⚡ WebGPU Benchmark Results (1.88x speedup) – M1 Max WebGPU up to bs=128
#54
by
pcuenq
HF staff
- opened
| Batch Size | WebGPU (fp16) | WebGPU (fp32) |
| 1 | 15.90 | 19.20 |
| 2 | 23.30 | 31.20 |
| 4 | 35.60 | 52.80 |
| 8 | 53.30 | 99.70 |
| 16 | 94.20 | 197.20 |
| 32 | 192.70 | 389.20 |
| 64 | 399.60 | 764.40 |
| 128 | 796.60 | 1494.60 |
- Model: Xenova/all-MiniLM-L6-v2
- Tests run: WebGPU (fp16), WebGPU (fp32)
- Sequence length: 512
- Browser: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/122.0.0.0 Safari/537.36
- GPU: vendor=apple, architecture=common-3, device=, description=