>Did you really mean a 30x speedup?? On Thu, 18 Sep 2003, Lucas Carey wrote: > I've built it with icc, but I no longer have the .mk file. > I seem to remember ~30 speedup for the more computationally intensive blast This is supposed to be ~30%. If I'd gotten a 30x speedup there'd be no need for parallel blast. as for blastn, these are STREAMS measurements I did on a dual P4-2.0Ghz with icc6. No clue as to what version of gcc I'd have been using. icc -O3 -xW ------------------------------------------------------------- Function Rate (MB/s) RMS time Min time Max time Copy: 1452.4363 0.0146 0.0110 0.0264 Scale: 1446.1244 0.0225 0.0111 0.0601 Add: 1647.6759 0.0235 0.0146 0.0529 Triad: 1624.4763 0.0242 0.0148 0.0456 gcc -O3 ------------------------------------------------------------- Function Rate (MB/s) RMS time Min time Max time Copy: 1433.1806 0.0112 0.0112 0.0114 Scale: 555.7492 0.0288 0.0288 0.0288 Add: 1648.0107 0.0146 0.0146 0.0146 Triad: 691.8828 0.0347 0.0347 0.0347 gcc ------------------------------------------------------------- Function Rate (MB/s) RMS time Min time Max time Copy: 943.6746 0.0170 0.0170 0.0173 Scale: 931.7509 0.0173 0.0172 0.0173 Add: 1278.3637 0.0188 0.0188 0.0188 Triad: 1269.7732 0.0189 0.0189 0.0189 -Lucas