All of lore.kernel.org
 help / color / mirror / Atom feed
* [BENCHMARK] reaim 2x,4x,8x with various SMP balancing patches
@ 2003-09-14  9:44 Con Kolivas
  0 siblings, 0 replies; only message in thread
From: Con Kolivas @ 2003-09-14  9:44 UTC (permalink / raw)
  To: linux kernel mailing list; +Cc: Andrew Morton

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

With the help of the OSDL hardware I've thrown a bootload of reaim benchmarks 
at these various patches because of the confusion regarding their usefulness. 
2.6.0-test5 is the baseline, and the 3 unique patches CMT,A3,BT were added by 
themselves and all together to 2.6.0-test5 for comparison. Finally the full 
2.6.0-test5-mm1 patch was compared. The A3 patch was modified slightly to 
overcome the one minor magnitude error for fairness. Looking at the url for 
each benchmark and examining the "Graph - Jobs per minute" is useful to 
assess where the performance hovers. 


Legend

260t5 - vanilla 2.6.0-test5
CMT - sched-CAN_MIGRATE_TASK-fix.patch
A3 - sched-2.6.0-test2-mm2-A3.patch
BT - sched-balance-tuning.patch
all - CMT + A3 + BT
mm1 - 2.6.0-test5-mm1


2 CPU:

Kernel	Throughput	URL
260t5	1337.11		http://khack.osdl.org/stp/279474/
CMT	1321.92		http://khack.osdl.org/stp/279796/
A3	1304.82		http://khack.osdl.org/stp/279799
BT	1326.27		http://khack.osdl.org/stp/279802
All	1337.79		http://khack.osdl.org/stp/279805/
mm1	1316.95		http://khack.osdl.org/stp/279588/


4 CPU:

Kernel	Throughput	URL
260t5	5406.68		http://khack.osdl.org/stp/279883
CMT
A3	5099.14		http://khack.osdl.org/stp/279800
BT	5721.79		http://khack.osdl.org/stp/279803
All	4919.36		http://khack.osdl.org/stp/279806
mm1	5360.32		http://khack.osdl.org/stp/279887


8 CPU:

Kernel	Throughput	URL
260t5	8812.21		http://khack.osdl.org/stp/279448/
CMT	8794.14		http://khack.osdl.org/stp/279798
A3	7084.15		http://khack.osdl.org/stp/279801
BT	8615.13		http://khack.osdl.org/stp/279804
All	7629.14		http://khack.osdl.org/stp/279919
mm1	8478.10		http://khack.osdl.org/stp/279562/


4xCMT seemed to not successfully finish.

In summary, it appears that all the patches cause detriment to performance, 
except the combination of BT with 4 cpus. The A3 patch is particularly 
detrimental, but it is reassuring to see that the extra patches in mm1 (which 
includes O1int patches) recover a lot of that performance, even with the 
other balancing patches.

Con
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQE/ZDhtZUg7+tp6mRURAmJAAJ4zMYHg7iSuVHi7SpiS5hkmoaQl9wCggrcB
SSH6dbJthgABVnEIn4K/z3M=
=Jz7G
-----END PGP SIGNATURE-----


^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2003-09-14  9:36 UTC | newest]

Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-09-14  9:44 [BENCHMARK] reaim 2x,4x,8x with various SMP balancing patches Con Kolivas

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.