[Bioclusters] altivec-HMMER bechmark results from Erik Lindahl - 6x faster than dual Athlon

Dave Waddell bioclusters@bioinformatics.org
Wed, 11 Dec 2002 11:40:40 -0500

Don't want to start a processor war but the current Intel chip is a
3.06GHz Pentium 4 not a 1GHz Pentium 3. IMHO, benchmark results should
match current hardware or don't make them at all.

-----Original Message-----
From: chris dagdigian [mailto:dag@sonsorol.org]=20
Sent: Wednesday, December 11, 2002 10:56 AM
To: bioclusters@bioinformatics.org; biodevelopers@bioinformatics.org
Subject: [Bioclusters] altivec-HMMER bechmark results from Erik Lindahl
- 6x faster than dual Athlon

Some nice data from Erik Lindahl at Stanford regarding the new apple=20
altivec enhancements that were added to the hmmer codebase:

> Hmmsearch on the dual-1.25 GHz G4 is 6.2 times faster than on a dual
> Hmmpfam is less impressive, mainly because it spends less time in the
> core routines, but as far as I know hmmsearch is the most important
program anyway.
> Sean Eddy has also confirmed that he will be including this altivec
> optimization in the next official version of HMMer, whenever that is
released. In the
> meantime I've put my optimized version at
> http://csb.stanford.edu/lindahl/altivec/
> Since it is licensed under the GPL anybody can use and distribute it.

Benchmark data from Erik is enclosed in PDF form.

BioTeam has built a double-clickable Mac OS X package installer for this

code but I've been a slacker and have yet to put it on our website.=20
Interested parties should email me directly if they want the package=20
version of these binaries.


Chris Dagdigian, <dag@sonsorol.org>
Independent life science IT & informatics consulting
Office: 617-666-6454, Mobile: 617-877-5498, Fax: 425-699-0193
PGP KeyID: 83D4310E Yahoo IM: craffi Web: http://bioteam.net