lkml.org 
[lkml]   [2015]   [Mar]   [2]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH v2 00/15] x86, alternatives: Instruction padding and more robust JMPs

Hi Borislav,

At Thu, 26 Feb 2015 19:13:38 +0100,
Borislav Petkov wrote:
>
> Hi all,
>
> So this alternatives patchset breaks perf bench mem, here are a couple
> of patches ontop, you guys tell me whether it makes sense. I wanted to
> make it run all memset/memcpy routines so here are a couple of patches
> which do this:
>
> ./perf bench mem memset -l 20MB -r all
> # Running 'mem/memset' benchmark:
> Routine default (Default memset() provided by glibc)
> # Copying 20MB Bytes ...
>
> 1.136000 GB/Sec
> 6.026304 GB/Sec (with prefault)
> Routine x86-64-unrolled (unrolled memset() in arch/x86/lib/memset_64.S)
> # Copying 20MB Bytes ...
>
> 5.333493 GB/Sec
> 5.633473 GB/Sec (with prefault)
> Routine x86-64-stosq (movsq-based memset() in arch/x86/lib/memset_64.S)
> # Copying 20MB Bytes ...
>
> 5.828484 GB/Sec
> 5.851183 GB/Sec (with prefault)
> Routine x86-64-stosb (movsb-based memset() in arch/x86/lib/memset_64.S)
> # Copying 20MB Bytes ...
>
> 5.553384 GB/Sec
> 5.956465 GB/Sec (with prefault)
>
> This way you can see all results by executing one command only with "-r
> all".
>
> Patches coming as a reply to this message.

I'm not sure I'm a suitable person for reviewing your patch, but I
tested this patchset for perf bench with your latest (v2) patchset for
x86 alternatives. It looks good to me.
Reviewed-by: Hitoshi Mitake <mitake.hitoshi@lab.ntt.co.jp>

Thanks,
Hitoshi


\
 
 \ /
  Last update: 2015-03-02 16:01    [W:0.326 / U:1.096 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site