Adding cuda kernel (optimized for sm80) for block-wise 4b quantized float 16 GEMM. #18619
Azure Pipelines / Windows GPU CI Pipeline
succeeded
Feb 29, 2024 in 1h 49m 6s
Build #20240228.27 succeeded
Details
- Failed: 0 (0.00%)
- Passed: 39,210 (97.15%)
- Other: 1,150 (2.85%)
- Total: 40,360
Loading