CUDA Error: no kernel image is available for execution on the device

[AdamBear]

Failed to run with P100 GPU which works fine with other pytorch cuda code, GPU info:
GPU Device 0: "Pascal" with compute capability 6.0
Compute 6.0 CUDA device: [Tesla P100-PCIE-16GB]

error trace:

/opt/conda/lib/python3.8/site-packages/cpm_kernels/library/cuda.py in checkCUStatus(error)
    214 def checkCUStatus(error : int) -> None:
    215     if error != CUDA_SUCCESS:
--> 216         raise RuntimeError("CUDA Error: %s" % cuGetErrorString(error))
    217 
    218 @cuda.bind("cuDriverGetVersion", [ctypes.POINTER(ctypes.c_int)], CUresult)

RuntimeError: CUDA Error: no kernel image is available for execution on the device

What the minimal compute version does cpm-kernels need?

[a710128]

BMInf requires compute capability 6.1. The description in the current documentation may inaccurate.

[AdamBear]

I see, thanks!