Commit graph

1822 commits

Author SHA1 Message Date
bunnei
12445b476d
Merge pull request #1285 from bunnei/depth-fix
gl_rasterizer_cache: Only use depth for applicable texture formats.
2018-09-10 20:28:40 -04:00
bunnei
d884e805c5
Merge pull request #1284 from bunnei/bgra8_srgb
gl_rasterizer_cache: Implement RenderTargetFormat::BGRA8_SRGB.
2018-09-10 20:28:00 -04:00
Markus Wick
c1b8cd9058 video_core: Refactor command_processor.
Inline the WriteReg helper as it is called ~20k times per frame.
2018-09-10 22:06:16 +02:00
Markus Wick
0cfb0bacb2 video_core: Move command buffer loop.
This moves the hot loop into video_core. This refactoring shall reduce the CPU overhead of calling ProcessCommandList.
2018-09-10 22:06:13 +02:00
Markus Wick
c560043581 rasterizer: Drop unused handler.
This virtual function is called in a very hot spot, and it does nothing.

If this kind of feature is required, please be more specific and add callbacks
in the switch statement within Maxwell3D::WriteReg. There is no point in having
another switch statement within the rasterizer.
2018-09-10 22:03:10 +02:00
bunnei
4c0b1cc1ae gl_rasterizer_cache: Only use depth for applicable texture formats.
- Fixes an issue with Octopath Traveler leaving stale data here.
2018-09-10 00:50:38 -04:00
bunnei
1c34498368 gl_rasterizer_cache: Implement RenderTargetFormat::BGRA8_SRGB.
- Used by Octopath Traveler (with multiple render targets).
2018-09-10 00:37:52 -04:00
bunnei
49b15af054 gl_rasterizer: Implement multiple color attachments. 2018-09-09 22:48:28 -04:00
bunnei
e58855c7a4
Merge pull request #1268 from FernandoS27/tmml
shader_decompiler: Implemented TMML
2018-09-09 21:39:39 -04:00
FernandoS27
00131e752d Implemented TMML 2018-09-09 20:46:31 -04:00
bunnei
223ddb2008
Merge pull request #1272 from Subv/dma_2d
GPU/DMA: Partially implemented the 'enable_2d' bit in the DMA engine.
2018-09-09 19:53:17 -04:00
bunnei
fcf81147e7
Merge pull request #1280 from zero334/improvements
video_core: fixed arithmetic overflow warnings & improved code style
2018-09-09 19:51:46 -04:00
FernandoS27
073a21ac0b Implemented TXQ dimension query type, used by SMO. 2018-09-09 11:59:01 -04:00
Patrick Elsässer
64e45b04e0 video_core: fixed arithmetic overflow warnings & improved code style
- Fixed all warnings, for renderer_opengl items, which were indicating a
possible incorrect behavior from integral promotion rules and types
larger than those in which arithmetic is typically performed.
- Added const for variables where possible and meaningful.
- Added constexpr where possible.
2018-09-09 17:51:43 +02:00
FernandoS27
82a313a14c Change name of TEXQ to TXQ, in order to match NVIDIA's naming 2018-09-08 18:08:57 -04:00
Subv
fdb199290b GPU/DMA: Partially implemented the 'enable_2d' bit in the DMA engine.
When not set, this tells the GPU to only use the X size when performing a DMA copy.
This is only implemented for linear->linear and tiled->tiled copies. Conversion copies still retain the assert.

This bit is unset by some games for various purposes, and by nouveau when copying the vertex buffers.
2018-09-08 16:02:16 -05:00
bunnei
af074ee422
Merge pull request #1256 from bunnei/tex-target-support
Initial support for non-2D textures
2018-09-08 16:14:46 -04:00
bunnei
8b08cb925b gl_rasterizer: Use baseInstance instead of moving the buffer points.
This hopefully helps our cache not to redundant upload the vertex buffer.

# Conflicts:
#	src/video_core/renderer_opengl/gl_rasterizer.cpp
2018-09-08 04:05:56 -04:00
Patrick Elsässer
a8974f0556 video_core: Arithmetic overflow warning fix for gl_rasterizer (#1262)
* video_core: Arithmetic overflow fix for gl_rasterizer

- Fixed warnings, which were indicating incorrect behavior from integral
promotion rules and types larger than those in which arithmetic is
typically performed.

- Added const for variables where possible and meaningful.

