So, after tinkering for a while on PhysX, I have a script that build "noodles" of capsules and I am quite happy with it.
However, I would like to try using the GPUs (I have 1 Geforce and 3 tesla cards) to do the simulation instead of the CPU.
Here is what I have...
//sceneDesc.cpuDispatcher = PxDefaultCpuDispatcherCreate(1); //Creates default CPU dispatcher for the scene
//testing GPU dispatcher
PxProfileZoneManager* profileZoneManager = &PxProfileZoneManager::createProfileZoneManager(gFoundation);
PxCudaContextManagerDesc cudaContextManagerDesc;
PxCudaContextManager* cudaContextManager = PxCreateCudaContextManager(*gFoundation,cudaContextManagerDesc,profileZoneManager);
sceneDesc.gpuDispatcher = cudaContextManager->getGpuDispatcher();
//testing GPU dispatcher
As you can see, while using the cpuDispatcher is easy, using the gpuDispatcher requires quite a number of lines of codes.
However when the script is launched I get 1 error lines...
..\..\PhysX\src\NpPhysics.cpp (229) : invalid parameter : Physics::createScene: desc.isValid() is false!
... along with many warnings...
'phystest3.exe' : Chargé 'C:\Windows\SysWOW64\<multiple name>.dll', Cannot find or open the PDB file
But these warning don't stop the script from working properly when using the CPU.
Before anyone asks, I updated Cuda to the latest available on nVidia site, and updated my graphic drivers because I had an error message from the script itself telling me the Cuda version and graphic drivers on my machine were a bit old.
Searching google for this specific error return few useful links. I am at my wit's end as to what to do now.