I am trying to write CUDA code in Visual Studio 2019 but the NSight debugger seems to freeze when I try to step over a
extern __shared__ float s_data;. If I remove the shared memory from the code, it seems to work fine. I can set break point later in the code however when I take a step from there, it freezes again.
- Windows 10 Pro
- Evga RTX 3090 FTW3 Ultra
- Visual Studio 2019 Community Edition
- Cuda 11.2
- Compute version 8.6
I am not sure if this is a bug in the debugger if there is special settings I need to debug shared memory.