User reported: the drudge statue on top of the Foundry (a multi-part
live-spawned entity with AnimPartChange + texChanges) renders only
PARTIALLY — some parts visible, some missing.
Root cause: the dispatcher's slow path skips a MeshRef when
_meshAdapter.TryGetRenderData returns null (mesh still async-decoding
via ObjectMeshManager.PrepareMeshDataAsync). The classified-batches
collector accumulates only the MeshRefs that DID resolve. At entity
boundary, the cache populates with the PARTIAL set. Frame-2 cache hits
serve that partial entry forever — even after the missing mesh loads,
the cache continues to skip those parts because classification never
reruns for cached entities.
Fix: track currentEntityIncomplete during the foreach. Set it true on
any null renderData. At entity boundary (and at end-of-loop), if the
flag is set, DROP the accumulated populate scratch instead of writing
it to the cache. The slow path retries on the next frame; once all
meshes have loaded, the populate fires correctly with the complete
classification.
Adds a regression test pinning the contract — incomplete entities
produce zero cache entries.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
User confirmed via A/B test (ACDREAM_DISABLE_TIER1_CACHE=1) that the
visual bug — buildings rendering up in the air outside Holtburg — is in
the cache wiring, not elsewhere. The matrix math (restPose * entityWorld
== model) was provably correct, so the bug had to be cache key collision.
Stabs were namespaced in commit 71d0edc, but scenery (0x80LLBB00 +
localIndex) and interior (0x40LLBB00 + localCounter) still have the
same 256-overflow risk. Dense LBs outside Holtburg (forest, urban) push
localIndex past 255, wrapping into the lbY byte and creating cross-LB
collisions.
Fix: change the cache key from uint entityId to (uint, uint) tuple of
(EntityId, LandblockHint). The cache is now correct-by-construction
regardless of any hydration path's Id-generation strategy. Defensive
against future regressions in any ID namespace.
InvalidateEntity becomes a sweep (was O(1)), but it's called rarely
(only on live-entity despawn). InvalidateLandblock was already a sweep.
Updated 14 existing cache tests + 1 dispatcher integration test to thread
landblockHint through TryGet / DebugCrossCheck calls.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
LandblockLoader.BuildEntitiesFromInfo restarted nextId at 1 per landblock,
producing colliding entity.Id values across landblocks. EntityClassificationCache
keys by entity.Id alone, so cross-LB collisions caused cache pollution:
multiple stabs sharing id=1 -> cache entry for id=1 ended up with the
CONCATENATION of multiple entities' batches -> buildings rendered up in the
air with wrong textures (visual gate observation 2026-05-10).
Audit at docs/research/2026-05-10-tier1-mutation-audit.md did not verify
entity.Id uniqueness - that was an unchecked assumption. Cache design
trusted entity.Id was globally unique; for stabs it wasn't.
Fix: optional landblockId parameter on BuildEntitiesFromInfo. When non-zero,
stab Ids are namespaced as 0xC0XXYY00 + nextId, matching the scenery
(0x80XXYY00) and interior (0x40XXYY00) namespacing already in GameWindow.cs.
The 0xC0 top byte distinguishes stabs from those. Existing tests pass
landblockId=0 and keep their legacy starting-from-1 behavior.
Known latent: if any one landblock has >256 stabs, nextId overflows the
low byte. Same pattern + same limitation as scenery/interior. Out of scope
for the immediate Tier 1 cache bug; not affecting current Holtburg play.
Adds 2 regression tests pinning the namespacing + the legacy fallback.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Code review of f16604b flagged that DebugCrossCheck's XML doc claimed
"called once per static-entity cache hit per frame" — overstated. The
method is currently exercised by unit tests only; the dispatcher's
cache-hit branch fires a simpler predicate assert (!isAnimated) at
production hit time, not the full live-state cross-check. Wiring the
full cross-check is the spec section 6.5 stretch goal, kept open as a
follow-up.
