Repository navigation
MLPerf Client 0.6.0
Changes in v0.6
- Added support for accelerated execution on Intel NPUs.
- Added support for device enumeration and selection so users with multi-GPU systems can specify the devices on which to run the benchmark.
- Updated the ONNX Runtime, ONNX Runtime GenAI, and Intel OpenVINO software components.
Note: The benchmark workloads in v0.6 are the same as in v0.5, but performance may differ due to the updated software components. Please note this dynamic when making comparisons across versions.