Messed up F@H point scale

towert7

2[H]4U
Joined
Dec 18, 2003
Messages
2,930
I'm a little confused by stanfords point scale. They said that they do their testing on a P4 2.8GHz computer, and depending on how long it takes, they assign a point value to it.

So then, would'nt it make sense that my P4 2.8GHz should have a pretty constant PPD value? But this is far from it. Some days it gets 250 PPD, other days like today its doing 126PPD. Is it just me, or is stanford's point scale a little iffy?

And maybe I missed it, but is stanford switching away from the large work units for smaller point work units? Almost all of my boxen are running at 1/2 of their average PPD values because of the work units they keep getting. hm
 
I think the benchmark is with SSE2 turned off. Same chip, SSE2 on/off makes a big differance on Gromacs and alike.

edit: ohh yea, regardless of how this thread goes yea the points thing is, and has been, and always will be, screwed up at least to some.
 
Visit [H]ard|Folding, your PPU, PPD,PPW, and PPM is calculated for you.

1 [H]ard Team, 2 Great Projects, 1 Common Goal..... The Cure



Webmaster: www.hardfolding.com
Lurker Extraordinaire
 
It would if you got the same string of WU every day... then it would average out to be the same. I'm sure you don't get the same WU every day though...
 
In addition, some WUs have "bonus"es, either because they take lots of RAM, or are larger down/uploads. For example, the 1140-1150 (600 pointers) projects have a bonus because of their RAM requirements (~115MB), and the QMD WUs have an even larger bonus.

 
AtomicMoose said:
ooooo, ooooo, oooooo, it's majic!
:cool:

Speaking of which.... has anyone heard from him?


Keep on Folding!! For the [H]orde!!

 
OSUguy98 said:
Speaking of which.... has anyone heard from him?


Keep on Folding!! For the [H]orde!!


That's who I thought of too when I read that. Haven't seen him around much but I haven't been around much myself lately. Maybe he's lurking?

 
Back
Top