Re: Resolution to UDMA performance issues on 2.0.36

M.Brands (shrike@il.fontys.nl)
Wed, 16 Jun 1999 00:38:10 +0200


On Mon, Jun 14, 1999 at 06:49:04PM -0400, Victor Orlikowski allegedly wrote:
> <idiot mode>
> Victor's Law:
> *Never* run cpu-intensive tasks whilst benchmarking IDE drives.
> </idiot mode>
>
> cpu-intensive tasks == rc5des client.
> :)
> It swallowed *HALF* the UDMA drive performance marks on hdparm -Tt for my
> UDMA drive.
>
> Now in normal figures (I thought I wasn't crazy). Sheesh. Must remember I
> have that thing running there......
>
> Victor

I disagree. Have a look:

[root@bucephalus /root]# hdparm -tT /dev/md0

/dev/md0:
Timing buffer-cache reads: 128 MB in 0.96 seconds =133.33 MB/sec
Timing buffered disk reads: 64 MB in 2.97 seconds =21.55 MB/sec
[root@bucephalus /root]# killall rc5des
[root@bucephalus /root]# hdparm -tT /dev/md0

/dev/md0:
Timing buffer-cache reads: 128 MB in 0.93 seconds =137.63 MB/sec
Timing buffered disk reads: 64 MB in 2.95 seconds =21.69 MB/sec

Yes, the results without rc5des are a bit higher, but not that much.

I'm running two Maxtor DiamondMax 8.4 GB disk in RAID0 setup, each
one on it's own IDE controller. Doing a 'dd if=/dev/md0 of=/dev/null
bs=4096' gives an average transferrate of about 21 MB/sec with 26%
CPU use.

Btw. This box is sitting beside my desk, has lots of memory (>256M)
and is doing absolutely nothing (apart from running rc5des).

My $.02,

Mathijs

-
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/