[Beowulf] HPL Benchmarking and Optimization

[Ellis Wilson]

I'm currently running some benchmarks on my cluster,
one of special 
interest to me being High Performance Linpack,
implemented for parallel 
installations.  I've gotten a hang of the variables
(Ns, NBs, etc.) from 
information available online, however would be really
interested to hear 
back on whether my results are far under par and if
compilation 
optimizations (or different fortran compilers) will
increase my 
performance vastly.

Currently I get these kind of numbers from tested
computers using the 
same environment (gentoo, fortran in gcc, hpl, all
same compilation 
options):
1 x Core2Duo (2.1ghz/core, 2gigs ram) - 2.3Gflops
1 x Athlon 64 3500+ (2.2ghz, 1gig ram) - 1.0Glops
4 x Core2Duo (2.1ghz/core for a total of 8 cores,
2gigs ram/node, 
100mbit Ethernet interconnect) - 6.7Gflops

I'm finding numbers online that are well below, at,
and well above my 
own.  Are these numbers low and if so, how could I
improve the situation 
(or should I attempt to optimize at all, this seems a
bit backwards 
since it's making the benchmark akin to my
application)?

Thanks in advance,

Ellis




      ____________________________________________________________________________________
You rock. That's why Blockbuster's offering you one month of Blockbuster Total Access, No Cost.  
http://tc.deals.yahoo.com/tc/blockbuster/text5.com