 

tried. (on 8cpu/2socket host). It seems cache-miss decreases.
But IPC ..?
==
/root/bin/perf stat --repeat 5 -a -e cycles,instructions,cache-misses,L1-dcache-load-misses,L1-dcache-store-misses -- ./ctxt_sw.sh
[Before] patch
 Performance counter stats for './ctxt_sw.sh' (5 runs):

  1511260148530  cycles                     ( +-   0.025% )  (scaled from 63.49%             )
   470690642181  instructions             #      0.311 IPC     ( +-   0.098% )(scaled from 79.49%)
     1210051728  cache-misses               ( +-   0.629% )  (scaled from 79.00%             )
     3202978828  L1-dcache-load-misses      ( +-   1.118% )  (scaled from 78.00%             )
     1803963907  L1-dcache-store-misses     ( +-   0.728% )  (scaled from 42.99%             )

   60.161941918  seconds time elapsed   ( +-   0.029% )

[After] patch
 Performance counter stats for './ctxt_sw.sh' (5 runs):

  1511961867506  cycles                     ( +-   0.018% )  (scaled from 71.50%)
   448724406149  instructions             #      0.297 IPC     ( +-   0.133% )  (scaled from 75.49%)
     1184548041  cache-misses               ( +-   0.136% )  (scaled from 75.50%)
     3086357048  L1-dcache-load-misses      ( +-   0.822% )  (scaled from 77.50%)
     1708375493  L1-dcache-store-misses     ( +-   0.328% )  (scaled from 47.00%)

   60.179814774  seconds time elapsed   ( +-   0.052% )

Thanks,
-Kame

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/