Post Your GPU Folders and their PPD!
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
Post Your GPU Folders and their PPD!
An NVidia 8800GS, low speed, shaders speeded up a bit, about 4189PPD so far.
Added - Increased shaders again, about 3/4 of the way up the slider, now at 4822 PPD.
Added - Increased shaders again, about 3/4 of the way up the slider, now at 4822 PPD.
Last edited by aristide1 on Sat Aug 23, 2008 9:53 am, edited 1 time in total.
2 * XFX 256MB 8800GT's @ 687/1718/752 (core/shader/mem) in one box. Both get around 4,850ppd but I have seen up to 5,200ppd on 'easier' WUs.
For info the 2nd 8800GT is running in an open-ended PCI-E 2.0 4x slot but achieves almost identical ppd to it's twin running in the regular PCI-E 2.0 16x slot.
For info the 2nd 8800GT is running in an open-ended PCI-E 2.0 4x slot but achieves almost identical ppd to it's twin running in the regular PCI-E 2.0 16x slot.
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
Re: Post Your GPU Folders and their PPD!
Added - An 8800GTS OC'd - About 5200PPD. It's about to get interesting.aristide1 wrote:An NVidia 8800GS, low speed, shaders speeded up a bit, about 4189PPD so far.
Added - Increased shaders again, about 3/4 of the way up the slider, now at 4822 PPD.
-
- Posts: 22
- Joined: Thu Apr 24, 2008 5:31 am
- Location: Massachusetts
Re: Post Your GPU Folders and their PPD!
You have a ways to go. This is my overclock on an 8800GS:aristide1 wrote:An NVidia 8800GS, low speed, shaders speeded up a bit, about 4189PPD so far.
Added - Increased shaders again, about 3/4 of the way up the slider, now at 4822 PPD.
http://www.techpowerup.com/gpuz/34vev/
Here are my benchmarks from FahMon:
Code: Select all
Project : 5015
Core : GPUv2 Gromacs
Frames : 100
Credit : 480
-- FAH-NV --
Min. Time / Frame : 1mn 20s - 5184.00 ppd
Avg. Time / Frame : 1mn 20s - 5184.00 ppd
Project : 5504
Core : GPUv2 Gromacs
Frames : 100
Credit : 480
-- FAH-NV --
Min. Time / Frame : 1mn 14s - 5604.32 ppd
Avg. Time / Frame : 1mn 14s - 5604.32 ppd
Project : 5506
Core : GPUv2 Gromacs
Frames : 100
Credit : 480
-- FAH-NV --
Min. Time / Frame : 1mn 20s - 5184.00 ppd
Avg. Time / Frame : 1mn 20s - 5184.00 ppd
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
Dang, there's a decent amount of new folders that I wush would post their info.
I don't want to OC very much because my experience has been that the eneergy costs rise faster than the amounts of points gained. In fact that will probably be adquate reason to return this 8800GTS G92, but I will measure its actual consumption before the do that. And man can it create heat. Big time.
I don't want to OC very much because my experience has been that the eneergy costs rise faster than the amounts of points gained. In fact that will probably be adquate reason to return this 8800GTS G92, but I will measure its actual consumption before the do that. And man can it create heat. Big time.
-
- Posts: 22
- Joined: Thu Apr 24, 2008 5:31 am
- Location: Massachusetts
I happened to have my Kill-A-Watt on my pc so I measured the difference between stock clocks and my current overclock. It looks like it takes an additional 12-14 watts for ~1500PPD. YMMV.aristide1 wrote:Dang, there's a decent amount of new folders that I wush would post their info.
I don't want to OC very much because my experience has been that the eneergy costs rise faster than the amounts of points gained. In fact that will probably be adquate reason to return this 8800GTS G92, but I will measure its actual consumption before the do that. And man can it create heat. Big time.
