performance of 2.4.13pre4 with Jens blockhighmem

DICKENS,CARY (cary_dickens2@hp.com)
Thu, 25 Oct 2001 17:22:57 -0400


Here is some more performance data.

I had some problems with the megaraid portion of the patch and removed that
part of it since most of the storage was connected via the qlogicfc card.
What I measured was a maximum throughput of 125% that of the 2.4.13pre4
without the patch. The run was also much more stable. (No dropped packets
because of response times.)

The formula used is:

Max throughput of the 2.4.13pre4 with patch
--------------------------------------------------------------- * 100
Max throughput of the' 2.4.13pre4 w/o patch

The problem that I see is that 2.4.13preX were all really slow. I'm seeing
less than 50% of the throughput that was seen in 2.4.5pre1 and the response
times increased by 50% in the 2.4.13pre2 time frame. Where we were getting
1 ms response times, we now see 1.5 to 2. This multiplicative factor grows
as the test becomes more aggressive.

The benchmark is SPEC SFS NFS benchmark testing. The filesystem is reiserfs.

I hope this information helps.

Cary
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/