ptxas error uses too much shared data

[BloodLemonS]

I'm really sorry to bother you again. I encountered "ptxas error uses too much shared data" in the final compilation process. I saw that the Cmakelist file needed to be modified in the previous communication with others. Could you please tell me how to make a detailed modification

[BloodLemonS]

I'm using CUDA11.2

[BloodLemonS]

I have also set up CUDa10.2 using Anaconda

[BruceCanovas]

Try to remove the line 40 of the CMakeLists.txt (include("${PROJECT_SOURCE_DIR}/cmake/UseCUDA.cmake")) and instead set the NVCC flags that fit the architecture of your CUDA GPU.

(PS : I apologize because my CMakeLists.txt is written in a pretty old fashioned way that should be deprecated now)