I have an 8600 "attempting" to fold but it crashes so often I only get 250-350PPD.
not sure if it was a bad batch of WU or what but the past few days it actually completed several WU in a row - then the weather turns hot and I shut my PC off last nite.
but the regular SMP client is still chugging along.
not sure if it was a bad batch of WU or what but the past few days it actually completed several WU in a row - then the weather turns hot and I shut my PC off last nite.
but the regular SMP client is still chugging along.
-
- Posts: 22
- Joined: Thu Apr 24, 2008 5:31 am
- Location: Massachusetts
I'm running at 750/1850/1045 settings in rivatuner. It reports that the card is running at 756/1836/1044. It appears that overclocking is done indirectly by changing a relatively coarse multiplier. The shader clock appears to step by 54MHz, the core and the memory clocks by 5.4MHz steps.aristide1 wrote:Hey SKeptical Thinker,
What are you speed settings? I left everything except the shaders stock:
550/1800/800
The card becomes unstable at shader of 1900 but no BSOD. I cranked up the fan and the temp dropped below 60C, but that made no difference.
When I increment any of the clocks to the next step, the card continues to run, but the folding core declares that the machine was unstable. EVGA's precision tool is even more conservative. According to it, all of the clocks I am running cause the card to be unstable.
In any case, on my card the most important clock is the shader, followed by the core, then the memory. It's hard to quantify due to the bizarre interactions between the core and shader clocks. It appears that you have to move them both up together to maximize performance. At some specific values for the shader clock, the core clock falls to 100MHz.
Fire up the hardware monitor in rivatuner to see what is happening. That helped me out considerably.
Of course, YMMV.
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
Well with the current batch of WU's my GPUs are worth barely more than 3000 PPD, a long ways down from the 4500+ PPD there were earning just a week ago.
And if that wasn't enough they also are causing large lags in 2D performance, which FAH team is well aware of, for all that's worth.
I was going to get a mobo with 4 PCI-E X16 slots, and 4 video cards, now I am far less excited about the whole thing.
And if that wasn't enough they also are causing large lags in 2D performance, which FAH team is well aware of, for all that's worth.
I was going to get a mobo with 4 PCI-E X16 slots, and 4 video cards, now I am far less excited about the whole thing.
-
- Patron of SPCR
- Posts: 749
- Joined: Sat Nov 11, 2006 6:02 pm
I'm having a problem with my reserator, I can't fold until I solved this and I don't know what to look for.
Current temps:
49/40 °C idle (core/ambient) (heavily underclocked: 268 core / 675 shader / 450 memory - rivatuner).
As said, the aluminium reservoir in the back does all the cooling.
Open it up, ditch the default pump and somehow (this is the thing) connect the outside tube connector to the radiator, otherwise the water will never go through it
Current temps:
49/40 °C idle (core/ambient) (heavily underclocked: 268 core / 675 shader / 450 memory - rivatuner).
As said, the aluminium reservoir in the back does all the cooling.
Open it up, ditch the default pump and somehow (this is the thing) connect the outside tube connector to the radiator, otherwise the water will never go through it
-
- Patron of SPCR
- Posts: 744
- Joined: Tue Mar 04, 2008 4:05 am
- Location: London
- Contact:
-
- Patron of SPCR
- Posts: 744
- Joined: Tue Mar 04, 2008 4:05 am
- Location: London
- Contact:
-
- Posts: 22
- Joined: Thu Apr 24, 2008 5:31 am
- Location: Massachusetts
The benchmark machine has an ATI card in it. The differences in the architectures of the NVidia and ATI cards affect the relative performance on proteins of different sizes. NVidia has faster, but fewer, shaders so it can fold small proteins much faster than ATI. When the proteins are complex enough that NVidia cards have too few shaders to efficiently parallize the operations, ATI starts to have the advantage because it can do more operations in parallel.FartingBob wrote:After 24 hours of folding on my 4850 (8.9 drivers) ive only got around 2200 points. This is much lower than what others are getting with older, slower cards. Am i doing something wrong, or are nvidia cards just far more efficient than ATI cards at this?
The view from Stanford is that once the software for both cards is fully optimized and the proteins are complex enough to saturate both architectures, the performance differences will be relatively small.
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
Re: Post Your GPU Folders and their PPD!
Temps?SKeptical Thinker wrote:You have a ways to go. This is my overclock on an 8800GS:aristide1 wrote:An NVidia 8800GS, low speed, shaders speeded up a bit, about 4189PPD so far.
Added - Increased shaders again, about 3/4 of the way up the slider, now at 4822 PPD.
http://www.techpowerup.com/gpuz/34vev/
Not too shabby for a $64 card.
-
- Posts: 323
- Joined: Wed Jul 20, 2005 10:56 am
- Location: USA
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
Sometime the servers that accept completed WUs or the reports get backed up, and then everything gets counted all at once. I once completed like 53 WUs in a 3 hour period. Well, on paper I did.warriorpoet wrote:I had a very strange day about two weeks ago where I did 11k. Now I'm down around 5300PPD again. Wonder what happened...
8800GTX 630/1566/ 2006
e6600 @ 3GHz
There's a GTX280 and a 3GHz quad about to be added to the SPCR ranks, gentlemen
-
- Posts: 323
- Joined: Wed Jul 20, 2005 10:56 am
- Location: USA
-
- *Lifetime Patron*
- Posts: 4284
- Joined: Fri Apr 04, 2003 6:21 pm
- Location: Undisclosed but sober in US
Welcome Einstein. Oh you'll be bright red in no time (a week) over here:
http://folding.extremeoverclocking.com/ ... s=&t=31574
Then you will race up the ranks and end up on page one, just like WP. He's on page 2 already.
Are you einsteingodel? I don't see any other einsteins on our team at the moment.
http://folding.extremeoverclocking.com/ ... s=&t=31574
Then you will race up the ranks and end up on page one, just like WP. He's on page 2 already.
Are you einsteingodel? I don't see any other einsteins on our team at the moment.
-
- Posts: 323
- Joined: Wed Jul 20, 2005 10:56 am
- Location: USA
-
- Posts: 323
- Joined: Wed Jul 20, 2005 10:56 am
- Location: USA
I'm jealous - I picked up an open-box special 9800GTX+ and even with it overclocked to 1890mhz shaders it still only does 4100ppd on the 511 pointers. The other bummer is that apparently there's a bug where if you mix cards with different shader amounts, one card runs dramatically slower than normal. Case in point, I had two 9800GT's doing 32-3500ppd on the 511's. I replace the slower of the two 9800GT's with the GTX+ and suddenly the GT that was doing 3500 drops to 2200. I check on folding forum and find out after the fact that it's a known problem.warriorpoet wrote:Heh. The GTX is cruising at 5580 PPD on a 511 point WU as we speak
-
- Posts: 323
- Joined: Wed Jul 20, 2005 10:56 am
- Location: USA
That's no good. I have a 8800GTS 512 for dedicated folding coming Monday. Wonder if I should send it back...AZBrandon wrote:I'm jealous - I picked up an open-box special 9800GTX+ and even with it overclocked to 1890mhz shaders it still only does 4100ppd on the 511 pointers. The other bummer is that apparently there's a bug where if you mix cards with different shader amounts, one card runs dramatically slower than normal. Case in point, I had two 9800GT's doing 32-3500ppd on the 511's. I replace the slower of the two 9800GT's with the GTX+ and suddenly the GT that was doing 3500 drops to 2200. I check on folding forum and find out after the fact that it's a known problem.warriorpoet wrote:Heh. The GTX is cruising at 5580 PPD on a 511 point WU as we speak
BTW, FAHMon just reported average speeds of 6100PPD on a 511 GPU WU. This is with 3x GMP cores on the CPU
240 processing cores FTW!