The 32_32_32_FLOAT format seems to be vertex-only, so it looks like there
can't be storage elements smaller than a single texel.
So, use a more precise name that can't be confused with "picture element"
(pixel) or "texture element" (texel) that represents a single logical pixel
rather than a storage block of pixels.
Switch between even and odd 16-byte element sequences along X by simply
flipping a bit rather than going to a different resolution-scaled group of
pixels, by increasing the size of the group within the constraints imposed
by tiling.
This fixes Watchman - The End is Nigh which relay on UB on boot (OOR)
They're checking in hardcoded "game:" for '/' character.
If it doesn't exist then adds 0x7FFFFFFE to start address and start searching from that location downward
`spirv-remap` is not present in modern Vulkan SDK versions, it was replaced
with the `--canonicalize-ids` pass in `spirv-opt`.
Overall, canonicalization provides a significant compression improvement,
which is important considering that currently Xenia is distributed in a ZIP
archive and contains many very similar shaders.
With normal DEFLATE compression, canonicalization reduced the size of a ZIP
with `xenia.exe` from 3.54 MB to 3.45 MB in a test done before committing.
Also disable stripping of debug information from shaders, which apparently
was among what `spirv-remap` was doing with `--do-everything`, as binding
and uniform buffer member names heavily aid in debugging in RenderDoc.
Partially integrated from #2329.
Co-authored-by: Herman S. <429230+has207@users.noreply.github.com>
Co-authored-by: Gliniak <Gliniak93@gmail.com>
Replace the `SubmissionTracker`s with new `GPUCompletionTimeline`s with a
more unified interface (using a base class), and without the internal logic
for queue ownership transfers since that idea was scrapped during the
development of the `Presenter`.
Also use this fence management logic for GPU emulation, though without
architectural reworks for now, just on the bottom level.
Still very messy, but can be cleaned up in further GPU command processor
and presenter reworks.
- Fake: No decoding
- Master: 1:1 from base version of Xenia
- Old: Master with improvements
- New: Completely new decoder with better stability (but not compability)
Co-authored-by: Herman S. <429230+has207@users.noreply.github.com>
- Added codepath for "consume only" pass. This should resolve constant clicking in Source engine games.
- Added smaller performance improvements to decoder
