Henrik Rydgård
b65bbaf92c
Merge pull request #21678 from nintendo424/gpu-save-first-frame
...
gpu: Preserve first-frame font render targets
2026-05-27 12:22:47 +02:00
Henrik Rydgård
2eca0123f1
Correct culling in case of some weird viewport setups
2026-05-23 14:49:17 +02:00
Henrik Rydgård
fa856fbf68
Pass the culling unit tests.
2026-05-23 14:40:15 +02:00
Henrik Rydgård
0d33cd0a65
Convert the BBOX/BJUMP culling to use the worldviewproj matrix and a screen space check.
2026-05-23 14:40:15 +02:00
Henrik Rydgård
893bc732a1
Fast culling: Use the worldviewproj matrix to avoid the need for planes
2026-05-23 14:40:15 +02:00
Henrik Rydgård
5cfbf50111
Keep updated products of view*proj and world*view*proj matrices. Use to simplify bbox culling.
2026-05-21 19:18:32 +02:00
nintendo424
e5689d1adc
gpu: Preserve first-frame font render targets
...
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.
2026-05-15 12:23:14 -04:00
Henrik Rydgård
10759dc8b9
DrawEngineCommon: Avoid using member variables as loop counters, bad for perf
2026-01-30 14:10:36 +01:00
Henrik Rydgård
7853881210
GPU header cleanup: Avoid including GPUCommon.h in some places
2025-11-23 20:37:37 +01:00
Henrik Rydgård
44d06ec42f
Add some sanity checks guided by Android crash reports
2025-06-07 16:35:19 +02:00
Henrik Rydgård
9d1f8a2575
Don't forget to update the vertex addr even when culling 2D draws
...
Fixes #20454
2025-06-06 17:53:49 +02:00
Henrik Rydgård
8ed898c5e6
GPU: Constify VertexDecoder pointers, remove in some cases
2025-05-14 15:13:59 +02:00
Henrik Rydgård
5d85d5ed22
Depth raster: Rename some variables, fix incorrect draw limit
2025-04-10 17:49:06 +02:00
Henrik Rydgård
b080f7da6b
Small depth raster refactor
2025-04-10 17:13:48 +02:00
Henrik Rydgård
c23c3e8d03
Measure the available time for "background depth raster". The idea is viable.
2024-12-29 09:45:55 +01:00
Henrik Rydgård
6d539cc5e0
More fixes
2024-12-28 23:50:42 +01:00
Henrik Rydgård
aec17d8829
Actually queue up depth draws
2024-12-28 23:21:54 +01:00
Henrik Rydgård
25d79afd47
Small rename, remove some obsolete reporting
2024-12-28 23:05:07 +01:00
Henrik Rydgård
1e8eb88bf0
Unify the ends of the two depth draw paths
2024-12-28 19:22:23 +01:00
Henrik Rydgård
3c3dfe5273
Reorganize depth raster draw state handling
2024-12-28 18:51:34 +01:00
Henrik Rydgård
2371fdfedd
Rework depth raster to fully support software transform
2024-12-21 21:22:22 +01:00
Henrik Rydgård
a344d0225f
DepthRaster: Fix bug where we used the wrong vertex count.
