public_git
/
whisper.cpp
mirror of https://github.com/ggml-org/whisper.cpp.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

doc: benchmark step by step doc

This commit is contained in:
ciricc 2025-09-20 16:31:45 +03:00
parent 334f40b7ea
commit 50caee3467
1 changed files with 2 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
bindings/go/README.md
View File

@ -207,35 +207,11 @@ make examples
# optionally: ./build/go-model-download -out models
```
### Run CPU benchmarks
### Run benchmarks
```bash
cd bindings/go/pkg/whisper
go test -bench=ContextProcessCPU -benchmem -run=^$ .
go test -bench=ContextProcessBigCPU -benchmem -run=^$ .
# or run all
go test -bench=. -benchmem -run=^$ .
```
### Run GPU benchmarks
GPU runs toggle `UseGPU` in model params. Ensure your build has GPU backends enabled:
- CUDA: build the C libs with CUDA
```bash
cd bindings/go
GGML_CUDA=1 make whisper
```
- macOS Metal: supported by default in the build flags on Darwin.
Then run:
```bash
cd bindings/go/pkg/whisper
go test -bench=ContextProcessGPU -benchmem -run=^$ .
go test -bench=ContextProcessBigGPU -benchmem -run=^$ .
make benchmark
```
### What the benchmarks measure