search

TiledTensor / TiledCUDA

TiledCUDA is a highly efficient kernel template library designed to elevate CUDA C’s level of abstraction for processing tiles.
MIT License
159 stars 10 forks source link

feat(cmake): port googletest into the project #21

Closed haruhi55 closed 7 months ago

haruhi55 commented 7 months ago

resolve https://github.com/TiledTensor/TiledCUDA/issues/18 https://github.com/TiledTensor/TiledCUDA/issues/20 https://github.com/TiledTensor/TiledCUDA/issues/17

I am trying to improve the organization of CMakeLists, but I have run into an issue that I haven't been able to resolve yet.

Specifically, when compiling with nvcc, I need to explicitly set std=c++17 as a compiling option. If I don't do this, I get errors when compiling with torchlibs that require the c++ standard to be set to 17.