error in nnd updateOutput: invalid device function ; error in nnd get grad: invalid device function

[a4152684]

after I compile Chamfer 2D, 3D, 5D, when I run the "unit_test.py"， it shows me such error:

Loaded compiled 2D CUDA chamfer distance
Loaded compiled 3D CUDA chamfer distance
Loaded compiled 5D CUDA chamfer distance
testing Chamfer 2D
error in nnd updateOutput: invalid device function
error in nnd get grad: invalid device function

AssertionError: chamfer cuda and chamfer normal are not giving the same results

my environment: python3.7 + cuda 10.1 + pytorch 1.6

can you help me? thanks

[ThibaultGROUEIX]

Hi @a4152684 ,

I would start by looking at the output of the cpu/gpu approaches and check which one is failing for you.
Let me know if you get to the bottom of it;

Best regards;
Thibault

[a4152684]

Hi @ThibaultGROUEIX
thank you for your reply
I think the problem lies in the output of the gpu, the output of the gpu is all 0

[ThibaultGROUEIX]

I see, this is very strange. It works on my end so it is hard for me to debug in your stead.
I recommend line-by-line debugging with PyCharm or VS Code to understand what is going on,

Kind regards

[nicolasugrinovic]

I get similar errors:

error in nnd updateOutput: no kernel image is available for execution on the device
error in nnd get grad: no kernel image is available for execution on the device

But it is very weird, this only happens if I execute the code from PyCharm and not when I do from the terminal.