* Changed the casts from C to C++ style

Changed the C-style casts to C++ casts as proposed.
Took also care about signed / unsigned behaviour.
2018-09-08 02:59:59 -04:00
bunnei
23ae7cf9db gl_rasterizer_cache: Improve accuracy of RecreateSurface for non-2D textures. 2018-09-08 02:53:39 -04:00
bunnei
fdd5c97a14 maxwell_3d: Remove assert that no longer applies. 2018-09-08 02:53:39 -04:00
bunnei
f165a85398 gl_rasterizer_cache: Partially implement several non-2D texture types. 2018-09-08 02:53:38 -04:00
bunnei
0731383124 gl_shader_decompiler: Partially implement several non-2D texture types (Subv). 2018-09-08 02:53:38 -04:00
bunnei
05f6f59ffb gl_rasterizer: Implement texture wrap mode p. 2018-09-08 02:53:38 -04:00
bunnei
ce8291f6c5 gl_rasterizer_cache: Track texture depth. 2018-09-08 02:53:38 -04:00
bunnei
9dccf7e1fa gl_rasterizer_cache: Remove impl. of FlushGLBuffer.
- Will not work for non-2d textures, and was not used anyways.
2018-09-08 02:53:37 -04:00
bunnei
030676b95d gl_rasterizer_cache: Keep track of texture type per surface. 2018-09-08 02:53:37 -04:00
bunnei
a439f7b6e1 gl_rasterizer_cache: Remove unused DownloadGLTexture. 2018-09-08 02:53:37 -04:00
bunnei
b56e5edafc gl_state: Keep track of texture target. 2018-09-08 02:53:37 -04:00
bunnei
9947c6ad59
Merge pull request #1252 from lioncash/header
video_core/CMakeLists: Add missing gl_buffer_cache.h
2018-09-06 19:19:43 -04:00
bunnei
9b50dca2bb
Merge pull request #1253 from lioncash/explicit
video_core/gl_buffer_cache: Minor tidying changes
2018-09-06 19:19:35 -04:00
bunnei
009a2cc9cc
Merge pull request #1255 from bunnei/minor-opt
gl_rasterizer: Call state.Apply only once on SetupShaders.
2018-09-06 19:19:16 -04:00
bunnei
820f646458 gl_rasterizer: Call state.Apply only once on SetupShaders. 2018-09-06 17:41:53 -04:00
bunnei
948f6c0738 gl_shader_decompiler: Implement saturate mode for IPA. 2018-09-06 17:40:03 -04:00
Lioncash
ddcdbce067 gl_buffer_cache: Default initialize member variables
Ensures that the cache always has a deterministic initial state.
2018-09-06 15:07:15 -04:00
Lioncash
8d685a29bc gl_buffer_cache: Make GetHandle() a const member function
GetHandle() internally calls GetHandle() on the stream_buffer instance,
which is a const member function, so this can be made const as well.
2018-09-06 15:07:15 -04:00
Lioncash
14230fe2af gl_buffer_cache: Remove unnecessary includes 2018-09-06 15:05:52 -04:00
Lioncash
68296d9474 gl_buffer_cache: Make constructor explicit
Implicit conversions during construction isn't desirable here.
2018-09-06 14:54:49 -04:00
Lioncash
8f4e09ba07 video_core/CMakeLists: Add missing gl_buffer_cache.h
Without this, the header file won't show up by default within IDEs such
as Visual Studio.
2018-09-06 14:49:51 -04:00
Markus Wick
a781042700 gl_shader_gen: Initialize position.
IMO the old code is fine, but nvidia raises shader compiler warnings.
Trivial fix through...
2018-09-06 13:37:50 +02:00
bunnei
77554ac773
Merge pull request #1243 from degasus/VAO_cache
gl_rasterizer: Implement a VAO cache.
2018-09-05 22:50:52 -04:00
bunnei
6f09c5b128
Merge pull request #1244 from FernandoS27/ipa
shader_decompiler: Implemented IPA Properly (Stage 1)
2018-09-05 21:20:40 -04:00
FernandoS27
e63b229f4a Implemented IPA Properly 2018-09-05 20:15:47 -04:00
Markus Wick
7f15306f78 gl_rasterizer: Skip TODO log.
This is called ~3k times per frame in SMO ingame.
My laptop spends ~3ms per frame on allocating and freeing this string.

Let's just stop printing this kind of redundant information.
2018-09-05 20:20:20 +02:00
Markus Wick
d3ad9469a1 gl_rasterizer: Implement a VAO cache.
This patch caches VAO objects instead of re-emiting all pointers per draw call.
Configuring this pointers is known as a fast task, but it yields too many GL
calls. So for better performance, just bind the VAO instead of 16 pointers.
2018-09-05 18:46:35 +02:00
Markus Wick
50a806ea67 renderer_opengl: Implement a buffer cache.
The idea of this cache is to avoid redundant uploads. So we are going
to cache the uploaded buffers within the stream_buffer and just reuse
the old pointers.
The next step is to implement a VBO cache on GPU memory, but for now,
I want to check the overhead of the cache management. Fetching the
buffer over PCI-E should be quite fast.
2018-09-05 08:03:50 +02:00
Markus Wick
99a71580c4 gl_shader_cache: Use an u32 for the binding point cache.
The std::string generation with its malloc and free requirement
was a noticeable overhead. Also switch to an ordered_map to
avoid the std::hash call. As those maps usually have a size of
two elements, the lookup time shall not matter.
2018-09-04 21:04:41 +02:00
bunnei
9a07e9f805
Merge pull request #1237 from degasus/optimizations
Optimizations
2018-09-04 12:16:06 -04:00
bunnei
26e96d16d0
Merge pull request #1232 from lioncash/copy
gl_shader_decompiler: Use used_shaders member variable directly within GenerateDeclarations()
2018-09-04 11:52:25 -04:00
Markus Wick
2081ed7db2 command_processor: Use std::array for bound_engines.
subchannel is a 3 bit field. So there must not be more than 8 bound engines.
And using a hashmap for up to 8 values is a bit overpowered.
2018-09-04 14:10:05 +02:00