⚡ WebGPU Benchmark Results (1.32x speedup)
#91
by
Xenova
HF staff
- opened
| Batch Size | WebGPU (fp16) | WebGPU (fp32) |
| 1 | 37.10 | 26.00 |
| 2 | 31.00 | 38.40 |
| 4 | 56.20 | 66.70 |
| 8 | 95.80 | 118.40 |
| 16 | 176.10 | 230.30 |
| 32 | 344.00 | 454.50 |
- Model: Snowflake/snowflake-arctic-embed-s
- Tests run: WebGPU (fp16), WebGPU (fp32)
- Sequence length: 512
- Browser: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/123.0.0.0 Safari/537.36
- GPU: vendor=nvidia, architecture=turing, device=, description=