LLM Training Throughput on Qwen2.5-7B 64K context (train)
1,560.38Throughput (tokens/sec)OOMB (Sparse Attn)
Evaluation results
| Method | Links | |
|---|---|---|
| OOMB (Sparse Attn)Hardware=1xH200, Attention Type=Sparse, Tensor Parallel=false2026.02 | 1,560.38 | |
| OOMB (Sparse Attn)Hardware=4xH200, Attention Type=Sparse, Tensor Parallel=true2026.02 | 1,542.85 | |
| OOMB (Dense Attn)Hardware=1xH200, Attention Type=Dense, Tensor Parallel=false2026.02 | 936.22 | |
| OOMB (Dense Attn)Hardware=4xH200, Attention Type=Dense, Tensor Parallel=true2026.02 | 933.84 | |
| RFAHardware=4xH2002026.02 | 872.51 |