feat(streaming): GPU-side routed-expert slot remap + residency shed - #22
Merged
Conversation
…op (gpu-dispatch Tier 2 front end) kq.route_shed(indices, scores, slot_table) -> (slots, mix, miss_ids, miss_scores): remaps routed ids to arena slots on GPU, sheds every non-resident expert (lazy graphs cannot demand-read disk), renormalizes kept gate weights mass-preserving, reports misses front-packed descending-score for between-token prestage. Shed entries reuse the row's first kept slot with zero mix weight so downstream gather kernels need no changes. CPU eval mirrors the kernel's f32 accumulation order bit-for-bit; GPU parity tests gated on Metal (deferred while a decode run holds the box). Claude-Session: https://claude.ai/code/session_01Szxi6DtQUJcrCgTFUmWcfR
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
No description provided.