MCPcopy Create free account
hub / github.com/NVIDIA/cutlass / _loop_and_call_kernel

Function _loop_and_call_kernel

python/CuTeDSL/cutlass/cute/testing.py:903–908  ·  view source on GitHub ↗
(iterations: int, workspace_index: int = 0)

Source from the content-addressed store, hash-verified

901 _cuda_success(err, "Error on freeing memory")
902
903 def _loop_and_call_kernel(iterations: int, workspace_index: int = 0) -> int:
904 for _ in range(iterations):
905 current_workspace = workspaces[workspace_index]
906 callable(*current_workspace.args, **current_workspace.kwargs)
907 workspace_index = (workspace_index + 1) % workspace_count
908 return workspace_index
909
910 # Create CUDA events for timing
911 err, start_event = cuda_driver.cuEventCreate(

Callers 2

benchmarkFunction · 0.85
run_profiling_iterationsFunction · 0.85

Calls 1

rangeClass · 0.85

Tested by

no test coverage detected