Attention Operator Performance on MLA 128 Head Dimension Sequence Length 4k
138.6TFLOPSLLM-TL
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LLM-TLHead Dimension=128, Mask=Causal, GPU=A100, Base Model=DeepSeek-V32025.06 | 138.6 | 8.61 | |
| cuDNNHead Dimension=128, Mask=Causal, GPU=A1002025.06 | 70.3 | — | |
| torchHead Dimension=128, Mask=Causal, GPU=A1002025.06 | 26.7 | — | |
| DeepSeek-V3Head Dimension=128, Mask=Causal, GPU=A1002025.06 | 16.1 | — |