lkml.org 
[lkml]   [2019]   [Dec]   [10]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [pipe] 3c0edea9b2: lmbench3.PIPE.bandwidth.MB/sec -17.0% regression
On Mon, Dec 09, 2019 at 10:35:35AM -0800, Linus Torvalds wrote:
> On Mon, Dec 9, 2019 at 12:56 AM Michal Hocko <mhocko@kernel.org> wrote:
> >
> > Huh. How can something like that can even can get merged? No changelog,
> > no s-o-b?
>
> Yeah, I missed that when pulling.
>
> I wish that had been the only problem we had with that code, but it wasn't.
>
> Anyway, I'm fairly certain the current git head should fix this
> lmbench3 performance regression too.
>
> Linus

Hi Linus,

Sorry for the inconvenience, indeed, the regression has been fixed.

commit:
cefa80ced5 ("pipe: Increase the writer-wakeup threshold to reduce context-switch count")
3c0edea9b2 ("pipe: Remove sync on wake_ups")
6794862a16 ("Merge tag 'for-5.5-rc1-kconfig-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux")

cefa80ced57a2917 3c0edea9b29f9be6c093f236f7 6794862a16ef41f753abd75c03 testcase/testparams/testbox
---------------- -------------------------- -------------------------- ---------------------------
%stddev change %stddev change %stddev
\ | \ | \
21204 -17% 17604 23% 26123 ± 3% lmbench3/performance-development-100%-PIPE-50%-ucode=0x7000019/lkp-bdw-de1
21204 -17% 17604 23% 26123 GEO-MEAN lmbench3.PIPE.bandwidth.MB/sec

Best Regards,
Rong Chen

\
 
 \ /
  Last update: 2019-12-10 15:15    [W:0.083 / U:0.184 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site