Fix algorithm name in dataflow walkthrough - #2
Open
RaviTejGuntuku wants to merge 1 commit into
Open
Conversation
the paper that proposed the online softmax algorithm to avoid manifesting the entire O(N^2) QK^T matrix was the ORIGINAL FlashAttention paper, not FlashAttention3.
themoddedcube
added a commit
that referenced
this pull request
Jul 20, 2026
Re-derive the MatE integer accumulator for Qwen2 dims (was "TBD pending re-derivation"). The two integer reduction axes need different widths: - INT24: W4A8 GEMM + Q·Kᵀ (reduce over hidden/head dim ≤4096; 10b product × 4096 = 22b). - INT32: INT8 P·V (reduces over the TOKEN dim -> width scales with context). Worst-case flat attention needs 14+ceil(log2 L) bits; INT24 only covers ~520 tokens, INT32 ~133k. Empirically real Qwen2 stays at 21b and doesn't grow with context (peaky softmax), so INT24 works in practice, but INT32 is required for guaranteed correctness. - FP32 for FP16 tiles (unchanged). Resolves MatE README open-question #2. Evidence: adaptive-precision-attention/analysis/pv_accumulator_width.py. Updates arch.yml (pe_op_attention_pv, accumulator_rationale, key_decisions), MatE README, STATUS log. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
the paper that proposed the online softmax algorithm to avoid manifesting the entire O(N^2) QK^T matrix was the ORIGINAL FlashAttention paper, not FlashAttention3.