We are incrementally improving the cuML benchmark harness so it can support declarative benchmark suites, cleaner result artifacts, better console output, stronger CPU/GPU comparisons, and cuml.accel benchmarking.
This issue tracks the benchmark-harness improvement work across the current PR stack
Tasks:
We are incrementally improving the cuML benchmark harness so it can support declarative benchmark suites, cleaner result artifacts, better console output, stronger CPU/GPU comparisons, and cuml.accel benchmarking.
This issue tracks the benchmark-harness improvement work across the current PR stack
Tasks: