Hello,
I have a problem to run two instance of DirectX 9 applications on the same time; both applications used CUDA interoperability with the driver API layer.
From the NVIDIA_CUDA_ProgrammingGuide_3.0.pdf:
“A CUDA context may interoperate with only one Direct3D device at a time …”
Means this restriction system wide? And then yes it is possible to use a global lock or something to synchronize them?
On the other hand I found this line in chapter “C for CUDA” not for “Driver API”.
And I have tested this problem with Windows7 and its works with more then 2 applications parallel.
System:
-WindowsXP SP 3.
- Quadro FX 3700
- NVIDIA driver: 191.87 and 195.39
- Cuda: 2.3 and 3.0beta
- TestApp: NVIDIA Corporation\NVIDIA CUDA SDK\projects\cudaDecode\
Can anybody help me?
Thanks Mirek