So, based on my quick initial tests, I'm guessing this is a longshot, but I figure, what the heck
As the subject suggests, I'm trying to run GPGPU apps on a remote box running windows. I installed cygwin on it some time ago, but just decided to try OpenSSH on it. OpenSSH itself is surprisingly easy to setup. I checked, locally, with a cygwin xterm, I can run a GPGPU program no problem. However, I pull up an xterm on the remote windows box, and try to run my program with no options, which should just print a usage statement, and it just hangs. Anyone else tried this? Got any hints as to what may be wrong?
Yeah, I'd done xhost+, ran an xterm, etc. I hadn't tried the compute=:0 option, but it made no difference. I'm assuming cygwin's SSH server is getting in the way somehow, perhaps its "privilege seperation", since the program will run under a local xterm under cygwin, just not a remote one. It's odd though that it just hangs. I pulled up a task manager, and I can see the process was started, it just doesn't do anything! CTRL-C isn't enough to kill it either, I have to use the task manager...though, perhaps its my rerouting of ctrl-c under cygwin/ssh causing the issue....for now I'm just using VNC. If I get more time to fight with it, I'll try...
According to this, Knowledge Base article, you'll need to set up VNC so that your local machine can gain access to the GPU: http://developer.amd.com/support/KnowledgeBase/Lists/KnowledgeBase/DispForm.aspx?ID=19
"AMD’s OpenCL implementation relies on CAL to manage the AMD GPU for general-purpose computations. AMD CAL uses existing API hooks into the display driver to access the GPU. For doing so, CAL needs to have direct access to the display driver."
When the applications are run locally on the graphics console, CAL has direct access to the display driver, but not so when the applications are remotely, this article explains how to set this up using VNC.