Doc-only change. No behavior change. 1708 / 8 baseline preserved.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds EntityClassificationCache.DebugCrossCheck(entityId, liveBatches) that
asserts cached state matches a live re-classification. Wires a simpler
predicate assert into WbDrawDispatcher's cache-hit branch (asserts
isAnimated == false on cache hit). Tests #13a and #13b cover the
batch-count mismatch and clean-match cases via a custom TraceListener
that captures Debug.Assert calls.
Zero cost in Release. In DEBUG, the assert fires immediately if a future
regression mutates static-entity state outside the audit's known write
sites — the same failure mode that bit the prior Tier 1 attempt.
Phase 4 complete. Cache + invalidation + safety net all in place.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
GpuWorldState.RemoveEntitiesFromLandblock now invokes an optional
Action<uint> callback before zeroing the entity list. GameWindow wires
this to EntityClassificationCache.InvalidateLandblock so cache entries
get swept on LB demote (Near to Far) and unload. Per spec section 5.3 W3b.
The callback receives the canonicalized landblock id (low 16 bits forced
to 0xFFFF), matching the LandblockHint stored at Populate time. Trace:
GpuWorldState._loaded keys are canonical (set by AppendLiveEntity),
LandblockEntries yields kvp.Key as LandblockId, WalkEntitiesInto
propagates entry.LandblockId into _walkScratch, the dispatcher's
populateLandblockId reads that tuple and stores it as LandblockHint.
Phase 3 (invalidation hooks) complete. The cache now stays correct across
all spec-identified mutation events: despawn, ObjDescEvent (despawn+
respawn), LB demote, LB unload.
Two integration tests added:
- RemoveEntitiesFromLandblock_FiresUnloadCallbackWithCanonicalId asserts
the callback fires once with the canonical id even when called with a
cell-resolved input (low 16 bits non-FFFF).
- RemoveEntitiesFromLandblock_NotLoaded_DoesNotFireCallback asserts the
early-return path doesn't fire the callback for unknown landblocks.
Tests: 1706 passed / 8 failed (baseline). Sentinel: 110/110.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
GameWindow.RemoveLiveEntityByServerGuid now invalidates the entity's
cache entry next to the existing _animatedEntities.Remove(). Fires for
DeleteObject (0xF747) and the dedup leg of ObjDescEvent (0xF625).
Adds test #15 (despawn-respawn under reused id repopulates fresh) per
spec section 7.5 — pins the audit's ObjDescEvent-as-despawn-respawn contract.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Task 10 (commit 0cbef3c) called ApplyCacheHit inside the per-(entity, partIdx)
foreach loop, but cachedEntry.Batches is flat across all MeshRefs of the
entity. For a 3-MeshRef static building on frame 2: 3 tuples times 6 cached
batches per call = 18 instances drawn instead of 6. Severe Z-fighting and
3x perf hit on every multi-part static entity (buildings, statues, multi-
MeshRef NPCs).
This is the symmetric mirror of the Task 9 bug fixed at 00fa8ae. Both
spec section 5.2 and the plan describe the foreach as per-entity, but
_walkScratch has been per-tuple since Task 6. The implementation
faithfully ported the buggy spec.
Fix: track lastHitEntityId; the cache-hit fast path fires only on the
first tuple of each entity, and subsequent tuples skip the iteration
body via continue. Adds a regression test pinning the per-entity
amplification invariant.
Caught by code review (subagent-driven-development) before Phase 3
dispatched. The bug was invisible in the no-multi-frame-test 1702/8
baseline; would have manifested as visible Z-fighting on every multi-
part building on second-and-subsequent frames once Task 13 perf gate
captured live runs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
WbDrawDispatcher.Draw now branches on cache hit before running classification:
on hit, walks the cached flat batch list and appends RestPose times entityWorld
to the matching groups; on miss, runs today's classification and populates
the cache (Task 9). Animated entities skip the cache entirely.
