Adding cuda kernel (optimized for sm80) for block-wise 4b quantized float 16 GEMM. #18619
Azure Pipelines / Linux DNNL CI Pipeline
succeeded
Feb 29, 2024 in 1h 13m 11s
Build #20240228.26 succeeded
Details
- Failed: 0 (0.00%)
- Passed: 14,224 (99.95%)
- Other: 7 (0.05%)
- Total: 14,231
Loading