|
Download benchmarks/README.md from flashrt/fp8-gemm: direct link, hf CLI and curl.
- Browser
- Download file 395 Bytes
-
https://huggingface.co/flashrt/fp8-gemm/resolve/main/benchmarks/README.md
- Command line
-
hf download hf://flashrt/fp8-gemm/benchmarks/README.md
-
curl -L -o README.md https://huggingface.co/flashrt/fp8-gemm/resolve/main/benchmarks/README.md
395 Bytes
| # Benchmarks | |
| ```bash | |
| python fp8-gemm/benchmarks/benchmark.py --backend source --mode headline | |
| ``` | |
| The benchmark sweeps dispatcher rows and explicit M=1 GEMV variants. | |
| On NVIDIA Thor, use `--mode thor-full`. The runner records normal launches, | |
| CUDA Graph replay, all SM110 Sq/T1/Wide diagnostic tiles, and an optional | |
| original FlashRT pointer-API comparison when `FLASHRT_NATIVE_ROOT` is set. | |