Adds dispatcher integration tests #11 (static entity populates + reuses)
and #12 (animated bypasses) per spec test plan section 7.2, plus the
multi-MeshRef regression test that would have caught the bug fixed in
commit 00fa8ae (cache populate must flush at entity boundary, not per-tuple).
Phase 2 (dispatcher integration) complete. End-to-end caching now live.
Invalidation hooks (Phase 3) ensure correctness across despawns + LB demotes.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Task 9 (commit 2f489a8) called _cache.Populate inside the per-tuple
foreach loop, but _walkScratch contains one tuple per (entity, MeshRefIndex)
and the cache is keyed by entity.Id. For multi-MeshRef entities (multi-part
Setup buildings, statues, multi-MeshRef NPCs), each iteration's Populate
OVERWROTE the previous one — only the last MeshRef's batches survived.
The bug was invisible at commit time because Task 10 had not landed
(cache populates but isn't read). It would have manifested the moment
Task 10 wired the cache-hit fast path: every multi-part static building
in Holtburg would render as N stacked copies of its last part.
Fix: restructure the per-entity loop with a flush-on-entity-change pattern.
Track the previous entity's Id; when the iteration moves to a different
entity, flush the previous entity's accumulated _populateScratch via one
Populate call. After the loop, flush the final entity. _populateScratch
is now cleared at flush time, not per-iteration.
Caught by code review (subagent-driven-development) before Task 10 dispatched.
Verified: 1699/8 baseline preserved, sentinel 105/105 unchanged.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Restructures Draw's per-entity loop: animated entities still skip the
cache entirely, but static entities now collect their classification into
_populateScratch and call cache.Populate at the end of the iteration.
Cache fast-path (skip slow classification on cache hit) lands in Task 10.
This intermediate state is verifiable: behavior unchanged, but the cache
is being populated as entities render. Diagnostic-friendly split.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
ClassifyBatches now accepts a restPose parameter (the model-matrix
component without entityWorld baked in) and an optional collector. When
collector is non-null, each classified batch is appended as a CachedBatch
record. Defaults preserve today's behavior. Used in Task 9 to populate
the cache on a static-entity miss.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds the cache as a constructor parameter on WbDrawDispatcher and a
private field on GameWindow. The cache is passed through but not yet
consumed by Draw — that wires up in Task 9 (cache miss / populate) and
Task 10 (cache hit / fast path).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Extends the walk scratch tuple from (entity, meshRefIndex) to
(entity, meshRefIndex, landblockId). The dispatcher's per-entity loop now
has the landblock id available for EntityClassificationCache.Populate's
landblockHint argument (consumed in Task 9). No behavior change.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Idempotent removal of a cached entry by entity id. Tests #4 and #5 from
spec section 7.1 lock in the contract.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Implements Populate (insert-or-overwrite) and adds 5 tests covering the
populate->TryGet round-trip including the Setup pre-flatten shape. Per
spec test plan section 7.1 tests #2, #3, #9, #10, #14.
Tests use xUnit Assert.* (not FluentAssertions) to match the Task 2
implementer's choice and the existing 149 sibling assertions in the Wb
test directory.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds CachedBatch, EntityCacheEntry, and EntityClassificationCache with
just TryGet (returns false on empty). The skeleton compiles and the first
test (TryGet_EmptyCache_ReturnsFalse) passes. Subsequent tasks add
Populate, InvalidateEntity, InvalidateLandblock, and the dispatcher
integration. Per spec design Section 6.1.
Note: CachedBatch / EntityCacheEntry / EntityClassificationCache are
internal (not public as the plan snippet showed). Their members
transitively reference the internal GroupKey type, so promoting them to
public produces CS0051 inconsistent-accessibility errors. The cache is
dispatcher-internal coordination state anyway, and the AcDream.App
csproj already exposes internals to AcDream.Core.Tests via
InternalsVisibleTo, so the test sees everything it needs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Mechanical refactor: GroupKey was a private nested record struct on
WbDrawDispatcher. The upcoming EntityClassificationCache (ISSUE #53) needs
to store GroupKey inside CachedBatch records, so it must be visible to
both the dispatcher and the cache. Promoting to internal at file scope is
the smallest change that achieves this.
