Kernel Optimization on KernelBench 1.0 (test)
0.0063Latency (us)CodeEvo
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CodeEvoKernel=MLP2026.01 | 0.0063 | — | |
| PACE-SingleKernel=GELU2026.01 | 0.0079 | 1.12 | |
| CodeEvoKernel=BMMNorm2026.01 | 0.0095 | — | |
| CodeEvoKernel=Softmax2026.01 | 0.0114 | — | |
| PACE-MultiKernel=Softmax2026.01 | 0.0114 | 1.48 | |
| PACE-MultiKernel=MeanRedu2026.01 | 0.0117 | 1.72 | |
| PACE-MultiKernel=BMMNorm2026.01 | 0.0121 | 2.36 | |
| PACE-MultiKernel=MLP2026.01 | 0.0227 | 1.55 | |
| PACE-MultiKernel=RNN2026.01 | 0.0258 | 1.35 | |
| PACE-MultiKernel=MaxPool2026.01 | 0.0424 | 1.3 | |
| PACE-MultiKernel=ConvT2D2026.01 | 0.181 | 2.79 | |
| PACE-SingleKernel=ConvMax2026.01 | 0.309 | 3.95 | |
| CodeEvoKernel=BatchNorm2026.01 | 0.391 | — | |
| PACE-MultiKernel=BatchNorm2026.01 | 0.391 | 2.81 | |
| PACE-MultiKernel=RMSNorm2026.01 | 0.397 | 2.59 | |
| PACE-SingleKernel=BatchNorm2026.01 | 0.399 | 2.81 | |
| ShinkaKernel=BatchNorm2026.01 | 0.476 | — | |
| PACE-MultiKernel=AlexNet2026.01 | 0.502 | 1.94 | |
| PACE-MultiKernel=LayerNorm2026.01 | 0.581 | 17.38 | |
| PACE-MultiKernel=ConvDiv2026.01 | 0.609 | 1.82 | |
| KBenchKernel=BatchNorm2026.01 | 0.683 | — | |
| PACE-MultiKernel=MatMul2026.01 | 1.06 | 0.97 | |
| PyTorchKernel=BatchNorm2026.01 | 1.1 | — | |
| OpenEvoKernel=BatchNorm2026.01 | 1.1 | — | |
| PyTorchKernel=ConvDiv2026.01 | 1.11 | — | |
| PyTorchKernel=ConvMax2026.01 | 1.22 | — | |
| PACE-SingleKernel=VGG162026.01 | 2.16 | 2.96 |