>>>>> "A" == Alan Cox <alan@lxorguk.ukuu.org.uk> writes:
>> tasks. IMHO with the reliable active_mm information we have in 2.3.x (so
>> being sure to do the right thing) using a +5 should not hurt (maybe we
>> could remove 4 points from the per-processor penality?).
A> Search the kernel archives for late 2.1.x about jumpy SMP behaviour. People
A> traced problems back to the processor favouritism.
A> I also saw no difference on performance when testing multi-cpu load stuff
A> with different values. I think Ingo also did some stuff with this
Any effect will be dependent on the size of the caches. Any effect
would be very small (might be non-existent) on a Celeron (128kB L2
cache), and might be quite noticable on a 2MB L2 Xeon. Cache sizes
inbetween would be expected to have intermediate affect.
Anybody benchmark how much of the L2 cache gets dirtied in a
"typical" (probably read/write) syscall that also goes through
schedule()?
-- Scott Henry <scotth@sgi.com> / Help! My disclaimer is missing! IRIX MTS, / GIGO *really* means: Garbage in, Gospel Out Silicon Graphics, Inc / http://reality.sgi.com/scotth/- 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/
This archive was generated by hypermail 2b29 : Sat Jan 15 2000 - 21:00:16 EST