There was an error fetching the commit references. Please try again later.
removed the unused `cuda_block_size` for the `gpu.generate_benchmark`
function. For now the only way to set the cuda_block_size size is to use pass it in the `ps.KernelConfig` to the generated kernel