On Wed, 2008-02-27 at 23:30 -0800, Bill Broadley wrote: > > I don't see any particular reason why memory bandwidth can go through a full > doublings in the near future if there was a market for it, last I checked > nvidia was doing pretty well ;-) > > [1] Sorry to use marketing bandwidth, I've not seen stream numbers for CUDA > yet. I hope to work on one though. If anyone has numbers please speak > up.
Some numbers here: http://forums.nvidia.com/index.php?showtopic=52686 Running this on my 8600 card I get: STREAM Benchmark implementation in CUDA Array size (single precision)=2000000 using 128 threads per block, 15625 blocks Function Rate (MB/s) Avg time Min time Max time Copy: 291777.6696 0.0001 0.0001 0.0001 Scale: 291777.6696 0.0001 0.0001 0.0001 Add: 437666.5043 0.0001 0.0001 0.0001 Triad: 437666.5043 0.0001 0.0001 0.0001 All the 0.0001 don't look right to me, but its 8:30 here and I should get on with the day job. _______________________________________________ Beowulf mailing list, Beowulf@beowulf.org To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
