String hashing was: Thou shalt have no other gods before the ANSI C standard 1539

Interesting.

I tried the test code from your web page to see how it liked various configurations... the results weren't quite what I expected. Notice that the order of the tests by speed is different in several of these.

Dual Processor Xeon 2.0GHz-512KB, 533Mhz FSB 1GB RAM, Hyperthreading disabled XP Pro SP2

-march=i686 -O3 pinned to a single CPU (CPU 1, not 0).

CRC32 : 6.593s oneAtATimeHash : 9.176s FNVHash : 10.110s BobJenkins : 3.681s SuperFastHash : 2.473s

Same as above, but allowed to bounce across both CPUs

CRC32 : 6.538s oneAtATimeHash : 9.066s FNVHash : 10.055s BobJenkins : 3.681s SuperFastHash : 2.418s

AMD Athlon 64 Mobile 3200+ Notebook, 512MB RAM XP Home (32-bit OS, 64-bit CPU)

CRC32 : 13.187s oneAtATimeHash : 11.648s FNVHash : 8.352s BobJenkins : 6.044s SuperFastHash : 3.516s

Note: this one is really disappointing, despite not being a 64-bit OS, especially for the first two tests, I suspect some sort of mobile power management throttling is getting in the way of them all. I'll need to check into that some more, even though I had it plugged into the AC adapter at the time.

The 64-bit Linux hard drive for this notebook is currently in need of a reinstall, but I want to try that soon to see what happens there. I don't have a true Opteron system handy right now, but that would make for a more interesting test anyway, along with an EM64t system.

Celeron (Coppermine) 700Mhz 256MB RAM Linux 2.6

-march=i586 -O3

CRC32 : 15.620s oneAtATimeHash : 14.180s FNVHash : 15.040s BobJenkins : 9.390s SuperFastHash : 10.870s

Seems as if you need cache for SuperFastHash to show its stuff relative to the others.

-- Randy Howard (2reply remove FOOBAR) "Making it hard to do stupid things often makes it hard to do smart ones too." -- Andrew Koenig