linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [BENCHMARK] dbench based
@ 2002-10-09 20:34 Paolo Ciarrocchi
  0 siblings, 0 replies; 3+ messages in thread
From: Paolo Ciarrocchi @ 2002-10-09 20:34 UTC (permalink / raw)
  To: linux-kernel

Hi all,
here the results of a simple dbench test.
The "core" of script is the following:

for value in 2 8 16 24 32;
do 
	log=${kern}$1.dbench${value}.$date.log
	echo $log
		> ./log/$log
		for i in `seq 1 1 3`;
		do 
			echo $i ":" $value
			sync;
			sync;
			sync;
			./dbench $value |grep Th >> ./log/$log
		done;
	
	 awk '{tot+=$2}; END {print "Average: " tot/NR " MB/sec"}' ./log/$log >>./log/$log

The scripts evaluates the average of 3 runs of dbench N.
The missing part of the script is just "logging".
If someone is interested I can send or post to lkml the whole script.




--- 2.4.19 ---
Istances Throughput
2 	 46.6689
8 	 25.5343
16 	 20.7133
24 	 16.2473
32 	 14.2351

--- 2.5.34 ---
Istances Throughput
2 	 28.675
8 	 26.7106
16 	 21.0888
24 	 13.9644
32 	 12.6921

--- 2.5.41 ---
Istances Throughput
2 	 31.0127
8 	 32.0934
16 	 29.3058
24 	 20.291
32 	 19.157


Ok, dbench is not the best benchmark but, finally, 2.5.41 is faster then 2.4.19 (from the "dbench" point of view ;-)


HW PIII@800, 256 MiB RAM
FS ext3


Comments ?

Ciao,
		Paolo

-- 
Get your free email from www.linuxmail.org 


Powered by Outblaze

^ permalink raw reply	[flat|nested] 3+ messages in thread

* [BENCHMARK] dbench based
@ 2003-03-05  9:36 Paolo Ciarrocchi
  0 siblings, 0 replies; 3+ messages in thread
From: Paolo Ciarrocchi @ 2003-03-05  9:36 UTC (permalink / raw)
  To: linux-kernel

Hi all,
here are the results of a test based on dbench.
It runs dbench N three times and evaluate the average.

Something happened between 2.5.59 and 2.5.61.

Kernel version: report%2.4.19
2 	 46.6689
8 	 25.5343
16 	 20.7133
24 	 16.2473
32 	 14.2351

Kernel version: report%2.5.34
2 	 28.675
8 	 26.7106
16 	 21.0888
24 	 13.9644
32 	 12.6921

Kernel version: report%2.5.41
2 	 31.0127
8 	 32.0934
16 	 29.3058
24 	 20.291
32 	 19.157

Kernel version: report%2.5.42
2 	 29.8424
8 	 33.2596
16 	 27.0958
24 	 23.7237
32 	 20.4011

Kernel version: report%2.5.44
2 	 66.2467
8 	 32.0691
16 	 28.6565
24 	 20.3913
32 	 17.6516

Kernel version: report%2.5.45
2 	 44.8421
8 	 30.7749
16 	 26.6394
24 	 22.0835
32 	 17.3273

Kernel version: report%2.5.46
2 	 50.5605
8 	 32.2306
16 	 27.0133
24 	 22.4922
32 	 21.5183

Kernel version: report%2.5.53
2 	 50.4847
8 	 50.3427
16 	 36.5902
24 	 40.0419
32 	 35.5747

Kernel version: report%2.5.54bk4
2 	 46.7899
8 	 49.8274
16 	 41.2771
24 	 41.6313
32 	 39.2595

Kernel version: report%2.5.59
2 	 49.3527
8 	 47.5257
16 	 39.307
24 	 41.0944
32 	 35.264

Kernel version: report%2.5.61
2 	 47.9871
8 	 44.9921
16 	 32.549
24 	 27.1713
32 	 24.1933

