Page 2 of 2

Re: GTS450 Folding = 12,000 PPD

Posted: Wed Feb 02, 2011 2:32 am
by Flying Fox
GT 440: http://techreport.com/discussions.x/20340

Assuming the current WU mix still favours higher clocks, does this mean the 440's even higher clock will at least match the 450 in terms of output? I don't see a PCIe connector there so that means 75W TDP for a possible 10k+ ppd? New ppd/watt/$ king?

Re: GTS450 Folding = 12,000 PPD

Posted: Wed Feb 02, 2011 3:24 am
by StuG
Can you fold with 5870's?

Re: GTS450 Folding = 12,000 PPD

Posted: Wed Feb 02, 2011 2:14 pm
by StuG
Bump?

Re: GTS450 Folding = 12,000 PPD

Posted: Wed Feb 02, 2011 2:30 pm
by DancinJack
StuG wrote:
Can you fold with 5870's?


Yes. V6 GPU2 client for ATI 2xxx-5xxx. NV cards are going to get more PPD, but every point counts dude.

http://folding.stanford.edu/English/DownloadWinOther
DL link: http://www.stanford.edu/group/pandegrou ... ay-623.msi

Don't forget you can run an smp client on that PII.

Re: GTS450 Folding = 12,000 PPD

Posted: Wed Feb 02, 2011 2:44 pm
by steelcity_ballin
I've got a quadcore 2500k and gtx560 that should start folding for an indeterminate amount of time starting early next week. My goal is to hit 4ghz+ on the 2500. The gpu probably won't be overclocked.

Re: GTS450 Folding = 12,000 PPD

Posted: Wed Feb 02, 2011 2:57 pm
by DancinJack
steelcity_ballin wrote:
I've got a quadcore 2500k and gtx560 that should start folding for an indeterminate amount of time starting early next week. My goal is to hit 4ghz+ on the 2500. The gpu probably won't be overclocked.


Sweet! At around 4ghz you should get about 10-12K PPD. The 560 should put up around 15K too!

Re: GTS450 Folding = 12,000 PPD

Posted: Wed Feb 02, 2011 3:50 pm
by Ragnar Dan
Flying Fox wrote:
GT 440: http://techreport.com/discussions.x/20340

Assuming the current WU mix still favours higher clocks, does this mean the 440's even higher clock will at least match the 450 in terms of output? I don't see a PCIe connector there so that means 75W TDP for a possible 10k+ ppd? New ppd/watt/$ king?
I'm not so confident this will do that well. My GTS 450 has 192 SP's, runs 930 MHz core/1860 MHz shader, and generally produces in the 9300-10000 range, depending on what else is going on and the WU. Yesterday I changed my SMP client to "low" priority from "idle", though, and the GPU is going slower since, and it seems like the P 6900 SMP WU I'm running is going faster, even though it didn't seem to do much when I ran procexp.exe as Admin to increase the priority of the SMP client to "Normal". So I don't know what to say, but in general I'm doubtful the new card will perform as well as mine does, though it may well be a better buy in energy cost and purchase price terms (mine cost $126 new).

DancinJack wrote:
steelcity_ballin wrote:
I've got a quadcore 2500k and gtx560 that should start folding for an indeterminate amount of time starting early next week. My goal is to hit 4ghz+ on the 2500. The gpu probably won't be overclocked.


Sweet! At around 4ghz you should get about 10-12K PPD. The 560 should put up around 15K too!
I expect he should do better than that with the CPU, depending. Since they're more efficient than Bloomfields, he may well be noticeably in excess of 30K PPD if he runs -bigadv.

Re: GTS450 Folding = 12,000 PPD

Posted: Wed Feb 02, 2011 4:00 pm
by DancinJack
Ragnar Dan wrote:
I expect he should do better than that with the CPU, depending. Since they're more efficient than Bloomfields, he may well be noticeably in excess of 30K PPD if he runs -bigadv.


I think he's probably going to get closer to 20K on bigadv @4ghz(assuming they will finish). From what I've been reading (random folding forums), people are getting somewhere between 25-30K depending on clock speed with -bigadv. I think one that I read was right about 25k @4.5Ghz on project 6900.

Re: GTS450 Folding = 12,000 PPD

Posted: Wed Feb 02, 2011 4:05 pm
by Ragnar Dan
Well that's odd. My 930, @ 3906 MHz, is delivering ~30-31K PPD depending on if I'm doing anything on the machine (when I go away the frames are taking under 34 minutes) on a P 6900. But I'm still unsure how OCing works on those SandyBridges. I bought faster-than-stock memory so I could OC my CPU, but if you can't increase the base clock any more but only increase the multiplier, it would seem like that would make things significantly more difficult for OCing.

Re: GTS450 Folding = 12,000 PPD

Posted: Wed Feb 02, 2011 4:24 pm
by flip-mode
Ragnar Dan wrote:
Well that's odd. My 930, @ 3906 MHz, is delivering ~30-31K PPD depending on if I'm doing anything on the machine (when I go away the frames are taking under 34 minutes) on a P 6900. But I'm still unsure how OCing works on those SandyBridges. I bought faster-than-stock memory so I could OC my CPU, but if you can't increase the base clock any more but only increase the multiplier, it would seem like that would make things significantly more difficult for OCing.

Sandy Bridge overclocking with the base clock is essentially a no-go, which makes higher clocking memory even more useless (if that's possible!). If you want to overclock SB you have to either get one of the unlocked models or, for a much less satisfying overclock, you can get one of the models with turbo boost and essentially tell the chip to run at the turbo frequency all the time.

Re: GTS450 Folding = 12,000 PPD

Posted: Mon Feb 21, 2011 9:00 am
by JPinTO
Those initially high GTS450 PPD seemed to only apply to the initial beta WU's. These things deliver 8-9000PPD.

On sandybridge OCing, you need to get the K-unlocked multiplier. As mentioned, a i7 running ~4Ghz on 8 "cores" is good for ~30k PPD.

My initial tinkering got a stable 4.45Ghz on a 2600K which is good for ~40k PPD on P6900. Nice!

CPU Multiplier @ 44 * 101 BCLK. Some people seem to be able to run the CPU multiplier at 45-48 but I haven't been able to get stable 24x7 yet. Not sure If I want to push it to the bitter edge when running FAH 24x7. The cool temperatures that the 2600K does this overclock at is amazing. Full burn 8 threads at 4.5Ghz the cores run ~60C with a Corsair H70 cooler.

- JP