Commit Graph

3581 Commits

Author SHA1 Message Date
JosJuice
9c70036105
Merge pull request #6503 from lioncash/brace-warn
PixelShaderGen/UberShaderPixel: Silence -Wmissing-braces warnings
2018-03-23 15:29:39 +01:00
Lioncash
a52cc8d52b
PixelShaderGen/UberShaderPixel: Silence -Wmissing-braces warnings 2018-03-23 10:06:27 -04:00
Lioncash
2ab29a40eb
AbstractFramebuffer: Silence a -Wlogical-op-parentheses warning in ValidateConfig() 2018-03-23 09:58:19 -04:00
Léo Lam
f335790623
Merge pull request #6460 from lioncash/datareader
DataReader: Minor API changes
2018-03-19 15:02:50 +01:00
Markus Wick
98b4716902
Merge pull request #6442 from stenzek/async-compiler-priority
ShaderCache: Implement compile priority
2018-03-19 09:16:53 +01:00
Markus Wick
79b21e1381
Merge pull request #6459 from lioncash/enum
VertexShaderGen: Convert defines to an enum
2018-03-19 08:48:25 +01:00
Lioncash
b21a00d290 DataReader: Provide a const qualified variant of GetPointer()
This wouldn't be much of a data reader if it can't access the
read-only data pointer in read-only contexts. Especially if it
can get a writable equivalent in contexts that aren't read-only.
2018-03-18 16:53:04 -04:00
Lioncash
ce29c1c42f DataReader: Correct return type of operator=
It's questionable to not return a reference to the instance being
assigned to. It's also quite misleading in terms of expected behavior
relative to everything else. This fixes it to make it consistent with
other classes.
2018-03-18 16:50:59 -04:00
Lioncash
ffade65c55 DataReader: Remove __forceinline from trivial functions 2018-03-18 16:49:50 -04:00
Lioncash
4c2ec39199 DataReader: In-class initialize member variables where applicable
Allows the default constructor to be defaulted and ensures the default
values are associated with the member variables directly.

Also corrects a prefixed underscore in the two parameter constructor.
2018-03-18 16:34:51 -04:00
Lioncash
fd956f6c69 DataReader: Make size() and Peek() const member functions
These don't modify class state, so they can be const qualified.
2018-03-18 16:33:14 -04:00
Lioncash
2b3b1e8d09 VertexShaderGen: Convert defines to an enum
Unlike defines, these will actually obey namespacing (should one be
added), and also provide a symbol when debugging, as opposed to a magic
value.
2018-03-18 15:45:20 -04:00
Lioncash
b68e8b872e VideoBackendBase: Migrate functions from MainBase.cpp to VideoBackendBase.cpp
Given that this only contains functions from the VideoBackendBase class,
it makes more sense to move these to the relevant cpp file to keep them
all together.
2018-03-18 15:33:59 -04:00
Lioncash
13ed716419 MainBase: Remove effectively unused s_FifoShuttingDown variable
This is only ever set and cleared, which is fine, however nothing else
ever actually reads its state to perform differing behavior.
2018-03-17 18:10:43 -04:00
Lioncash
826c11ba3b MainBase: Remove unused s_beginFieldArgs struct
This is only ever memset to zero and never used again.

This also gets rid of an instance of undefined behavior considering the
draft standard for C++17 (N4659) states at [dcl.type.cv] paragraph 5:

