[FP16] Improved performance by fusing dequantize with compute in kernels: 20-30% Inference Speedup #137
mikepapadimbuild-and-run.yml
on: pull_request
code-quality
13s
Matrix: build-and-run
Annotations
2 errors
|
build-and-run (ptx)
Process completed with exit code 1.
|
|
build-and-run (opencl)
The strategy configuration was canceled because "build-and-run.ptx" failed
|