Hey folks, my PPD blows right now. I've got a bunch of Quadro 4000, 5000, 6000 and K2000's that can be used but on all the remote clients it says CUDA: not detected. I'm testing on one specific machine just now and have just updated to the latest NVIDIA drivers but still no CUDA detection. Could this be because I'm usingRDP or other remote tools to connect? Or am I missing something else really simple? I've not done much GPU folding to be honest so could be me just being a bit of a spanner. Docs producing 5x what I am at the moment, HELP ME haha
Nelio - I might be producing 5x the points, but its a couple of years for me to catch you up First thing to do is to check that the Folder your client is in contains an up-to-date 'GPUs.txt' file - you may need to download the latest file from the Official Forum - GPUs.txt.
.... Then check that your GPU's are listed in the file. If they are great, if they are not, then they will not work until Stanford add them to the list.
If they are in the GPUs.txt list, you will need to Stop the client and Restart it in order for the file to take effect. The next thing is that, for the older GPU's, you might need to add the -forcegpu flag (see Flag options). Are you running Windows or Linux? If Linux, did you download the nVidia drivers directly from nVidia, or via a PPA? If the latter, I have had CUDA problems with these - so go straight to the nVidia website to get them.