lkml.org 
[lkml]   [2009]   [Jul]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [PATCH 1/2 -tip] perf_counter: Add generalized hardware vectored co-processor support for AMD and Intel Corei7/Nehalem
From
Date
Hello Ingo,

On Thu, 2009-07-02 at 15:14 +0530, Jaswinder Singh Rajput wrote:
> This output is from AMD box:
>
> $ ./perf stat -e add -e multiply -e divide -e vec-idle-cycles -e vec-stall-cycles -e vec-ops -- ls -lR /usr/include/ > /dev/null
>
> Performance counter stats for 'ls -lR /usr/include/':
>
> 4218 vec-adds (scaled from 66.60%)
> 7426 vec-muls (scaled from 66.67%)
> 5441 vec-divs (scaled from 66.29%)
> 821982187 vec-idle-cycles (scaled from 66.45%)
> 2681 vec-stall-cycles (scaled from 67.11%)
> 7887 vec-ops (scaled from 66.88%)
>
> 0.417614573 seconds time elapsed
>
> $ ./perf stat -e add -e multiply -e divide -e vec-idle-cycles -e vec-stall-cycles -e vec-ops -- /usr/bin/rhythmbox ~jaswinder/Music/singhiskinng.mp3
>
> Performance counter stats for '/usr/bin/rhythmbox /home/jaswinder/Music/singhiskinng.mp3':
>
> 17552264 vec-adds (scaled from 66.28%)
> 19715258 vec-muls (scaled from 66.63%)
> 15862733 vec-divs (scaled from 66.82%)
> 23735187095 vec-idle-cycles (scaled from 66.89%)
> 11353159 vec-stall-cycles (scaled from 66.90%)
> 36628571 vec-ops (scaled from 66.48%)
>
> 298.350012843 seconds time elapsed
>
> $ ./perf stat -e add -e multiply -e divide -e vec-idle-cycles -e vec-stall-cycles -e vec-ops -- /usr/bin/vlc ~jaswinder/Videos/Linus_Torvalds_interview_with_Charlie_Rose_Part_1.flv
>
> Performance counter stats for '/usr/bin/vlc /home/jaswinder/Videos/Linus_Torvalds_interview_with_Charlie_Rose_Part_1.flv':
>
> 20177177044 vec-adds (scaled from 66.63%)
> 34101687027 vec-muls (scaled from 66.64%)
> 3984060862 vec-divs (scaled from 66.71%)
> 26349684710 vec-idle-cycles (scaled from 66.65%)
> 9052001905 vec-stall-cycles (scaled from 66.66%)
> 76440734242 vec-ops (scaled from 66.71%)
>
> 272.523058097 seconds time elapsed
>
> $ ./perf list shows vector events like :
>
> vec-adds OR add [Hardware vector event]
> vec-muls OR multiply [Hardware vector event]
> vec-divs OR divide [Hardware vector event]
> vec-idle-cycles OR vec-empty-cycles [Hardware vector event]
> vec-stall-cycles OR vec-busy-cycles [Hardware vector event]
> vec-ops OR vec-operations [Hardware vector event]
>
> Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com>
> ---
> arch/x86/kernel/cpu/perf_counter.c | 45 +++++++++++++++++++++++++++++
> include/linux/perf_counter.h | 15 ++++++++++
> kernel/perf_counter.c | 1 +
> tools/perf/util/parse-events.c | 55 ++++++++++++++++++++++++++++++++++++
> 4 files changed, 116 insertions(+), 0 deletions(-)
>
> diff --git a/arch/x86/kernel/cpu/perf_counter.c b/arch/x86/kernel/cpu/perf_counter.c
> index 36c3dc7..48f28b7 100644
> --- a/arch/x86/kernel/cpu/perf_counter.c
> +++ b/arch/x86/kernel/cpu/perf_counter.c
> @@ -372,6 +372,22 @@ static const u64 atom_hw_cache_event_ids
> },
> };
>
> +/*
> + * Generalized hw vectored co-processor event table
> + */
> +
> +static u64 __read_mostly hw_vector_event_ids[PERF_COUNT_HW_VECTOR_MAX];
> +
> +static const u64 nehalem_hw_vector_event_ids[] =
> +{
> + [PERF_COUNT_HW_VECTOR_ADD] = 0x01B1, /* UOPS_EXECUTED.PORT0 */
> + [PERF_COUNT_HW_VECTOR_MULTIPLY] = 0x0214, /* ARITH.MUL */
> + [PERF_COUNT_HW_VECTOR_DIVIDE] = 0x0114, /* ARITH.CYCLES_DIV_BUSY */
> + [PERF_COUNT_HW_VECTOR_IDLE_CYCLES] = 0x0,
> + [PERF_COUNT_HW_VECTOR_STALL_CYCLES] = 0x60A2, /* RESOURCE_STALLS.FPCW|MXCSR*/
> + [PERF_COUNT_HW_VECTOR_OPS] = 0x0710, /* FP_COMP_OPS_EXE.X87|MMX|SSE_FP*/
> +};
> +

Have you tested this patch on Intel Corei7/Nehalem.

Thanks,
--
JSR

http://userweb.kernel.org/~jaswinder/





\
 
 \ /
  Last update: 2009-07-03 09:41    [W:0.973 / U:0.028 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site