No behavior change. 1688 tests pass; 8 pre-existing failures unchanged.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The audit at docs/research/2026-05-10-tier1-mutation-audit.md enumerates
every entity.MeshRefs write site (5 STATIC at hydration, 1 DYNAMIC at
GameWindow.cs:7580 inside TickAnimations) and verifies that all 7
Position/Rotation write sites only touch entities in _animatedEntities.
Establishes the load-bearing invariant: an entity's renderer state is
stable from spawn to despawn iff entity.Id is NOT in _animatedEntities.
The spec at docs/superpowers/specs/2026-05-10-issue-53-tier1-cache-design.md
locks in the design from brainstorming on 2026-05-10:
- Static-only cache + DEBUG cross-check (option c) — catches future
regressions of the prior bug class without paying perf cost in Release
- Separate EntityClassificationCache class injected into WbDrawDispatcher
- Cache the rest pose, not the full model matrix (Position/Rotation read
live each frame so Release stays correct even if the invariant breaks)
- Pre-flatten Setup multi-parts at populate time (the bulk of the win)
- 15 new tests covering all invalidation paths + DEBUG cross-check +
Setup pre-flatten + lifecycle pin
Closes the audit + design steps of the post-A.5 polish Priority 3 work.
Implementation plan owned by superpowers:writing-plans next.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
After the post-A.5 lifestone (#52) + JobKind plumbing (#54) work shipped,
only Priority 3 (Tier 1 entity-classification cache retry, ISSUE #53)
remains. This handoff captures the audit insights gathered during the
#52 investigation that the original post-A.5 handoff didn't have:
- MeshRef is a `readonly record struct` — its fields can NOT be mutated
in place. The actual per-frame mutation for animated entities is the
entire MeshRefs LIST replacement at GameWindow.cs:7474-7553. This
reframes the cache design.
- _animatedEntities dict at GameWindow.cs:160 is the source of truth
for which entities go through the per-frame rebuild path.
- Static entity = entity.Id NOT in _animatedEntities. Its MeshRefs is
the same instance from spawn until rare events (ObjDesc / palette
swap / part hide / scale apply).
- Recommended cache approach: static-only with explicit invalidation
hooks on the network/spawn-time write sites enumerated in the doc.
Doc covers: where main is, what shipped this session, why the first
Tier 1 attempt failed, the pre-started audit, cache design options,
acceptance criteria, files to read, workflow for the next session, and
things-to-NOT-do.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Move ISSUE #54 to Recently closed referencing commit `bf31e59`. Drop
#54 from CLAUDE.md "Currently in flight" — only #53 (Tier 1 retry)
remains open in the post-A.5 polish phase.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Bug A's fix (commit `9217fd9`) patched at the worker output by stripping
entities from far-tier `LoadedLandblock`s after the full `LoadNear` path
ran. The worker still wasted CPU on `LandBlockInfo` reads + entity
hydration + `SceneryGenerator` math + interior-cell walks for ~544
far-tier LBs at radius=12, just to throw the work away.
This commit plumbs `LandblockStreamJobKind` through to the factory so the
worker can branch at the source:
- `LandblockStreamer.cs`: replace the `Func<uint, LoadedLandblock?>`
factory with `Func<uint, LandblockStreamJobKind, LoadedLandblock?>` as
the primary ctor signature. Add a back-compat overload that wraps the
old single-arg signature (`(id, _) => loadLandblock(id)`) so existing
test code keeps compiling without modification — the 5 ctor sites in
`LandblockStreamerTests.cs` now resolve to the overload. `HandleJob`
passes `load.Kind` to the factory; the post-load entity-strip is
retained as a `Debug.Assert` + Release safety net.
- `GameWindow.cs`: `BuildLandblockForStreaming(uint, JobKind)` branches
on `kind == LoadFar` at the top — reads only the `LandBlock` heightmap
dat and returns a `LoadedLandblock` with `Array.Empty<WorldEntity>()`.
