Re: BLAS recommendations SUMMARY
"Robert G. Brown" wrote:
> To close with a mildly political statement, I'd recommend that in cases
> where folks have energy to put into LAPACK/BLAS for linux and
> beowulfery, they should focus on open source solutions (GPL or not) as
> ATLAS appears to be and avoid the "easy" route of using binary libraries
> with uncertain licensing. If just a few folks who are interested join
> Emil in his efforts to optimize for the common processor families it is
> likely that the libraries will very rapidly become extremely good, while
> in the case of closed/proprietary libraries (however nice they may be)
> you are always at the mercy of the owners with respect to bugs and
> performance both. Remember, a beowulf is a COTS, OPEN SOURCE based
> cluster supercomputer, at least to the extent possible.
I'd add to your list then the Alpha assembler BLAS routines from Kazushige
Goto. On a 600 MHz LX164 with 2MB L3 cache (which can be had for under $2000,
and uses standard ECC DRAM max 1GB), dgemm gets 815-860 MFlops! I think this
beats anything else in price/performance pretty easily. And it's GPL.
ftp://www.netstat.ne.jp/pub/Linux/Linux-Alpha-JP/BLAS
<fantasy>I'd love to see this running in a parallel app on the latest four-Alpha
system from AlphaData... Won't give the advertised 5.33 GFlops (I suspect
nothing will), but will come pretty close! http://www.alphadata.co.uk/</fantasy>
This is the single-processor result, more coding would be required to
efficiently parallelize it. I'm interested in adding it to the Debian blas1
package (so it would use these routines only on Alpha, and build the Fortran
sources for other platforms), but am awaiting a reply from that package's
maintainer before starting.
Zeen,
Adam Powell http://lyre.mit.edu/~powell/
Thomas B. King Assistant Professor of Materials Engineering
77 Massachusetts Ave. Rm. 4-117 Phone (617) 452-2086
Cambridge, MA 02139 USA Fax (617) 253-5418
Reply to: