Projects 7624-25 Points

[DoctorsSon]

I have been getting 7624 & 25 wu's and have noticed a 4-5k ppd drop from my usual average.
Some others have reported the same with 7623.
While the drop may not seem like a lot it is about a 25% drop for me on that card.
Anyone else notice this yet?

[Zagen30]

Not as big a drop for me, but yes, they do run slower. My GTX 460s are running at 820 core/1640 shaders, and they normally get 14-15.5k PPD. All the 762x get around 11k.

[ArVee]

Yes they're terrible. Same %age losses on gtx465's as well as a 550ti. For a couple months now too, not unbroken but pretty regular.

[k1wi]

I take it we are talking GPU WUs and not CPU projects (perhaps a mod could move this to a more appropriate place?) Edit by Mod: Done.

It's well known that due to the massively different charactersitics of different GPUs (particularly in terms of the number of shaders) that there can be correspondingly large differences in their performance with particular WUs.

This is because WUs with smaller protein counts are less efficient about fully utilising GPUs with a large number of shaders, while larger protein count WUs tend to utilise more of the top end GPU's resources (while the previously already fully utilised smaller GPUs experience no change in utlisation). The reason for the drop in performance is due to the GPU that they use to benchmark the WU responds to the gain in utlisation.

I would be interested to know whether this is largely limited to the sub GT *60 cards, or whether the *70/*80/*90 cards also experience a drop in ppd...

[HaloJones]

Perhaps it would be sensible if the benchmark gpu were more representative of the mass of contributors i.e. a mid-range card such as a *50 so that points are more consistent for the majority,

[7im]

PG did downgrade already. 480 to a 460 quite some time ago. But now with 5xx and 6xx cards on the market, it's a low end card, by comparison. It likely won't be changing again soon.

[gwildperson]

If PG does benchmark with the 460 as youre saying and as everything else seems to say, how is it possible for Zagen30 (above) to say his PPD went down so much?

[7im]

FAH Points FAQ: GPU3 client benchmark

Benchmark machine: GeForce GTX460 . Details:

GPU: VGA GeForce GTX460 768MB DDR5 PCI-Express 2.0 Graphics Card 768-P3-1360-TR
Processor: Dual Core AMD Opteron(tm) Processor 275 (2210 MHz)
Operating System: Windows 7 Professional, 32-bit
DirectX version: 11.0
GPU processor: GeForce GTX 460
Driver version: 258.96
CUDA Cores: 336
Core clock: 720 MHz
Shader clock: 1440 MHz
Memory clock: 1800 MHz (3600 MHz data rate)
Memory interface: 192-bit
Total available graphics memory: 2047 MB
Dedicated video memory: 768 MB GDDR5
System video memory: 0 MB
Shared system memory: 1279 MB
Video BIOS version: 70.04.1B.00.60
Bus: PCI Express x16
PPD: 9787

To be getting 11K PPD seems just fine to me. Only when compared to an overbenchmarked project getting 14K PPD does this seem out of line.

That's how.

[artoar_11]

Yes, PG benchmark machine is GTX 460. I just want to support the words of Zagen30. I have 2x GTX 460/1MB @ 700 MHz (very hot summer). With other WUs I get ~ 12.5k PPD. With these WUs 7623-25 ​​get ~ 9.2-9.6k PPD.
The problem for me is very big lag. For now clench teeth and not stopping GPU.

[DoctorsSon]

That seems to be about a 25% drop for the GTX 460's at lower clock rates also over the normal average.
My 460 is cranked up higher.

[Nathan_P]

Back when the 460's were first released they were only getting 10k-12k in points so the are not far off the mark, your 14k and 9.5k WU should even out and bring you back to a 11-12k average

[Jesse_V]

The problem for me is very big lag. For now clench teeth and not stopping GPU.

If it's any consolation, sometimes I get lag as well. The PG is aware of this issue and in this forum they've pointed out some interesting solutions that they may be pursuiting. I hope that it's taken care of!

[artoar_11]

Just a few thoughts out loud. When running the benchmark machine, it is loaded or only GPU, or CPU. We donors usually CPU+GPU fully loaded. If I think properly, should and bechmark machine to be CPU+GPU loaded. That would be more correct result.

Correct me if I'm wrong.

Jesse_V, I am also in the queue waiting for a decision of the lag problem

[heikosch]

To be getting 11K PPD seems just fine to me. Only when compared to an overbenchmarked project getting 14K PPD does this seem out of line.

That's how.

No! Folding with a GTX460@800MHz for years I can tell you that in 2012 ~14k PPD is typically and 11k PPD is below average.

@artoar_11
The nVidia core doesn´t use much CPU time. There´s no significant difference between GPU and GPU+CPU.
If you really need the extra 100 PPD with the GPU you can use a program to set high CPU priority for the GPU core automatically.

Heiko

[k1wi]

To be getting 11K PPD seems just fine to me. Only when compared to an overbenchmarked project getting 14K PPD does this seem out of line.

That's how.

No! Folding with a GTX460@800MHz for years I can tell you that in 2012 ~14k PPD is typically and 11k PPD is below average.

@artoar_11
The nVidia core doesn´t use much CPU time. There´s no significant difference between GPU and GPU+CPU.
If you really need the extra 100 PPD with the GPU you can use a program to set high CPU priority for the GPU core automatically.

Heiko

You seem to be overlooking the benchmark information that was quoted by 7im - 9787ppd is how much a stock GTx460 should get. The extra 10% that yours is over clocked by would put the 11k bang in line with the stock standard benchmark card... Anything higher than that would be 'above average'. I don't know what WUs earn the 14k PPD but perhaps they're smaller WUs that are on the older 480 benchmarked WUs.