OpenCL via Remote Desktop not running?
Hi all,

Under both Windows 7 and Windows Server 2008 R2, using the driver linked in the download page (nvdrivers_2.3_winvista_64_190.89_general.exe), I am trying to run the OpenCL samples over Remote Desktop.

They run fine in both OSes when logged in locally to the machine, but when I try to run the programs (command line) when remotely accessing the machine, they fail (appears to not find the device).
Both OS installs are fresh OS installs.

System is a Core 2 Duo with a Geforce 8800 GTS 512.

Specifically, I try to run the "OpenCL Bandwidth Test " and the "OpenCL Device Query".
Output from the bandwidth test app:
"oclBandwidthTest.exe Starting...

Error -1 in clGetDeviceIDs call !!!


TEST FAILED !!!


Press <Enter> to Quit...
-----------------------------------------------------------"

(The app runs fine when run locally).


If this is not a valid usage scenario, could you please explain what scenario is supported for remotely executing/coding/debugging/running OpenCL code under Windows?

Thank you very much!

--Joe
Hi all,



Under both Windows 7 and Windows Server 2008 R2, using the driver linked in the download page (nvdrivers_2.3_winvista_64_190.89_general.exe), I am trying to run the OpenCL samples over Remote Desktop.



They run fine in both OSes when logged in locally to the machine, but when I try to run the programs (command line) when remotely accessing the machine, they fail (appears to not find the device).

Both OS installs are fresh OS installs.



System is a Core 2 Duo with a Geforce 8800 GTS 512.



Specifically, I try to run the "OpenCL Bandwidth Test " and the "OpenCL Device Query".

Output from the bandwidth test app:

"oclBandwidthTest.exe Starting...



Error -1 in clGetDeviceIDs call !!!





TEST FAILED !!!





Press <Enter> to Quit...

-----------------------------------------------------------"



(The app runs fine when run locally).





If this is not a valid usage scenario, could you please explain what scenario is supported for remotely executing/coding/debugging/running OpenCL code under Windows?



Thank you very much!



--Joe

#1
Posted 11/13/2009 08:28 PM   
It's not a valid usage scenario because RDP/services are limited to session 0, which has no access to WDDM devices. GPUs currently always appear as WDDM devices.

This is solved in a future compute-only driver for Tesla cards/servers.
It's not a valid usage scenario because RDP/services are limited to session 0, which has no access to WDDM devices. GPUs currently always appear as WDDM devices.



This is solved in a future compute-only driver for Tesla cards/servers.

#2
Posted 11/13/2009 08:37 PM   
Some people have successfully used VNC instead of Remote Desktop.
Some people have successfully used VNC instead of Remote Desktop.

#3
Posted 11/14/2009 11:06 AM   
[quote name='tmurray' post='950566' date='Nov 13 2009, 12:37 PM']It's not a valid usage scenario because RDP/services are limited to session 0, which has no access to WDDM devices. GPUs currently always appear as WDDM devices.

This is solved in a future compute-only driver for Tesla cards/servers.[/quote]

Is there an eta or beta program for this driver?
I am in an environment that is remote desktop-only and
this is a showstopper for me- I do dev and deployment on
remotely connected machines.

Thanks,
--Joe
[quote name='tmurray' post='950566' date='Nov 13 2009, 12:37 PM']It's not a valid usage scenario because RDP/services are limited to session 0, which has no access to WDDM devices. GPUs currently always appear as WDDM devices.



This is solved in a future compute-only driver for Tesla cards/servers.



Is there an eta or beta program for this driver?

I am in an environment that is remote desktop-only and

this is a showstopper for me- I do dev and deployment on

remotely connected machines.



Thanks,

--Joe

#4
Posted 11/16/2009 08:32 AM   
Scroll To Top