⚡ WebGPU Benchmark Results (233.94x speedup) - Ubuntu 3090 Ti
#51
by
pcuenq
HF staff
- opened
Batch Size | WASM (int8) | WASM (fp16) | WASM (fp32) | WebGPU (int8) | WebGPU (fp32) |
1 | 485.50 | 706.90 | 664.20 | 793.60 | 30.00 |
2 | 934.40 | 1425.30 | 1330.80 | 1249.80 | 51.40 |
4 | 1904.80 | 2929.60 | 2771.00 | 2100.90 | 97.70 |
8 | 3898.60 | 5863.20 | 5502.90 | 3860.90 | 287.60 |
16 | 7886.60 | 12020.40 | 11108.70 | 7249.20 | 330.30 |
32 | 16368.70 | 24586.70 | 22498.90 | 13512.10 | 105.10 |
- Model: Xenova/all-MiniLM-L6-v2
- Tests run: WASM (int8), WASM (fp16), WASM (fp32), WebGPU (int8), WebGPU (fp32)
- Sequence length: 512
- Browser: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/122.0.0.0 Safari/537.36
- GPU: vendor=nvidia, architecture=lovelace, device=, description=
e/acc