Remove MMCE one-at-a-time blocking throttle in art loader#11
Merged
NathanNeurotic merged 4 commits intoincebtion-round-fourfrom Mar 19, 2026
Merged
Conversation
Use cacheAdvanceGenerationPreservePrefetch() for MMCE mode in both menuInvalidateArtSelection and menuAdvanceArtSelectionOnMove. Previously each navigation step (scroll within MMCE list or device switch) called cacheAdvanceGeneration() which downgrades ALL CACHE_ENTRY_READY/PRIMED entries to DISPLAYABLE. For MMCE this prevented surviving art from being pre-uploaded to VRAM via cachePrimeReadyTexture, causing a first-draw stall on the 3rd pass after returning from another device. cacheAdvanceGenerationPreservePrefetch() still aborts in-flight MMCE IO, drops queued MMCE requests, and advances the generation. The only difference is READY/PRIMED entries are preserved so cachePrimeReadyTexture can upload them to VRAM during idle time before they are needed. Co-authored-by: NathanNeurotic <109461996+NathanNeurotic@users.noreply.github.com>
… worse) Co-authored-by: NathanNeurotic <109461996+NathanNeurotic@users.noreply.github.com>
Co-authored-by: NathanNeurotic <109461996+NathanNeurotic@users.noreply.github.com>
…queue together The `cacheHasActiveInteractiveModeLocked` + `cacheHasQueuedInteractiveModeLocked` guard in cacheGetTextureInternal prevented BG and SCR art from queuing until the active MMCE request (typically COV) fully finished loading. This added a dead render-frame gap between each art type and meant a slow or missing COV file held up all subsequent art. The art thread already processes requests one-at-a-time (FIFO) so the serial constraint existed only in the queue submission, not execution. With the 1-frame COV head-start delay still in place, COV queues on frame 8 and BG/SCR on frame 9; the art thread then loads them in that order without any extra render-frame waits. On navigation cacheAdvanceGeneration() drops all queued requests and signals abort on the active one — generation tracking discards stale completions correctly. The game-change drop (drop queued request for a different game) is kept as a safety net with an explanatory comment. Co-authored-by: NathanNeurotic <109461996+NathanNeurotic@users.noreply.github.com>
Copilot created this pull request from a session on behalf of
NathanNeurotic
March 19, 2026 16:45
View session
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
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.
MMCE art almost never popped in and navigation felt clogged. The root cause: a blocking guard in
cacheGetTextureInternalprevented BG and SCR from queuing until COV fully finished loading — a slow or missing COV (open() failure on MMCE can take 100–500 ms) stalled all subsequent art types, and an extra render-frame dead gap separated each type.Changes
cacheHasActiveInteractiveModeLocked+cacheHasQueuedInteractiveModeLockedblocking guard from the MMCE throttle block incacheGetTextureInternalcacheAdvanceGeneration()on navigation drops all queued requests at once and handles abort/discard via generation tracking