README: Kakeya Attention vs PagedAttention / RadixAttention comparison - #129
Merged
Merged
Conversation
…ttention / SGLang RadixAttention / CXL-Ollama (bounded vs grows-with-conversation) Co-authored-by: FluffyAIcode <FluffyAIcode@users.noreply.github.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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What
Adds an attention-memory-management comparison to the README design-philosophy section: vLLM PagedAttention, SGLang RadixAttention, Kakeya Attention, and CXL/Ollama.
Key framing: other engines optimise how the KV is stored/laid out but still store the whole history (memory grows with the conversation → must provision for the total). Kakeya optimises how much is stored (sliding-window bound + global-attention restoration → bounded footprint, provision only for the peak window). The approaches are orthogonal/composable; the cost is restoration compute (recall 1.0; ~AR-parity/1.79–2.06× on CUDA; ~4× more agents/GB).
Files
README.md(new comparison subsection + table)Testing