Skips `LandblockLoader.Load` (which reads `LandBlockInfo`),
`BuildSceneryEntitiesForStreaming`, and `BuildInteriorEntitiesForStreaming`
entirely. Near-tier path is unchanged. Both call sites updated to pass
the kind through the lambda: `(id, kind) => BuildLandblockForStreaming(id, kind)`.
Tests: 1688/1696 (8 pre-existing physics/input failures unchanged).
Streaming-targeted filter (30 tests covering LandblockStreamer +
StreamingController + StreamingRegion) all green via the back-compat
overload — no test code needed updating.
Per-LB worker cost on far-tier: was ~tens of ms (full hydration,
including LandBlockInfo + scenery generation + interior cells); now a
single `LandBlock` dat read (~sub-ms).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Move ISSUE #52 from Active to Recently closed with full root-cause writeup
referencing commit `e40159f`. Strip lifestone reference from CLAUDE.md
"Currently in flight"; remaining post-A.5 polish scope is #53 (Tier 1
retry) + #54 (JobKind plumbing).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three root causes regressed the Holtburg lifestone since the WB rendering
migration (Phase N.5 retirement amendment, commit dcae2b6, 2026-05-08).
All confirmed via temporary [LIFESTONE-DIAG] instrumentation and visually
verified by the user through the +Acdream test character.
1. **Alpha-test discard** in mesh_modern.frag transparent pass killed
high-α pixels of dat-flagged transparent surfaces. Native AC
transparent surfaces routinely include effectively-opaque pixels —
e.g. the lifestone crystal core (surface 0x080011DE) — that compose
correctly under (SrcAlpha, 1-SrcAlpha) blending. The original N.5
§2 rationale ("high-α belongs in opaque pass") doesn't hold for
surfaces flagged transparent at the dat level: those pixels can't
reach the opaque pass at all. Fix: remove `α >= 0.95 discard` from
the transparent pass, keep `α < 0.05 discard` as a fragment-cost
optimization (skip totally-empty pixels).
2. **Cull state** for the transparent pass was unset by
WbDrawDispatcher after the N.5 retirement amendment deleted
StaticMeshRenderer.cs (which had the Phase 9.2 setup at commit
6f1971a, 2026-04-11). Closed-shell translucents — lifestone crystal,
glow gems — need GL_CULL_FACE + GL_BACK + GL_CCW in the transparent
pass; otherwise back faces composite over front faces in iteration
order under DepthMask(false). Fix: re-establish Phase 9.2's exact
GL state setup at the top of Phase 8.
3. **uDrawIDOffset uniform** was missing from mesh_modern.vert.
gl_DrawIDARB resets to 0 at the start of each
glMultiDrawElementsIndirect call, so the transparent pass — which
begins later in the indirect buffer — was fetching
Batches[0..transparentCount) instead of its actual section at
Batches[opaqueCount..end). The lifestone crystal ended up reading
the FIRST OPAQUE batch's TextureHandle every frame; as the camera
moved and the front-to-back opaque sort reordered which group
landed at BatchData[0], the crystal's apparent texture flickered to
whatever sat first — typically the player character's body parts.
Fix: add `uniform int uDrawIDOffset` to the vertex shader, change
Batches[gl_DrawIDARB] → Batches[uDrawIDOffset + gl_DrawIDARB], and
set the uniform per-pass in WbDrawDispatcher (0 for opaque,
_opaqueDrawCount for transparent). Mirrors WorldBuilder's
BaseObjectRenderManager.cs line 845.
Tests: 1688/1696 passing (8 pre-existing physics/input failures
unchanged). N.5b conformance sentinel 94/94 clean.
