linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [BENCHMARK] 2.5.38-mm3 with contest 0.37
@ 2002-09-26 12:33 Con Kolivas
  2002-09-27 14:41 ` Jens Axboe
  0 siblings, 1 reply; 2+ messages in thread
From: Con Kolivas @ 2002-09-26 12:33 UTC (permalink / raw)
  To: linux-kernel; +Cc: Andrew Morton



Here are the contest results (http://contest.kolivas.net) for 2.5.38-mm3:

noload:
Kernel                  Time            CPU             Ratio
2.4.19                  72.21           99%             1.00
2.5.38                  73.56           99%             1.02
2.5.38-mm2              73.88           99%             1.02
2.5.38-mm3              74.10           99%             1.03

process_load:
Kernel                  Time            CPU             Ratio
2.4.19                  88.02           80%             1.22
2.5.38                  77.44           94%             1.07
2.5.38-mm2              77.88           94%             1.08
2.5.38-mm3              77.94           94%             1.08

io_load:
Kernel                  Time            CPU             Ratio
2.4.19                  170.06          45%             2.36
2.5.38                  283.27          28%             3.92
2.5.38-mm2              106             75%             1.47*
2.5.38-mm3              95.9            84%             1.33*

mem_load:
Kernel                  Time            CPU             Ratio
2.4.19                  100.52          78%             1.39
2.5.38                  116.94          67%             1.62
2.5.38-mm2              105             74%             1.45  
2.5.38-mm3              109             73%             1.50

IO load and mem_load results are averages of n=5 runs. The difference between
io_loads of mm2 and mm3 is statistically significant (p=0.022) but not mem_load.

Con.


P.S. For those who just want more info :-P

Hardware: 1133Mhz PIII, 256Mb Ram, IDE ATA5 5400rpm HD, IO load on same disk,
reiserFS


Statistics assuming normal distribution curve:

mm2:
Hi = 126. Low = 92.1
Mean = 106.
95% confidence interval for Mean: 87.52 thru 124.6
Standard Deviation = 14.9
Median = 103.
Average Absolute Deviation from Median = 11.5

mm3:
Hi = 107. Low = 82.1
Mean = 95.9
95% confidence interval for Mean: 80.51 thru 111.4
Standard Deviation = 12.4
Median = 101.
Average Absolute Deviation from Median = 9.64

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [BENCHMARK] 2.5.38-mm3 with contest 0.37
  2002-09-26 12:33 [BENCHMARK] 2.5.38-mm3 with contest 0.37 Con Kolivas
@ 2002-09-27 14:41 ` Jens Axboe
  0 siblings, 0 replies; 2+ messages in thread
From: Jens Axboe @ 2002-09-27 14:41 UTC (permalink / raw)
  To: Con Kolivas; +Cc: linux-kernel, Andrew Morton

On Thu, Sep 26 2002, Con Kolivas wrote:
> io_load:
> Kernel                  Time            CPU             Ratio
> 2.4.19                  170.06          45%             2.36
> 2.5.38                  283.27          28%             3.92
> 2.5.38-mm2              106             75%             1.47*
> 2.5.38-mm3              95.9            84%             1.33*

this is encouraging to see, 2.5.38-mm3 has the new deadline io scheduler
and it appears to be performing well here too. I'm especially looking
forward to 2.5.38 vs 2.5.39 too, as 2.5.39 will have some deadline
updates that are not in -mm3.

-- 
Jens Axboe


^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2002-09-27 14:35 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2002-09-26 12:33 [BENCHMARK] 2.5.38-mm3 with contest 0.37 Con Kolivas
2002-09-27 14:41 ` Jens Axboe

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).