First time GPU folder

[anandhanju]

G'day! Long time Folder, first time GPU Folder. I finally got an opportunity to try Folding on a GPU as I have some Google credits to spare. I set up two separate 4 core, 4 GB Ubuntu 18.04 instances, the first with a Tesla P100 and the second with a Tesla V100.

With no CPU slots running, and both instances working on WUs from the same project, I'm seeing 1.03M PPD on the P100 and 2.28M PPD on the V100. At this rate, I'll get as many points in a week as I did in the past 15 years. Are these the expected PPDs on a P100/ V100? Is there anything from the

Code: Select all

nvidia-smi

outputs below to indicate these numbers can be improved?

P100:

Code: Select all

root@instance-4:~# nvidia-smi
Thu Oct  4 18:35:19 2018       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 410.48                 Driver Version: 410.48                    |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  Tesla P100-PCIE...  Off  | 00000000:00:04.0 Off |                    0 |
| N/A   59C    P0   124W / 250W |    317MiB / 16280MiB |     95%      Default |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0      2079      C   ...VIDIA/Fermi/xxx/Core_21.fah/FahCore_21   307MiB |
+-----------------------------------------------------------------------------+

V100:

Code: Select all

root@instance-5:~# nvidia-smi
Thu Oct  4 18:36:16 2018       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 410.48                 Driver Version: 410.48                    |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  Tesla V100-SXM2...  On   | 00000000:00:04.0 Off |                    0 |
| N/A   53C    P0   145W / 300W |    395MiB / 16130MiB |     92%      Default |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0      1843      C   ...VIDIA/Fermi/xxx/Core_21.fah/FahCore_21   385MiB |
+-----------------------------------------------------------------------------+

[bruce]

My drivers are older than what you're showing ... and my GPUs are considerably older/slower so I don't have a very good perspective, but your GPU-Util of 92% and 95% is certainly in an appropriate range. My Temps are 72C and 80C so you have some headroom there.

Maybe somebody with a system more like yours will come along and offer better information.

What projects happen to be running?

[anandhanju]

Thanks for your comments Bruce.

What projects happen to be running? -- 11728 at the time these figures were posted.

[foldy]

Tesla P100-PCIE ~= gtx 1080ti = 1m PPD
Tesla V100-SXM2 ~= rtx 2080ti = 2m PPD

Looks good!

[anandhanju]

Thanks foldy!