Jean-Michel Merliot <[EMAIL PROTECTED]> writes:
> Hi everybody,
> Still related to my MB problem ( I changed my MB and
> Seti@home now computes a WU in 24 hours instead of 15, with same
> CPUs and same memory..., and of course same kernel ).
> When I look at /proc/interrupts I get that :
>
> CPU0 CPU1
> 0: 17866148 24644515 IO-APIC-edge timer
> 1: 5640 7257 IO-APIC-edge keyboard
> 2: 0 0 XT-PIC cascade
> 8: 1 0 IO-APIC-edge rtc
> 12: 152828 193402 IO-APIC-edge PS/2 Mouse
> 13: 1 0 XT-PIC fpu
> 14: 379862 550184 IO-APIC-edge ide0
> 15: 5 1 IO-APIC-edge ide1
> 17: 33101 38691 IO-APIC-level PCnet/PCI II 79C970A
> NMI: 0
> ERR: 0
>
> Apparently, one CPU proceeds much more interrupts than the other.
>
> Is that normal, I've always seen the numbers for both CPU roughly
> the same ?
mine doesn't always distribute interupts evenly either.
euler(jk)$ cat /proc/interrupts
CPU0 CPU1 CPU2 CPU3
0: 10934034 18769573 9977544 9847089 IO-APIC-edge timer
1: 189 208 168 167 IO-APIC-edge keyboard
2: 0 0 0 0 XT-PIC cascade
8: 1 0 0 0 IO-APIC-edge rtc
10: 3511 3608 3483 3545 IO-APIC-level eth0
11: 64243 78855 63136 62458 IO-APIC-level sym53c8xx, sym53c8xx
12: 19 19 19 21 IO-APIC-level PS/2 Mouse
13: 1 0 0 0 XT-PIC fpu
NMI: 0
ERR: 0
they are usually roughly equal and yours do not seem out of bounds.
> I wonder if that could explain the drop in Seti@home computing
> performance... ?
i don't know what is wrong.
could you try running just one setiathome with the new setup? maybe
the whole system is slow or maybe the smp is slow specifically.
here's a handy little script to estimate the time. (this assumes you
run setiathome in ~/Tmp/0 ~/Tmp/1 &c. change this to suit your
configuration.) this will let you see the performance a little more
quickly. let the setiathome run for at least an hour to let the
statistics settle.
---> setitime <---
#!/bin/bash
for d in 0 1 2 3; do
sed 's/=/ /' ~/Tmp/$d/state.txt | gawk '
/^cpu / {
cpusec = $2;
cpuhour = cpusec / 3600;
}
/^prog / {
prog = $2;
}
END {
printf "%6.2f hr\t%6.2f %% done\t %6.2f hr/block\n",
cpuhour, 100*prog, cpuhour / prog;
}'
done
------------------
--
J o h a n K u l l s t a m
[[EMAIL PROTECTED]]
Don't Fear the Penguin!
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/dmentre/smp-howto/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
