Time per frame.

bothered

Hi all,
I have three PCs folding. I recently upgraded two of them and thought they would fold faster, not so. PC3 has my old AMD 3200+. All three PCs are working on the same WU, p2107.
PC1, AMD 64 Venice 3500, TPF 51.59
PC2, AMD 64 Venice 3200, TPF 53.54
PC3 AMD 3200+ Socket A, TPF 50.55
Just thought it was interesting.

bikerboy

do you have them all set to recieve deadlineless wu's? i know that AMDs can run better on those wu's.

bikerboy

bothered

Hmmm, I don't know. Folding has been on these PCs for quite a while but I think deadlines were not selected. I didn't think there was anything wrong with the general times, just odd that the oldest, non 64 CPU, was faster. Do you think they should all be faster?

QCH

Now I'm not an expert but I've heard that the upper end 32-bit Athlons are a bit faster than the 64-bit at the same clock speed...

My limited research shows:

• Venice 3200+ = 2.0 GHz
• Venice 3800+ = 2.4 GHz
• Athlon 3200+ = 2.2 GHz

I would assume the Athlon 3200+ would beat the Venice 3200+ and be very close to the 3800+ Venice. I'm not sure if there is some wasted overhead on the 64-bit CPU's?

Esso

The compiler used to make the folding software is from Intel.
And Intel makes sure that SSE2 is not used if the program is running on the AMD platform.
Therefore some of the WU's runs faster on the Intel platform.

(Note. As far as I know AMD is doing the same with their compilers).

What I would like Stanford to do is to compile some WU's (with AMD compiler) ,
to be running exclusive on the AMD platforms.
And some WU's for Intel platforms.

We the people that are folding should be given the best WU's at any time.
Without the problem to configure the folding parameters correctly from time to time.

In the end, some people will dye from cancer because of the blocking of processor capabilities.

So Standford please find one human friendly compiler !