I compiled my brand new 2.2.12 kernel with Andrea Arcangeli's
SMP-scheduler-2_2_11-E patch and I was stunned. The system felt much faster
and much more responsive. XFree comes up much quicker, both CPUs are really
busy at 100% while X is starting. I was impressed. Almost the same
performace gain after moving from 2.0.x to 2.1.x/2.2.x. WOW!!!!!
The patch is about when waking up a process the kernel scheduler first
tries to find an idle CPU to run the process on.
You can get the patch at:
ftp://ftp.suse.com/pub/people/andrea/kernel-patches/my-2.2.12/
SMP-scheduler-2_2_11-E
AFAIK this patch can be used against 2.2.11 and 2.2.12 and speeds up both
kernels on SMP systems although I haven't tried it with a 2.2.11, only with
2.2.12.
The patch was merged into 2.3.15 but not yet in 2.2.x because it's a
performance issue only but I feel it deserves to be integrated into 2.2.x
because it speeds up so much on SMP machines.
So go, test it, enjoy the performace gain and tell the world about it!!!
I had my system up for >10h on the weekend with lots of processes (niced
and regular prio). It was rock stable.
This is my hardware setup:
- Tyan S1563D HX motherboard (~ 2 1/2 years old)
- 2 * Pentium-S @ 133 MHz
- 96 MB Ram
- Adaptec 2940
- 2G + 4G SCSI HD
- PCI Elsa Winner 2000AVI (S3 968) w/ 4MB Ram
- SCSI CD 12x Pioneer
- SCSI 100M Zip
- SCSI Sony SDT7000 DDS2 DAT drive (BTW: an excellent and very fast drive)
- SuSE Linux 6.1 - now with Kernel 2.2.12 and Andrea's SMP_schedule and
printk patch (see P.S. below).
I know, it's not a killer machine (note: it's old, 2 1/2 years) but fast
enogh for me.
Punchline: I will never again compile a 2.2.x kernel without this
performace booster.
P.S.:
Don't forget
ftp://ftp.suse.com/pub/people/andrea/kernel-patches/my-2.2.12/printk-races-
2.2.11-D
or you may have a klogd with 100% CPU again.
Mit freundlichen Gruessen / best regards
Michael Kwasigroch
FaxPlus/Open Development
________________________________________
eMail: mkwasigr@intercope.com
INTERCOPE GmbH, Hamburg, Bebelallee 72
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/