WSL2, GPU-API with cuDeviceGet failed to initialize cuda

[tyssjx]

Hi,

I have a fresh installed Physx 5.4.1 on a Win11/WSL2, Cuda 16.2, and RTX4060 Laptop. All snippets requiring GPU calls failed with the following message, eg:

SnippetDirectGPUAPIArticulation_64
/builds/omniverse/physics/physx/source/cudamanager/src/CudaContextManager.cpp (597) : warning : cuDeviceGet failed

Failed to initialize cuda context.
The direct GPU API feature is only supported on the GPU.

Note, other snippets such as HellowWorld runs OK. Docker run with Nvidia container and benchmark also runs OK, meaning GPU driver and cuda toolkit are probably working, eg:

---

sudo docker run --gpus all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark

Windowed mode
Simulation data stored in video memory
Single precision floating point simulation
1 Devices used for simulation
MapSMtoCores for SM 8.9 is undefined. Default to use 128 Cores/SM
MapSMtoArchName for SM 8.9 is undefined. Default to use Ampere
GPU Device 0: "Ampere" with compute capability 8.9

Compute 8.9 CUDA device: [NVIDIA GeForce RTX 4060 Laptop GPU]
24576 bodies, total time for 10 iterations: 18.842 ms
= 320.557 billion interactions per second
= 6411.130 single-precision GFLOP/s at 20 flops per interaction

---

Can someone please give some hint on the fix?

Thank you so much!

[tyssjx]

Setting the correct environment variable PHYSX_GPU_DEVICE in wsl2 system env. fixed my issue. Thanks!