Commit graph

5839 commits

Author SHA1 Message Date
Unknown W. Brackets
22e46b51c2 GPU: Centralize DestroyAllFBOs().
This cleans up the postshader update code.
2020-05-13 18:15:04 -07:00
Unknown W. Brackets
03e3a935da GPU: Cleanup presentation flipping a bit. 2020-05-13 18:11:25 -07:00
Unknown W. Brackets
884a64b932 GPU: Cleanup skip buffer texture draws.
These will be used when a game does multiple transfers from RAM to the
screen, rather than one big one.  Wasn't clearing some state, though.
2020-05-13 18:10:09 -07:00
Unknown W. Brackets
6b251dac4c GPU: Simplify direct render when skipping buffers.
This may have been uploading twice before, so there may be a small
performance improvement in some games now.
2020-05-13 18:10:09 -07:00
Unknown W. Brackets
a41fbb9225 softgpu: Fix postshader on 5551.
This also fixes rendering on Windows 7 Direct3D 11.
2020-05-13 18:10:09 -07:00
Unknown W. Brackets
2faab0e082 GPU: Use postshader for direct VRAM draws again.
Also centralize the pixel texture code while at it.
2020-05-13 18:10:09 -07:00
Unknown W. Brackets
762b656ea2 GPU: Use a texture directly for MakePixelTexture.
This makes it easier to do things with it.
2020-05-13 18:10:09 -07:00
Unknown W. Brackets
2653e50200 softgpu: Avoid RB swizzle when using a postshader.
So that it can post-process correctly.
2020-05-13 18:10:09 -07:00
Unknown W. Brackets
d4f4e87e66 Vulkan: Correct postshader sampler binding.
This was causing a lost device.
2020-05-13 18:10:09 -07:00
Unknown W. Brackets
3aa8287b74 softgpu: Enable postshader support. 2020-05-13 18:10:09 -07:00
Unknown W. Brackets
0e5b17eb9c GLES: Fix reuse of vertex data buffer.
Direct3D 11 and OpenGL had some issues with this being reused for both
renders.  Caused OpenGL to appear flipped.
2020-05-13 18:10:09 -07:00
Unknown W. Brackets
fa6544b737 GPU: Cleanup leftover postshader stuff. 2020-05-13 18:10:09 -07:00
Unknown W. Brackets
cb94487a16 GPU: Move post shader handling to new class.
Currently, Vulkan is not working properly and direct (RAM -> output) is
not hooked up.  But in general, it works.
2020-05-13 18:10:06 -07:00
Unknown W. Brackets
d17c5b30e1 D3D9: Enable postprocessing of post shaders. 2020-05-13 18:07:25 -07:00
Unknown W. Brackets
0b26eaaa7a GPU: Move a bit more to new display approach. 2020-05-13 18:07:25 -07:00
Unknown W. Brackets
57bd88fc33 softgpu: Allow display rotation. 2020-05-13 18:07:25 -07:00
Unknown W. Brackets
a03e368566 GPU: Move cardboard/etc. to PresentationCommon.
Now this works on softgpu as well.

Some hacks for backend differences...
2020-05-13 18:07:25 -07:00
Unknown W. Brackets
d39b0bdca2 GPU: Split FramebufferCommon into two classes.
Only some things moved over so far.

FramebufferCommon does too much, we want to share it with softgpu without
all the buffer management stuff.
2020-05-13 18:07:22 -07:00
Unknown W. Brackets
1b9440611a softgpu: Fix texture overlap.
Mainly happened when we had wide textures and split them up between GPUs.
2020-05-13 17:53:00 -07:00
Henrik Rydgård
009efa5c58
Merge pull request #12889 from sum2012/gpu-minor
Improved compatibility of sceGeListEnQueue: verify that stackDepth < 256
2020-05-13 23:25:56 +02:00
Unknown W. Brackets
a2b99e23d2 GPU: Handle buffer overhead in postshader uniforms.
Framebuffers may be temporarily larger than the estimated width, because
we estimate different sizes for them and resizing is expensive.

