xref: /aosp_15_r20/external/pytorch/benchmarks/inference/results/output_256_true.md (revision da0073e96a02ea20f0ac840b70461e3646d07c45) 
1## Batch Size 256 Compile true
2
3| Experiment | Warmup_latency (s) | Average_latency (s) | Throughput (samples/sec) | GPU Utilization (%) |
4| ---------- | ------------------ | ------------------- | ------------------------ | ------------------- |
5| original | 14.698 +/- 0.398 | 26.890 +/- 1.251 | 562.242 +/- 13.201 | 62.266 +/- 2.997 |
6| h2d_d2h_threads | 12.756 +/- 0.215 | 20.780 +/- 1.558 | 716.763 +/- 29.076 | 72.765 +/- 3.263 |
7