2024-12-21 14:28:00 +01:00
Henrik Rydgård
67078d439b
Depth raster: Switch to a SoA data layout for the screen space verts
2024-12-21 14:28:00 +01:00
Henrik Rydgård
bdb5f3a91b
Reorganize the depth vertex pipeline for future optimizations
2024-12-21 14:27:59 +01:00
Henrik Rydgård
c5ad81e3d5
Add DepthRaster.cpp/h. Rasterize depth rectangles, some triangles
2024-12-21 14:27:53 +01:00
Henrik Rydgård
df6ed8cfc9
Do some cleanup of #includes in GPU
2024-12-18 13:57:26 +01:00
Henrik Rydgård
ee297e974d
Move more code into SoftwareTransformCommon, cleanup
2024-12-18 13:57:26 +01:00
Henrik Rydgård
0b06cd1379
Actually avoid looking up the vertex decoder more than once
2024-12-17 22:42:07 +01:00
Henrik Rydgård
15b4f0c299
Lookup the decoder outside of DrawEngine
2024-12-17 19:27:01 +01:00
Henrik Rydgård
61b14df329
Delete obsolete empty function ClearTrackedVertexArrays
2024-12-17 19:15:25 +01:00
Henrik Rydgård
fe4504f612
Small simplification
2024-12-17 19:15:06 +01:00
Henrik Rydgård
a2a089b8c1
Remove the member function NormalizeVertices
2024-12-17 19:15:03 +01:00
Henrik Rydgård
a07da939ef
Some NormalizeVertices interface cleanup
2024-12-17 19:13:38 +01:00
Henrik Rydgård
fec232f8a8
GPU: Move things around a little, out of DrawEngine.
2024-12-17 19:13:38 +01:00
Henrik Rydgård
abb2558535
When overriding skinning in sw transform mode, don't write/restore member variables
2024-12-17 18:52:49 +01:00
Henrik Rydgård
20ecbfdf57
Pass the VertexDecoder pointer around
2024-12-17 18:24:26 +01:00
Henrik Rydgård
e74101a2fb
applySkinInDecode belongs in the VertexTypeID, not in the options.
2024-12-17 18:24:18 +01:00
Henrik Rydgård
190af89d9f
Plumb in a GPUCommon pointer in the draw engines. (Hopefully can get rid of this later)
2024-12-15 13:53:27 +01:00
Henrik Rydgård
11dbae3457
Remove the "DispatchFlush" mechanism, not convinced it's a win
2024-12-15 13:42:05 +01:00
Henrik Rydgård
d3b6d12067
Cull through-mode 2D draws against scissor rectangle
...
Helps texture replacement load performance in Fate Extra CCC (it does a lot
of off-screen drawing), and may help in other situations too.
2024-12-01 14:54:55 +01:00
Henrik Rydgård
858f37b8fc
Add a couple of asserts (to make things error out earlier)
2024-10-17 21:01:58 +02:00
Herman Semenov
b57dab2812
[GPU] Make static and const methods if possible
2024-04-05 17:04:31 +03:00
Henrik Rydgård
533fa14e27
Add a stat for number of verts decoded. It's quite high but not crazy for UFC 2010...
...
See #18806
2024-02-01 19:19:53 +01:00
Henrik Rydgård
db94b0b696
Pass the limit on the number of indices to generate to BuildDrawingParams.
2024-01-15 10:09:04 +01:00
Henrik Rydgård
61acce195c
Avoid decoding indices when we don't need them.
2023-12-20 14:25:19 +01:00
Henrik Rydgård
f86189c951
Show vertex decoders separately in profiles
2023-12-19 12:25:54 +01:00
Henrik Rydgård
aca3bbc9a0
DrawEngine: Remove the confusing MaxIndex accessor, replace with directly reading numDecodedVerts_
2023-12-10 11:58:47 +01:00
Henrik Rydgård
62c936babf
Flip the cull plane data around to avoid transforming each vertex multiple times.
2023-12-09 15:55:51 +01:00
Henrik Rydgård
89d8ef87ec
Use a less accurate but faster frustum cull for the general draws.
2023-12-09 15:55:51 +01:00
Henrik Rydgård
0905b6a5ad
Frustum-cull small draws
...
Some games do a poor job of culling stuff, and some transparent
sprites can be very expensive if they cause a copy.
Skipping them if outside the viewport makes sense in that case.
One example are the flame sprites in #17797 .
Additionally, we should be able to cull through-mode draws easily, this
one doesn't even try.
2023-12-09 15:55:51 +01:00
Henrik Rydgård