lkml.org 
[lkml]   [2009]   [May]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH tip 0/5] perf: make top share sym routines with report

* Arnaldo Carvalho de Melo <acme@redhat.com> wrote:

> Haven't done and perf stat runs tho. Perhaps ask for running it
> for 5 refreshes and then exit.

I've benchmarked it. With about half a million sample IRQs per
second on a 16way x86 box the overhead went from 22.3%:

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME COMMAND
4701 mingo 20 0 423m 61m 1796 R 22.3 0.5 0:08.11 perf

down to:

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME COMMAND
3386 mingo 20 0 88068 61m 1756 R 18.2 0.5 0:03.06 perf

So it's a ~22% speedup.

Also, there's much less virtual memory used as well - a factor of 5
reduction (!).

Applied and pushed out - thanks Arnaldo!

Ingo


\
 
 \ /
  Last update: 2009-05-29 00:03    [W:0.034 / U:0.604 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site