Visual: Holtburg lifestone now renders with the spinning blue crystal
correctly composed over the pedestal. Other transparent content (glass,
particle effects, NPC clothing) is unaffected — the same uniform fix
applies globally and is correct for all transparent draws.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Captures the three deferred items from A.5 ship:
- ISSUE #52: lifestone visual missing (1-3 hours, fast win)
- ISSUE #54: JobKind plumbing through BuildLandblockForStreaming
(~30 min - 1 hour, worker-thread efficiency cleanup)
- ISSUE #53: Tier 1 entity-classification cache retry (~5-7 days,
biggest perf win remaining; needs animation-mutation audit before
designing to avoid the freeze-pose bug from the first attempt)
Doc covers: A.5 final state + 3 high-value gotchas, files to read,
per-priority detail with effort estimates and acceptance criteria,
what NOT to do, the first-30-minute workflow, and the full A.5
commit chain for reference.
Phase is sized ~1 week if all three priorities land. The audit
step on Tier 1 is the highest-leverage investment.
Tier 2 + Tier 3 (static/dynamic split + GPU compute culling) are
explicitly out-of-scope for this phase — separate multi-week phases
per docs/plans/2026-05-10-perf-tiers-2-3-roadmap.md.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes ISSUES.md #13. PlayerDescriptionParser now walks the full
trailer (Options1 / Shortcuts / HotbarSpells / DesiredComps /
SpellbookFilters / Options2 / GameplayOptions blob / Inventory /
Equipped) ported from holtburger events.rs:503-625 +
shortcuts.rs:13-34. The trickiest piece — gameplay_options — uses a
4-byte-aligned forward heuristic (TryHeuristicInventoryStart)
mirroring holtburger's find_inventory_start_after_gameplay_options.
Trailer walk wrapped in its own inner try/catch so a malformed
trailer cannot destroy upstream attribute/skill/spell/enchantment
data; new Parsed.TrailerTruncated flag distinguishes clean parse
from graceful-degradation parse, with diagnostic log under
ACDREAM_DUMP_VITALS=1.
GameEventWiring registers parsed Inventory + Equipped into
ItemRepository at login (acceptance criterion: ItemRepository.Count
> 0 after login, exercised by GameEventWiringTests). 20 PD parser
tests + 1 wiring acceptance test; 282/282 AcDream.Core.Net.Tests
pass.
Plan: docs/superpowers/plans/2026-05-10-issue-13-pd-trailer.md.
Roadmap update: F.5a (visible-at-login dev panels) added as the
first deliverable that actually consumes the new trailer data —
ImGui dev panels under ACDREAM_DEVTOOLS=1 binding to
AcDream.UI.Abstractions ViewModels.
13 task commits + 1 review-followup + 1 nit-fix + 1 roadmap = 16
commits on the branch.
Sub-phase under existing F.5 (Core panels) capturing the immediate
follow-up to ISSUES.md #13: now that PlayerDescriptionParser surfaces
the full trailer (Inventory / Equipped / Shortcuts / HotbarSpells /
DesiredComps / Options1+2 / SpellbookFilters) and GameEventWiring
populates ItemRepository at login, F.5a wires that data into minimal
ImGui dev panels under ACDREAM_DEVTOOLS=1 so it's observable in-game.
Establishes the binding pattern (AcDream.UI.Abstractions ViewModels →
ImGui renderer) that the eventual D.2b retail-skinned F.5 panels
reuse. Spec to brainstorm before code.
Per docs/plans/2026-05-10-perf-tiers-2-3-roadmap.md Tier 1: cache the
per-(entity, meshRef, batch) classification (TextureCache lookup,
GroupKey hash, _groups dict insert) so the per-frame Draw inner loop
becomes "look up cache → walk assignments → append matrix to group's
Matrices list."
For static entities (~95% of world: trees, rocks, buildings, scenery),
the answer never changes between frames. Cache once at first visit;
reuse permanently. Per-frame work for static drops from 4 expensive
operations per (meshRef, batch) to 1 list-append.
Estimated entity dispatcher: 3.5ms → ~1-1.5ms median at radius=12.
Should land inside the 2.0ms spec budget.
