simon at himi.org
Fri Jun 28 13:31:05 EST 2002
On Fri, Jun 28, 2002 at 11:50:51AM +1000, Brett Worth wrote:
> I've just run some very rudimentary performance testing on a server in our
> office and see some poor SMP results. Can anyone suggest why?
> Kernel: 2.4.18
> Disk: Ultra Wide SCSI on AIC7XXX
> CPUs: 2 x 700Mhz PIII
> MB: Intel L440GX+
> RAM: 4 x 64Mb = 256M
> I'm just doing kernel compiles and timing them.
> time (make -jN dep; make -jN clean; make -jN ; make -jN module; ; make -jN install ; make -jN modules_install)
> N=1 264.79s real 251.83s user 20.74s system
> N=2 304 502 46
> N=3 402 660 53
> Is this a memory bandwidth problem? The machine is otherwise idle. I also
> ran the same command (j1) once before starting incase the cache made any
> # hdparm -T -t /dev/sda
> Timing buffer-cache reads: 128 MB in 0.79 seconds =162.03 MB/sec
> Timing buffered disk reads: 64 MB in 4.12 seconds = 15.53 MB/sec
This last one is rather slow - my 7200rpm IDE disk gets 22-23MB/s
off the platters, and I'd expect a good scsi disk to be a good bit
faster . . .
What filesystem are you using with this? ext3 is currently not at
all well optimised for SMP, so if you're using that you may be
seeing the results . . . Try mounting with ext2 instead, and see if
that fixes things.
Other than that, I can't think of anything right now, but you
definitely shouldn't be seeing that kind of slowdown.
PGP public key Id 0x144A991C, or ftp://bg77.anu.edu.au/pub/himi/himi.asc
(crappy) Homepage: http://bg77.anu.edu.au
doe #237 (see http://www.lemuria.org/DeCSS)
My DeCSS mirror: ftp://bg77.anu.edu.au/pub/mirrors/css/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 232 bytes
Desc: not available
Url : http://lists.samba.org/archive/linux/attachments/20020628/00771619/attachment.bin
More information about the linux