After the previous commit, the optimization is no longer needed, and
unnecessarily bloats the uniform buffer.
This reverts commit 62bcb1785e809649ba88b49e24c5217f457fa040.
Some games render HLE sceFont glyphs into small VRAM-backed framebuffers and immediately sample them as textures. This needs both the source glyph buffer and the temporary render target to be preserved before later commands overwrite or rebind them.
Save first-frame data from the previous render target before creating a new framebuffer, since ResizeFramebufFBO() makes the new framebuffer current. Also flush pending draws before PGF writes over a glyph buffer that may still be referenced by queued texture draws.
Backend software transform normally reports safe framebuffer size during draw flush, but first-frame readback can happen before queued draws reach that path. Estimate through-mode rectangle/triangle bounds before SubmitPrim() and feed those bounds to SetSafeSize(), using actual vertices clamped to scissor instead of a tiny-target heuristic.
Fixes missing text in Evangelion JO.
Reported by Joseph on Discord.
Sometimes, things could align perfectly so allocations happend exactly
at the end of a pushbuffer. At the same time, we allow our vertex decoder to write an extra few
bytes if it needs to for speed. Unfortunately I missed this interaction,
resulting in some uncommon crashes that were especially common with
heavy-geometry things like modified GTA LCS with PS2 assets, for
example.
The problem was reported with Vulkan, but our OpenGL backend had the
same issue too.
Fix#21363https://chat.deepseek.com/share/980xe4c2ue35te8cy2
Performance impact is minimal because sprites that require depth/fog are relatively rare, and the fallback still uses the per-pixel rectangle loop (not the full triangle rasterizer)
Remove two unused helper functions that were left behind after refactoring:
1. unpackUVec3: Left over from commit 4329aaa31c ("GPU: Apply color
test mask as a uint") which refactored color test to use
roundAndScaleTo8x4 with direct uint comparison instead of unpacking
to uvec3 components.
2. packFloatsTo8x4: Left over from commit aec22491fe ("Don't expand
alphaColorRef to 128 bytes") which changed u_alphacolorref from vec4
to uint, making the float-to-uint packing function unnecessary.
Both functions were defined but never called, wasting shader compilation
time and binary size for shaders that enable color testing.
Signed-off-by: Chris Healy <cphealy@gmail.com>
Fix block transfer clipping that happened due to code not taking block
transfer offsets into account when finding framebuffers for block
transfer and when creating ram framebuffer.
The game actually implements proper shadow mapping using some wicked
trickery, including framebuffer color reinterpretation. This finally
gets it working, by filling in some gaps in the existing mechanisms.
Unfortunately there's some precision issue that needs to be looked at,
as it can currently look quite stripy, very typical of shadow mapping.
Henrik Rydgård