Kernel version: report%2.5.62
2 	 62.2361
8 	 37.5537
16 	 30.384
24 	 22.9646
32 	 20.2702

Kernel version: report%2.5.63
2 	 61.2223
8 	 42.0958
16 	 30.3052
24 	 24.9342
32 	 19.6176

Kernel version: report%2.5.63-mjb2
2 	 59.8553
8 	 44.3926
16 	 32.0374
24 	 26.5789
32 	 23.7801

Ciao,
        Paolo

-- 
______________________________________________
http://www.linuxmail.org/
Now with e-mail forwarding for only US$5.95/yr

Powered by Outblaze

^ permalink raw reply	[flat|nested] 3+ messages in thread

* [BENCHMARK] dbench based
@ 2002-11-12 11:10 Paolo Ciarrocchi
  0 siblings, 0 replies; 3+ messages in thread
From: Paolo Ciarrocchi @ 2002-11-12 11:10 UTC (permalink / raw)
  To: linux-kernel

I ran a dbech based script.
The "core" of the script is:
for value in 2 8 16 24 32;
do 
	log=${kern}$1.dbench${value}.$date.log
	echo $log
		> ./log/$log
		for i in `seq 1 1 3`;
		do 
			echo $i ":" $value
			sync;
			sync;
			sync;
			./dbench $value |grep Th >> ./log/$log
done;
	
	 awk '{tot+=$2}; END {print "Average: " tot/NR " MB/sec"}' ./log/$log >>./log/$log

It runs dbench N 3 times and then it evaluates the average.


And here the results:
Kernel		N	Average		Baseline	Ratio
2.4.19		2	46.6689		46.6689		1.000
2.5.34		2	28.675		46.6689		0.614
2.5.41		2	31.0127		46.6689		0.665
2.5.42		2	29.8424		46.6689		0.639
2.5.44		2	66.2467		46.6689		1.420
2.5.45		2	44.8421		46.6689		0.961
2.5.46		2	50.5605		46.6689		1.083

2.4.19		8	25.5343		25.5343		1.000
2.5.34		8	26.7106		25.5343		1.046
2.5.41		8	32.0934		25.5343		1.257
2.5.42		8	33.2596		25.5343		1.303
2.5.44		8	32.0691		25.5343		1.256
2.5.45		8	30.7749		25.5343		1.205
2.5.46		8	32.2306		25.5343		1.262

2.4.19		16	20.7133		20.7133		1.000
2.5.34		16	21.0888		20.7133		1.018
2.5.41		16	29.3058		20.7133		1.415
2.5.42		16	27.0958		20.7133		1.308
2.5.44		16	28.6565		20.7133		1.383
2.5.45		16	26.6394		20.7133		1.286
2.5.46		16	27.0133		20.7133		1.304

2.4.19		24	16.2473		16.2473		1.000
2.5.34		24	13.9644		16.2473		0.859
2.5.41		24	20.291		16.2473		1.249
2.5.42		24	23.7237		16.2473		1.460
2.5.44		24	20.3913		16.2473		1.255
2.5.45		24	22.0835		16.2473		1.359
2.5.46		24	22.4922		16.2473		1.384

2.4.19		32	14.2351		14.2351		1.000
2.5.34		32	12.6921		14.2351		0.892
2.5.41		32	19.157		14.2351		1.346
2.5.42		32	20.4011		14.2351		1.433
2.5.44		32	17.6516		14.2351		1.240
2.5.45		32	17.3273		14.2351		1.217
2.5.46		32	21.5183		14.2351		1.512

I hope they are usefull for you.

Ciao,
          Paolo
-- 
______________________________________________
http://www.linuxmail.org/
Now with POP3/IMAP access for only US$19.95/yr

Powered by Outblaze

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2003-03-05  9:26 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2002-10-09 20:34 [BENCHMARK] dbench based Paolo Ciarrocchi
2002-11-12 11:10 Paolo Ciarrocchi
2003-03-05  9:36 Paolo Ciarrocchi

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).