(size)
| 100 | |
| 101 | |
| 102 | def device_mem_alloc(size): |
| 103 | if cutlass_cppgen.use_rmm: |
| 104 | return rmm.DeviceBuffer(size=size) |
| 105 | else: |
| 106 | err, ptr = cudart.cudaMalloc(size) |
| 107 | if err != cudart.cudaError_t.cudaSuccess: |
| 108 | raise Exception(f"cudaMalloc failed with error {err}") |
| 109 | return DevicePtrWrapper(ptr) |
| 110 | |
| 111 | |
| 112 | def align_size(size, alignment=256): |
no test coverage detected