Opened this issue 8 years ago · 0 comments
Consider using lazy_static for CUDA and OpenCL kernels.
lazy_static