稀疏 MLA 解码 (V3.2 / FP8 KV): B200 精度修复 (5aa668c) 的吞吐量成本与 topk 成正比

Sparse MLA decode (V3.2 / FP8 KV): throughput cost of the B200 accuracy fix (5aa668c) scales with topk

Hi FlashMLA team — first, thanks for the excellent kernels. While validating the sparse MLA decode path (DeepSeek-V3.2, FP8 KV cache) on B300, we measured the throughput impact of the recent accuracy fix 5aa668c — "Fix flashmla_kv nsa backend accuracy issue on B200" (#5) and found a measurable, reproducible latency cost that grows with topk (up to ~7%). We wanted to share the numbers and a self-contained benchmark for your reference.