"
The semantics of an access through a volatile glvalue are implementation-defined.
If an attempt is made to access an object defined with a volatile-qualified type
through the use of a non-volatile glvalue, the behavior is undefined.
"
2018-03-17 18:07:08 -04:00
Stenzek
93865b327f ShaderCache: Implement compile priority
Currently, when immediately compile shaders is not enabled, the
ubershaders will be placed before any specialized shaders in the compile
queue in hybrid ubershaders mode. This means that Dolphin could
potentially use the ubershaders for a longer time than it would have if
we blocked startup until all shaders were compiled, leading to a drop in
performance.
2018-03-17 01:53:11 +10:00
Lioncash
50a476c371 Assert: Uppercase assertion macros
Macros should be all upper-cased. This is also kind of a wart that's
been sticking out for quite a while now (we avoid prefixing
underscores).
2018-03-14 22:03:12 -04:00
Stenzek
517a977444 ShaderCache: Fix several issues in background shader compiling
- In D3D, shaders could be compiled on the main thread, blocking
startup.
- Reduced the latency between a pipeline being requested and used in all
backends in hybrid ubershader mode, when no shader stages were present.
- Fixed a case where async compilation could cause the same UID to be
appended multiple times to the UID cache.
- Fix incorrect number of threads being used when immediately compile
shaders was enabled.
2018-03-15 01:50:47 +10:00
Stenzek
51e4014a35 DriverDetails: Disable primitive restart on Vulkan with Mali driver 2018-03-14 02:56:24 +10:00
Stenzek
db810956ec Vulkan: Provide a more accurate method of detecting drivers/vendors
This is needed to differentiate between the open-source Mesa drivers and
their binary counterparts for Intel and AMD.
2018-03-14 02:48:53 +10:00
Henrik Rydgård
6cbb716f60 SSE-optimize UninitializeXFBMemory
Lowest hanging fruit I could find with a profiler.

Not sure this stuff actually needs to be done, but assuming it is, why
not do it quickly? 10x faster, goes from 1% CPU to 0.09%.
2018-03-13 11:55:28 +01:00
Stenzek
427aa188d4 ShaderCache: Use a version number for pipeline UID caches 2018-03-11 14:48:09 +10:00
Stenzek
e31cc1f679 ShaderCache: Implement background shader compilation
This enables shaders to be compiled while the game is starting, instead
of blocking startup. If a shader is needed before it is compiled,
emulation will block.
2018-03-10 16:11:19 +10:00
Stenzek
9fa24700b6 VideoConfig: Collapse ubershader configuration fields to a single value 2018-03-10 15:56:45 +10:00
Stenzek
590307b94c ShaderCache: Use memcmp for comparing pipeline UIDs
As these are stored in a map, operator< will become a hot function when
doing lookups, which happen every frame. std::tie generated a rather
large function here with quite a few branches.
2018-03-10 15:56:42 +10:00
Stenzek
41296db083 Renderer: Remove now-redundant Set{Rasterization,Depth,Blending}State 2018-03-10 15:56:40 +10:00
Stenzek
f9c829c7f7 OGL: Re-implement async shader compiling 2018-03-10 15:56:34 +10:00
Stenzek
dec0c3bce8 Move shader caches to VideoCommon 2018-03-10 15:56:30 +10:00
Stenzek
4c24a69710 VideoCommon: Add support for Abstract Framebuffers 2018-03-02 20:20:48 +10:00
Stenzek
2a6d9e4713 AbstractTexture: Add support for depth textures/formats 2018-03-01 17:31:24 +10:00
Stenzek
6374a4c4a8 AbstractTexture: Support multisampled abstract texture 2018-03-01 17:31:24 +10:00
Stenzek
4316f5f56b AbstractTexture: Add property/attribute accessor helpers 2018-03-01 17:31:24 +10:00
Stenzek
e125eaa237 VideoCommon: Drop references to AbstractRawTexture 2018-03-01 17:31:24 +10:00
Markus Wick
227db66e4f OGL: Use glBufferData on Mali.
tl;dr: This PR speedups dolphin on mobiles with the Mali GPU and ES 3.2
drivers by a factor of 10 by using the method with the biggest overhead.
Please keep care not to buy this shit!

The ARM driver team seems to care very well about their customers. But
bad luck, users and open source developers are *not* their customers. So
even device-independent feature requests are just ignored for *years*:

https://community.arm.com/graphics/f/discussions/4645/gl_ext_buffer_storage-support

The bad point, they neither implement any of the other common ways to
stream dynamic content in unextented GL:
- They just ignore the GL_MAP_UNSYNCHRONIZED_BIT flag
- They don't support on-device buffer updates and just stall with
glBufferSubData

It seems like no benchmark is using any dynamic content - and like no
customer cares about anything but benchmarks, or users...

We have a flag to disable the glBufferSubData way, this PR adds the flag
to also disable the unsychronized mapping way. The second one is
available since their ES 3.2 update, but slow as hell.