This moves accounting for that to texelDelta instead of pixelDelta.
2020-05-13 08:39:50 -07:00
sum2012
c7e58a79e3 Improved compatibility of sceGeListEnQueue: verify that stackDepth < 256
thanks gid15
2020-05-12 21:30:25 +08:00
Henrik Rydgård
c1d32e8b3e D3D11: Better set current texture to null first, otherwise we can run into another issue. 2020-05-10 23:42:52 +02:00
Henrik Rydgård
f708396bda Fix a comment 2020-05-10 23:10:57 +02:00
Henrik Rydgård
73c253e2ac D3D11: Fix a bind ordering issue in depal (only a problem with debug layer enabled) 2020-05-10 23:10:57 +02:00
Unknown W. Brackets
ba566ef26a GPU: Minor lighting cleanup in software transform.
In this case, unlitColor is already materialAmbient.
2020-05-08 23:09:24 -07:00
Unknown W. Brackets
ef43ec5f33 GPU: Split up software transform into phases. 2020-05-08 23:09:24 -07:00
Henrik Rydgård
bf6357a53a
Merge pull request #12875 from unknownbrackets/gpu-bof
GLES: Simplify the framebuf offset detection
2020-05-08 23:03:07 +02:00
Unknown W. Brackets
b413a58945 GLES: Add safety handling for offset depal.
Just in case...
2020-05-07 23:30:17 -07:00
Unknown W. Brackets
16e47f6333 GPU: Calc framebuf offset with right params.
It has nothing to do with the target or source framebuf, oops.
2020-05-07 23:22:51 -07:00
Unknown W. Brackets
4d11256807 GPU: Avoid a texture offset change in sw transform.
If we do it this late, it changes the shader, but that is tricky to handle
for all the backends.
2020-05-07 23:18:51 -07:00
Unknown W. Brackets
48980d8786 GLES: Simplify the framebuf offset detection.
It's clearly not been understood well as changes were made, so let's dump
a bunch of comments on it.
2020-05-07 23:12:14 -07:00
Henrik Rydgård
f9f568d266 Vulkan: Framebuffer manager: Use an allocator for "MakePixelTexture" images.
Fixes #12355 (or at a minimum, will improve it).
2020-05-06 22:24:10 +02:00
Henrik Rydgård
5508bfb596 Vulkan: Discard negative width/height framebuffer blits. Will likely help #12531. 2020-04-26 12:44:11 +02:00
Henrik Rydgård
5fbe369fd0
Merge pull request #12836 from unknownbrackets/d3d-depth
D3D11: Always use accurate depth
2020-04-19 14:02:50 +02:00
Unknown W. Brackets
4ea5d165f5 D3D11: Always use accurate depth.
The Direct3D APIs use a "min" and "max" value, and it's required that min
must be less than or equal to max.  If this requirement is not met, the
viewport parameters are not updated.

We forced accurate depth on for NVIDIA and AMD, but this just meant Intel
was still broken and less people were reporting bugs about it.

Although the PSP GE has depth viewport transform, this is not used for
clipping and is not actually the depth range.  The minz and maxz GE
registers are the depth range, and minz must be less than or equal to maxz
for drawing to occur.
2020-04-18 22:34:33 -07:00
Henrik Rydgård
c6cd45e933
Merge pull request #12813 from unknownbrackets/gles-state
GLES: Update shader uniforms after apply state
2020-04-12 15:21:23 +02:00
Henrik Rydgård
037d89906b
Merge pull request #12814 from unknownbrackets/debugger
GE Debugger: Prevent crash when stopping dump execution
2020-04-12 15:20:19 +02:00
Unknown W. Brackets
3b74d4b680 GE Debugger: Prevent crash when stopping execution. 2020-04-11 23:44:31 -07:00
Unknown W. Brackets
b57ba89b45 GLES: Update shader uniforms after apply state.
We may set shaderblend, projmatrix, or depthrange dirty in
ApplyDrawState().  These need to update shader uniforms to apply
correctly, but we were checking beforehand.
2020-04-11 17:40:39 -07:00
Unknown W. Brackets
a5ede2bed3 GLES: Re-enable non-float depal shader path.
Was accidentally always detected as unsupported.
2020-04-07 22:49:07 -07:00
Unknown W. Brackets
ac60e2ecd4 GPU: Track HW tess at start of frame too.
This also makes it so we don't force the setting off when you change
backends, and just ignore it if unsupported.
2020-04-04 11:52:32 -07:00
Unknown W. Brackets
46b9454e73 GPU: Reset hw transform once per frame.
Might prevent potential crashes if it changes mid-frame.
2020-04-04 11:21:22 -07:00
Unknown W. Brackets
30ede8240c GPU: Move hw transform decision to draw engine. 2020-04-04 11:14:32 -07:00
Unknown W. Brackets
c42fb72419 GPU: Update uniforms w/ consistent render mode too. 2020-04-04 11:03:07 -07:00
Unknown W. Brackets
ad98609819 GPU: Use consistent buffered rendering state.
The setting can be changed in Qt and Windows between frames, so let's
track the current setting in most places.  This is everywhere fbman is
easily accessible.
2020-04-04 10:51:47 -07:00
Henrik Rydgård
21eaef04fc Vulkan: Pick shaders after calling ConvertStateToVulkanKey because it may end up disabling shader blending
(due to too many copies, see DrawEngineCommon::ApplyShaderBlending).

(So much state leaking all over the place ... I want to redesign the
whole thing).

Typo fix in comment
2020-03-29 15:02:40 +02:00
Unknown W. Brackets
4a0109d273 GPU: Treat negative light exp same as 0.
Based on #12507 and some tests, seems like negative exponents are also
fixed to a 1.0f result.
2020-03-22 22:28:05 -07:00
Unknown W. Brackets
a95f3111cf GPU: Assume a scissor of 481x273 is a mistake.
We already do this for viewport, and only when region is still a correct
size.  Helps flicker in Everybody's Golf.
2020-03-21 23:47:50 -07:00
Unknown W. Brackets
4a8839c99d GPU: Avoid divide by zero in garbage displaylist. 2020-03-19 20:56:24 -07:00