Revision History
FlashMLA · 12 revisions
Sizes are character counts of the article source. The signed number is the change from the previous revision.
Recent edit summaries
Detailed summaries recorded by editors. Generic maintenance summaries are omitted here; every recorded revision remains below. Dates describe the edit, not necessarily the event it covers.
- Correction: the Sep 2025 row should not state H800 SXM5
Version 12 · Oct 1, 2026, 10:04 AM
- Correction: FlashMLA is no longer a Hopper-only BF16 decoding kernel (the 30 Sep 2026 release removed Hopper and V3/V3.2/V4.0 support and targets SM100/SM103 plus Ascend 950); removed the claim that MLA decode is memory-bound (DeepSeek's own deep-dive shows it is compute-bound at 128 query heads); dropped two dead references (InfoQ, vLLM 2025/03/10) and two quotes that could not be verified; fixed the DSA link target and mis-ordered citation numbers; added the Huawei Ascend 950 release, FP8/FP4 cache formats, current performance table and vendor ports.
Version 11 · Oct 1, 2026, 09:15 AM
- Added 6 contextual internal links
Version 10 · Jul 23, 2026, 03:53 PM