Ahhhh Killer project

P5900, R7, C722, G12
472 points

ppd whilst running project on 8800GS according to FAHMON 906.24

I don't like this one

 

Hmm that sounds a bit off. My 9600GT got one of those and managed 3741PPD with fewer SPs.

 

The 8800GS's don't like those units, I've had a few of them over the past day or so as well and my 8800GS's do the same as Saspros, the PPD drops like a stone when they are working on a 472 pointer. My GTS's run them much faster.

 

What ppd are you getting for the 1888 pointers (P5901-5)? Not too familiar with the GS but judging from other cards you should be getting around the same ppd for the smaller 472 point versions (P5900 & P5906).

 

Not sure about what I'm getting on the GS's with 1888 point WUs, but my trusty 9800GT is estimated to be producing 4215 PPD in Fahmon whilst working on one.

 

I would've thought the GS shouldn't be less than 25% of that. Something seems amiss.

 

Just picked up a 5905 on one of my 8800GT's, showing 4027 PPD completing in approx 11hrs 10mins (NO O/C).

 

Yes, that is curious why the 8800GS is so poor with those 5900's. I've not had one yet on the 8800GS or 9800GX2's I'm running.

Code:

 Project : 5900
 Core    : GPU2 NV MT
 Frames  : 100
 Credit  : 472

 -- c7p6t GPU3 GTX295 --
 Avg. Time / Frame : 1mn 00s  - 6796.80 ppd

 -- c2ep45ex GPU1 GTX260 --
 Avg. Time / Frame : 59s  - 6912.00 ppd

 -- c2p5kp GPU1 8800GTX --
 Avg. Time / Frame : 1mn 30s  - 4531.20 ppd

 

P5905 (a 1888 pointer) gets ~3200PPD on an 8800GS at stock.

 

Sorry to ask such a seemingly silly question, but which program are you getting the above stats from JackOfAll?

It seems there is something wrong with both Saspro's and my own 8800GS setup here... I'm going to have to investigate and see if I can find anything wrong.

 

Fahmon. Tools->Benchmarks.

 

It's FahMon's benchmark window:

Code:

 Project : 5905
 Core    : GPU2 NV MT
 Frames  : 100
 Credit  : 1888


 -- Gecko 5 --

 Min. Time / Frame : 8mn 03s  - 3377.29 ppd
 Avg. Time / Frame : 8mn 10s  - 3329.04 ppd
 Cur. Time / Frame : 8mn 03s  - 3377.29 ppd
 R3F. Time / Frame : 8mn 09s  - 3335.85 ppd
 Eff. Time / Frame : 8mn 15s  - 3295.42 ppd


 -- Ferret 1 --

 Min. Time / Frame : 8mn 18s  - 3275.57 ppd
 Avg. Time / Frame : 8mn 36s  - 3161.30 ppd
 Cur. Time / Frame : 8mn 32s  - 3186.00 ppd
 R3F. Time / Frame : 8mn 37s  - 3155.19 ppd
 Eff. Time / Frame : 8mn 36s  - 3161.30 ppd


 -- Ferret 2 --

 Min. Time / Frame : 8mn 25s  - 3230.16 ppd
 Avg. Time / Frame : 8mn 29s  - 3204.78 ppd
 Cur. Time / Frame : 8mn 32s  - 3186.00 ppd
 R3F. Time / Frame : 8mn 33s  - 3179.79 ppd
 Eff. Time / Frame : 8mn 28s  - 3211.09 ppd

 

uncle_fungus has machines called Ferret and Gecko. Anything else we should know?

 

Not sure there is anything wrong. Except that for some reason those Project 5900, 472 pointers, just dont agree with the 8800GS. I'm sure I found a post about this on another forum. Will have a search.

 

There's one called Iguana, and another called Salamander too

 

He likes lizards...

Although it says that if you look

 

You learn something new every day

Thanks!

 

That's probably one of the best features of Fahmon, IMHO. Its local stats database grows with every unit you fold, making it very easy to compare WU's on your specific hardware, look for problems, identify WU's that don't work so well on specific cards, etc. etc.

 

P5905 gets 3236 ppd on the 8800GS
I've managed to get the other project up to 2200 ppd more or less with some rebooting & refreshing of FAHMON

 

A 1888 pointer, I get between 3690-3802PPD with my 9600GSO's( the good ones)