⚡ WebGPU Benchmark Results (35.43x speedup) – M1 Max
#49
by
pcuenq
HF staff
- opened
| Batch Size | WebGPU (int8) | WebGPU (fp16) | WebGPU (fp32) |
| 1 | 356.90 | 26.60 | 18.40 |
| 2 | 652.90 | 30.40 | 23.30 |
| 4 | 1234.00 | 49.50 | 42.00 |
| 8 | 2410.90 | 73.80 | 77.30 |
| 16 | 4801.80 | 113.00 | 157.00 |
| 32 | 9923.90 | 224.50 | 343.50 |
| 64 | 20731.30 | 429.00 | 664.00 |
| 128 | 47839.60 | 2693.80 | 1350.40 |
- Model: Xenova/all-MiniLM-L6-v2
- Tests run: WebGPU (int8), 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=