Each ETC2 upload created and destroyed its own staging buffer on the render
thread. Uploads now take ranges of 16 MiB blocks, reused once every range
is released; four empty blocks are kept. Profile traces gain per-frame CPU
time counters and slices for staging and driver memory calls.
Uploads decode on a background dispatcher and the submit carries a
timeline semaphore wait, so the GPU waits for the texture rather than
the render thread. The gather of the application's compressed bytes
moves there too when they are already mapped; unmapping or freeing that
memory drains the decodes still reading it.
Staged copies gain an explicit host-write barrier. A queue submission
only makes host writes from before it visible to the device, and these
writes land after the submit.
Devices without timelineSemaphore, and submits whose pNext already sizes
arrays by their semaphore counts, decode inline as before.
Blade Ball, median of three: worst vkQueueSubmit 140ms -> 1ms, p99
4.9ms -> 0.05ms, total 2534ms -> 309ms. Frame interval unchanged; its
tail is the engine's own main-thread step.
ETC2 uploads read bufferRowLength/bufferImageHeight instead of assuming
packed rows. MakeFileObject retains the path field. The main-thread pump
caches its JNI attach, but not the GetJNIEnv fallback, so a VM swap retries.
An idle launch-broker drain skips both locks.
Release the GameMode request when the daemon narrows the CPU mask; the pin
concentrates Roblox's worker pool and is re-applied every few seconds, so
ending the request is the only thing that lifts it. `gamemode: on` keeps it.
Emit texture uploads on worker threads instead of serially inside
vkQueueSubmit, in two passes so mip 0 lands before the levels scaled from it.
Decode ETC2 through a persistent pool and lower the parallel floor to one
128x128 image, so small batches stop running single-threaded.
Document all three in docs/PERFORMANCE.md.
Staging buffers stay mapped and are reused instead of allocated per upload. Box resampling precomputes column spans and copies repeated rows, with identical output.
Unset or non-numeric MOCKTAIL_SMALL_TEXTURE_UPSCALE now means 4, so the
upscale applies on every launch path. Drop the variable from the desktop
entry Exec line.
Level-0 uploads are hashed against the override store; MOCKTAIL_SMALL_TEXTURE_UPSCALE
creates colour images up to 64px N times larger, snaps block-rounded regions to the
host mip, fills every host mip from level 0 and turns image copies touching scaled
images into blits. MOCKTAIL_ADVERTISE_ETC2=0 hides the feature bit.
Hosts without native ETC2 now report ETC2/EAC support. Uploads into emulated images are decoded to RGBA8/R16 staging buffers at submit. MOCKTAIL_DISABLE_ETC2_EMULATION=1 turns it off.
vii