So how to continue? The last remaining technical way to stream dynamic
content at all is to alloc a new buffer per draw call with glBufferData.
This is very gross, but still a factor 10 speedup compared to stalling
the GPU. Small tests shows that you can expect another 3-5 times speedup
with EXT_buffer_data, so Mali would be on pair with Adreno here. So if
you have bought such a device unfortunately, please try to make noise on
your vendor forums/support and ask for this extension. If you are going
to buy a new mobile, I'd recormend to avoid *any* mobile with a Mali GPU
in it.
2018-02-25 17:12:36 +01:00
Stenzek
fec6bb4d56 VideoBackends: Add AbstractShader and AbstractPipeline classes 2018-02-22 22:02:34 +10:00
Stenzek
de632fc9c8 Renderer: Handle resize events on-demand instead of polling
We now differentiate between a resize event and surface change/destroyed
event, reducing the overhead for resizes in the Vulkan backend. It is
also now now safe to change the surface multiple times if the video thread
is lagging behind.
2018-02-20 01:15:55 +10:00
Stenzek
c1b39ecc58 BPFunctions: Move upscaling of scissor rect to VideoCommon 2018-02-20 00:49:32 +10:00
Stenzek
5359396099 BPFunctions: Move GX viewport conversion to VideoCommon 2018-02-20 00:49:32 +10:00
Stenzek
340ee8fff8 PixelShaderGen: Implement table-based fog range as in software renderer 2018-02-15 22:19:21 +10:00
Stenzek
93fb0e1e1c TextureCache: Add an option to disable EFB copies to VRAM
The option is named DisableCopyToVRAM under the Hacks section in
GFX.ini. It is intentionally not exposed to the GUI, as users should not
need to use it under normal circumstances. The main use is debugging
issues in the EFB-to-RAM shaders.
2018-02-11 15:48:46 +10:00
Stenzek
84b990faa0 VideoConfig: Remove bForceCopyToRam field
It's the inverse of supports-copy-to-vram.
2018-02-11 15:29:37 +10:00
Anthony
096131c908
Merge pull request #6334 from stenzek/startup
Video Backend Initialization/Core Boot Improvements
2018-02-07 23:35:54 -08:00
Stenzek
260d5b7aa7 BPMemory: Handle fog configuration where both A and C are infinity/NaN
The console appears to behave against standard IEEE754 specification
here, in particular around how NaNs are handled. NaNs appear to have no
effect on the result, and are treated the same as positive or negative
infinity, based on the sign bit.

However, when the result would be NaN (inf - inf, or (-inf) - (-inf)),
this results in a completely fogged color, or unfogged color
respectively. We handle this by returning a constant zero for the A
varaible, and positive or negative infinity for C depending on the sign
bits of the A and C registers. This ensures that no NaN value is passed
to the GPU in the first place, and that the result of the fog
calculation cannot be NaN.
2018-02-01 17:40:39 +10:00
Stenzek
fe5150cc31
Merge pull request #6303 from TraceBullet/auto-adjust-window-size
Fix Auto-Adjust Window Size option making the window too large
2018-01-29 17:28:44 +10:00
Stenzek
c790077c13 VideoBackend: Remove PeekMessages method
The video thread and backend no longer create any windows, therefore
there will never be any messages dispatched to their thread.
2018-01-27 13:53:55 +10:00
Stenzek
d96e8c9d76 VideoBackends: Combine Initialize/Prepare and Cleanup/Shutdown methods
Also allows the work previously done in Prepare to return a failure
status.
2018-01-27 13:53:55 +10:00
TraceBullet
ab6f932347 Fix Auto-Adjust Window Size option making the window too large 2018-01-26 10:47:19 -05:00
Stenzek
3f197480ef Renderer: Fix crash on shutdown when frame dumping or taking screenshots 2018-01-26 12:12:00 +10:00
Stenzek
81ae88d2d5 AbstractTexture: Fix crash in Vulkan backend when freeing texture 2018-01-26 19:12:11 +10:00
Stenzek
38e0b6e2ab AbstractTexture: Move Bind() method to Renderer
This makes state tracking simpler, and enables easier porting to command
lists later on.
2018-01-22 13:22:09 +10:00