CUDA supports several cache load/store modifies (such as `ldg` and `ldlu`). Would be nice if these are supported in `memory.h`: https://docs.nvidia.com/cuda/cuda-programming-guide/05-appendices/cpp-language-extensions.html#low-level-load-store-functions
CUDA supports several cache load/store modifies (such as
ldgandldlu). Would be nice if these are supported inmemory.h:https://docs.nvidia.com/cuda/cuda-programming-guide/05-appendices/cpp-language-extensions.html#low-level-load-store-functions