Implementation:
- New EntityClassificationCache class (per-meshRef list of cached
(group ref, baked-PartTransform) tuples) keyed by entity.Id.
- ClassifyEntity does the one-time work; result populates _groups and
the cache.
- Draw inner loop: cache lookup → for each assignment, model =
PartTransform × entityWorld; group.Matrices.Add(model).
- Cache miss when ClassifyEntity finds NO mesh loaded yet (Vao == 0)
→ don't store; retry next frame. Avoids cache thrash during the
streaming-in window.
- Public InvalidateEntity(uint id) + ClearEntityCache() for explicit
invalidation hooks. Wiring (palette swap on ObjDescEvent, MeshRefs
hot-swap) is post-A.5 follow-up — for now, cache-stale entities
show their pre-swap appearance until next respawn.
Tier 2 (static/dynamic split with persistent groups) and Tier 3 (GPU
compute culling) tracked in the roadmap doc.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
After PlayerDescription is dispatched, the Inventory and Equipped lists
produced by the parser are now fed into ItemRepository via AddOrUpdate +
MoveItem so inventory/paperdoll panels see items after login.
Acceptance test PlayerDescription_RegistersInventoryEntries_InItemRepository
confirms ItemCount goes 0→2 for a synthetic PD with two inventory entries.
282 Net.Tests pass.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Captured 2026-05-10 during Phase A.5 polish discussion. User asked why
the 9070 XT @ 1440p doesn't hit Unreal-level FPS for an old game like
AC. Answer: architectural — we rebuild the entire draw plan from
scratch every frame instead of caching pre-baked static-world data.
Tier 1 (entity-classification cache) lands as A.5 polish (separate
commit). Tiers 2 + 3 documented here for future scheduling:
- Tier 2 — Static/dynamic split with persistent groups
~2-week phase. Static entities (~95% of world) get permanent GPU-
resident matrix slots, populated at spawn, dirty-tracked for delta
upload. Per-frame CPU cost for static = LB-cull + dirty-flag check
only. Estimated entity dispatcher: 3.5ms → 0.5-1ms median.
400-600 FPS at standstill, radius=12.
- Tier 3 — GPU-side culling (compute pre-pass)
~1-month phase. Per-instance frustum cull moves to GPU compute
shader. Compute writes draw-indirect buffer; rasterizer reads it.
Estimated CPU dispatcher: ~0.05ms (essentially free).
600-1000+ FPS at standstill, radius=12.
Doc captures effort estimates, sub-decisions, risks, mitigations, and
scheduling triggers for each tier. Also notes the architectural
ceiling (~800-1500 FPS for a C# + GL client; reaching native engine
performance requires becoming a different engine).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
T17's WalkEntities helper allocated a fresh List<(WorldEntity, int)>
per frame to hold the (entity, meshRefIndex) pairs that pass visibility
filters. At ~10K entities × ~3 mesh refs = ~30K tuples × 16 bytes =
~480 KB / frame of GC pressure on the render thread. The implementer's
self-review flagged this as a future N.6 optimization; the post-T26
diagnostic showed it materially contributing to the perf regression
(though Bug A — far-tier entity load — was the dominant factor).
Refactor: split WalkEntities into two overloads.
- WalkEntities(...) — test-friendly, allocates a fresh ToDraw list per
call. Tests keep using this signature unchanged.
- WalkEntitiesInto(..., scratch, ref result) — no-alloc, clears + populates
a caller-provided scratch list. Draw uses this with a per-dispatcher
_walkScratch field reused across frames.
Test count unchanged (40 streaming + 8 bucketing tests still pass).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Phase A.5's two-tier streaming spec promised that far-tier landblocks
ship terrain ONLY — no entities, no scenery, no interior cells. T13/T16
wired the controller side (RecenterTo emits ToLoadFar/ToLoadNear/ToPromote;
controller passes JobKind to the worker), but the worker's HandleJob
never branched on Kind: every load called BuildLandblockForStreaming
which runs the full hydration + scenery generation + interior cell path.
