Folding Forum

All questions have to do with the new SMP and GPU2 clients.

1. Firstly, how do I measure the PPD everybody is talking about? I have a feeling that while I run the GPU2 client on the 8800GT the addition of a Q6600 clocked at 3GHz does not add a lot to it.

2. I noticed in the new clients that If I run the SMP and the GPU2 clients together, the CPU usage is at 100% but there is a drop in temperatures for both CPU and GPU. If I run them alone the temps are usually higher.

3. In the new SMP version while setting up the client, it asks to set a size of WUs, Small/Normal/Big. This is new right?
I set it to big and it seems it goes slower than it used to with the old client with set to "Accept
Wus larger than **Mb=yes". Did WUs got bigger or the slightly slower progress is just in my mind?
Could this be why it seems to go slower?

1) Divide the points value of the unit by the time taken to process it (in days). Or install something like FahMon which will calculate PPD for you (and allow you to monitor any machine on your network from one place, if you fold on more than one machine).

2) I have no idea why that would occur.

3) It's just in your mind, lol. SMP clients are hardcoded to use the big unit option, so whatever you do will not have an effect. That option only works if the client is being used in uniprocessor mode.

Thanks a lot for the FahMon! Had no idea about it.

1. Now FahMon reports Total PPD: 5860.92 Is this ok?

2. The SMP client says Credit: 1920 points(and goes slow) but the GPU Client says Credit: 480 points(and goes fast). Is this normal? Does it mean that the GPU processes smaller WUs?

1) 5860PPD sounds about right, definitely.

2) That is normal, and comes from the huge differences in benchmarks vs. your machine. The SMP benchmark machine is in the higher-end, even for SMP computers, so chances are your machine will achieve a score near the SMP benchmark (currently 1760PPD is the benchmark score). For GPUs however, an 8800GT is significantly quicker than the benchmark graphics card (which for both ATI and nVidia is an ATI Radeon HD3850) so will score an enormous amount more than the 1500PPD benchmark. The GPU units at the moment are fairly small, larger ones will come later.

Thanks very much for the answers, you cleared up a small cloud of questions.

The total PPD went up to 6807.63 after some time and more overclocking.