lkml.org 
[lkml]   [2013]   [Aug]   [8]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [PATCH V3]hrtimer: Fix a performance regression by disable reprogramming in remove_hrtimer
From
Date
On Thu, 2013-08-08 at 17:04 +0800, ethan.zhao wrote:

> So, it's Peter's patch working………

Yes. My patch killed the nohz tick start/stop overhead prior to 3.8,
but in 3.8, the menu governor changes added more high frequency timer
banging that negated its effect. In 3.9, mwait went away, making core2
boxen very sad, and in 3.10 all of x86 got generic idle, making my boxen
even sadder.

select_idle_sibling() (the thing that makes us schedule pipe-test and
many other things cross core if there's an idle core available) has
always been two-faced, doing both good and evil.. but with recent work,
it's leaning more toward evil than ever before.

-Mike

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

\
 
 \ /
  Last update: 2013-08-08 14:41    [W:0.128 / U:0.024 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site