release 0.45.0: pgw#622 eager-while-compiling with hot-swap#369
Merged
Conversation
…erve immediately A novel input signature at a compiled target no longer stalls 30-60s behind Dynamo+Inductor: the guard routes it to the eager original, one background thread warms the compiled callable with a zero-filled dummy batch (separate CUDA stream, nice +10), and a successful warm atomically hot-swaps the signature to the compiled path. Each completed warm repacks the live cache root and republishes the cell under the same key so the fleet never compiles that (shape, GPU, lane) again. Sequential compile-then-serve is kept for: the boot warmup-proof window, mandatory quantized lanes (w8a8/w4a4), tight VRAM headroom (degrade, never OOM), regional mode, and signature-vocabulary explosions.
PaulFidika
added a commit
that referenced
this pull request
Jul 22, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Promotion chaos -> master: pgw#622 (release 0.45.0).
A compiled target's first call at a novel input signature no longer stalls
30-60s behind a Dynamo+Inductor compile. The consumer guard serves it (and
followups at that signature) through the eager original; one background
thread (nice +10, separate CUDA stream) warms the compiled callable with a
zero-filled dummy batch of the same signature; a successful warm atomically
hot-swaps the signature onto the compiled path. Each completed warm repacks
the live cache root and republishes the cell under the same ck1 key
(mode=replace) so the fleet never compiles that (shape, GPU, lane) again.
Sequential compile-then-serve is preserved for: the boot warmup-proof
window, mandatory quantized lanes (w8a8/w4a4), tight VRAM headroom
(degrade-don't-OOM), regional mode, and signature-vocabulary explosions.
Tests: tests/test_hot_swap_pgw622.py (18, incl. a real CPU torch.compile
round-trip and the race test vs an in-flight eager batch); adjacent
compile/cell suites green (248); mypy + ruff clean.
🤖 Generated with Claude Code