Inference Throughput on 7-layer 512 x 512 MLP
113.4Throughput (TOPS)AIE4ML
Evaluation Results
| Method | Links | |
|---|---|---|
| AIE4MLDevice=Versal VEK280, Generation=AIE-ML, Precision=INT82025.12 | 113.4 | |
| TensorRTDevice=Nvidia 3060 GPU, Generation=Ampere, Precision=INT82025.12 | 14.1 | |
| Core MLDevice=Apple M4 ANE, Generation=2024, Precision=INT82025.12 | 10.5 | |
| hls4mlDevice=VU13P FPGA, Generation=UltraScale+, Precision=INT82025.12 | 3.7 |