Result: at default radii (N₁=4 / N₂=12), 540 far-tier LBs each loaded
their full entity layer (~132 entities/LB → ~71K entities total) into
GpuWorldState. The dispatcher then walked all ~54K entities per frame
(post-frustum-cull), driving the entity dispatcher cpu_us from ~3.6ms
median (T24 baseline) to ~18-21ms (post-T22.5 horizon-test). User-
observed: 40 FPS / 25ms frame time at horizon-safe settings; system
crash at full High preset.
Minimum-diff fix: in LandblockStreamer.HandleJob, after
_loadLandblock returns, strip Entities to empty for LoadFar before
posting Loaded. Worker still does wasted hydration CPU (off the render
thread, harmless). Render-side dispatcher walk drops from ~54K to ~10K
entities/frame.
Math: post-fix entity dispatcher should drop to ~3-4ms median at N₁=4 /
N₂=12 (matches T24's 3.6ms at radius=5 single-tier, since N₁=4 has 33%
fewer near entities than N₁=5).
Future optimization (N.6 / A.6): plumb JobKind through
BuildLandblockForStreaming so the worker also skips the wasted CPU.
Out of A.5 scope.
Bug B (T17 WalkEntities allocation) is a smaller perf hit — defer if
post-Bug-A FPS is acceptable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
GameWindow.OnLoad resolves QualitySettings.From(_persistedDisplay.Quality)
+ WithEnvOverrides() immediately after LoadAndApplyPersistedSettings, stores
result in _resolvedQuality field. All six quality dimensions applied:
- NearRadius / FarRadius: replace old T16 env-var-only block; preset drives
the radii, legacy ACDREAM_STREAM_RADIUS override still honoured.
- MsaaSamples: WindowOptions.Samples reads from startup quality resolution
in Run() (pre-window-create read from SettingsStore). MSAA cannot change
at runtime; ReapplyQualityPreset logs a restart-required warning if the
new preset would change it.
- AnisotropicLevel: TerrainAtlas.SetAnisotropic() called after Build() and
again in ReapplyQualityPreset. Temporarily removes bindless residency
before the GL TexParameter call, re-makes resident after.
- AlphaToCoverage: WbDrawDispatcher.AlphaToCoverage property gates the
glEnable/glDisable(SampleAlphaToCoverage) pair around the opaque pass.
- MaxCompletionsPerFrame: set on StreamingController after construction
and after each mid-session restart.
ReapplyQualityPreset(QualityPreset) method handles mid-session changes
(Settings panel Quality dropdown Save): rebuilds streamer + controller for
radius changes, toggles A2C and aniso immediately, logs MSAA restart caveat.
onSaveDisplay callback updated to call ReapplyQualityPreset when Quality
field changes.
TerrainModernRenderer.Atlas property added to expose the atlas for
mid-session aniso updates.
991 tests passing, 8 pre-existing failures unchanged.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Add QualityPreset enum + QualitySettings readonly record struct with
From(preset) table and WithEnvOverrides() env-var override layer.
Four presets (Low/Medium/High/Ultra) drive NearRadius, FarRadius,
MsaaSamples, AnisotropicLevel, AlphaToCoverage, MaxCompletionsPerFrame.
Env vars (ACDREAM_NEAR_RADIUS, ACDREAM_FAR_RADIUS, ACDREAM_MSAA_SAMPLES,
ACDREAM_ANISOTROPIC, ACDREAM_A2C, ACDREAM_MAX_COMPLETIONS_PER_FRAME)
override individual preset fields for dev spot-testing.
DisplaySettings gains a Quality: QualityPreset field (default High);
SettingsStore persists/loads it under display."quality" as an enum
name string with Enum.TryParse fallback. 12 new QualityPresetTests
cover the preset table (radii, msaa, aniso, a2c, completions) and all
six env-var override paths. 415 UI.Abstractions tests passing.
Wiring into GameWindow / WbDrawDispatcher / TerrainAtlas follows in
commit 2 of this task.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
