Attention PRP Testers (with SSE2 non-P4 machines)
 

If you are running LLRNet or manual PRP using a non-P4 SSE2 machine (Athlon64, Opteron, Petium M, Core [B][COLOR="Blue"]*[/COLOR][/B]), could you try adding the following line to your llr.ini:
[code]CPUSupportsSSE2=0[/code]

You need to restart LLRNet/LLR/PRP after making the change. Please let me know if you see any improvement in the iteration times.

[B][COLOR="Blue"]*[/COLOR][/B] I don't know about Core 2. If anyone has Core 2 and willing to do the test, please let me know how that works out.

For me it's worst on an AMD64 3000+. From 3.3 ms/it to 3.9 ms/it.

Carlos

For Athlon-64 X2 3800+: from 3.4 to 4.3 mseconds per iteration

This way your are disabling SSE2 code. Why do you expect to gain any speed :question:

Core 2 E6600
Iteration time before: 1.5 ms
Iteration time after: 2.9ms

not really good ;)
as Kosmaj asked: Why would you expect an improvement, when you disable some optimizations?

AMD 64 3700+
before: 3,092 ms
after: 3,567 ms

a.