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.
Records presents, GPU waits, submits, ETC2 decode, and pipeline and shader creation for ui.perfetto.dev. Adds a benchmark guide and a trace summary script.
vii