Commit graph

2976 commits

Author SHA1 Message Date
Fernando Sahmkow 6b0695b3cd Deglobalize Memory Manager on texture cahe and Implement Invalidation and Flushing using GPUVAddr 2019-06-20 21:36:11 -03:00
ReinUsesLisp 6c410104f4 texture_cache: Remove execution context copies from the texture cache
This is done to simplify the OpenGL implementation, it is needed for
Vulkan.
2019-06-20 21:36:11 -03:00
ReinUsesLisp fa59a7b4d8 gl_texture_cache: Implement fermi copies 2019-06-20 21:36:11 -03:00
ReinUsesLisp 1b4503c571 texture_cache: Split texture cache into different files 2019-06-20 21:36:11 -03:00
ReinUsesLisp 5f3aacdc37 texture_cache: Move staging buffer into a generic implementation 2019-06-20 21:36:11 -03:00
ReinUsesLisp 2787a0c287 texture_cache: Flush 3D textures in the order they are drawn 2019-06-20 21:36:11 -03:00
ReinUsesLisp 4b396f375c gl_texture_cache: Minor changes 2019-06-20 21:36:11 -03:00
ReinUsesLisp 0cefb7bcb4 gl_texture_cache: Add copy from multiple overlaps into a single surface 2019-06-20 21:36:11 -03:00
ReinUsesLisp 84139586c9 gl_texture_cache: Attach surface textures instead of views 2019-06-20 21:36:11 -03:00
ReinUsesLisp fb94871791 gl_texture_cache: Add fast copy path 2019-06-20 21:36:11 -03:00
ReinUsesLisp bab21e8cb3 gl_texture_cache: Initial implementation 2019-06-20 21:36:11 -03:00
bunnei c28694d907
Merge pull request #2591 from lioncash/record
core: Remove unused CiTrace source files
2019-06-19 22:28:26 -04:00
Lioncash 61d2498f00 core: Remove unused CiTrace source files
These source files have been unused for the entire lifecycle of the
project. They're a hold-over from Citra and only add to the build time
of the project, so they can be removed.

There's also likely no way this would ever work in yuzu in its current
form without revamping quite a bit of it, given how different the GPU on
the Switch is compared to the 3DS.
2019-06-18 16:57:59 -04:00
bunnei c7b5c245e1
Merge pull request #2562 from ReinUsesLisp/split-cbuf-upload
video_core/engines: Move ConstBufferInfo out of Maxwell3D
2019-06-17 22:35:04 -04:00
Zach Hilman c0e7b91145
Merge pull request #2538 from ReinUsesLisp/ssy-pbk
shader: Split SSY and PBK stack
2019-06-15 20:30:13 -04:00
bunnei f981efdf8d
Merge pull request #2572 from FernandoS27/gpu-mem
GPUVM: Correct GPU VM virtual address space
2019-06-11 21:09:57 -04:00
Fernando Sahmkow f79823fda7 GPUVM: Correct GPU VM virtual address space 2019-06-09 17:47:15 -04:00
ReinUsesLisp 528c15051c kepler_compute: Use std::array for cbuf info 2019-06-07 20:36:22 -03:00
ReinUsesLisp 17d5fb6d06 kepler_compute: Fix block_dim_x encoding 2019-06-07 20:35:46 -03:00
ReinUsesLisp 5669ff3cbd gl_rasterizer: Remove unused parameters in descriptor uploads 2019-06-07 19:52:16 -03:00
ReinUsesLisp 2f2a61887a video_core/engines: Move ConstBufferInfo out of Maxwell3D 2019-06-07 19:47:15 -03:00
Zach Hilman de33ad25f5
Merge pull request #2514 from ReinUsesLisp/opengl-compat
video_core: Drop OpenGL core in favor of OpenGL compatibility
2019-06-07 17:23:25 -04:00
ReinUsesLisp fe8e6618f2 shader: Split SSY and PBK stack
Hardware testing revealed that SSY and PBK push to a different stack,
allowing code like this:

        SSY label1;
        PBK label2;
        SYNC;
label1: PBK;
label2: EXIT;
2019-06-07 02:18:27 -03:00
ReinUsesLisp 769a50661a shader/node: Minor changes
Reflect std::shared_ptr nature of Node on initializers and remove
constant members in nodes.

Add some commentaries.
2019-06-06 20:03:33 -03:00
ReinUsesLisp e1b3be7ced shader: Move Node declarations out of the shader IR header
Analysis passes do not have a good reason to depend on shader_ir.h to
work on top of nodes. This splits node-related declarations to their own
file and leaves the IR in shader_ir.h
2019-06-06 20:02:37 -03:00
ReinUsesLisp bf4dfb3ad4 shader: Use shared_ptr to store nodes and move initialization to file
Instead of having a vector of unique_ptr stored in a vector and
returning star pointers to this, use shared_ptr. While changing
initialization code, move it to a separate file when possible.

This is a first step to allow code analysis and node generation beyond
the ShaderIR class.
2019-06-05 20:41:52 -03:00
bunnei a20ba09bfd
Merge pull request #2520 from ReinUsesLisp/vulkan-refresh
vk_device,vk_shader_decompiler: Miscellaneous changes
2019-06-05 18:10:00 -04:00
bunnei 55c5029171
Merge pull request #2540 from ReinUsesLisp/remove-guest-position
gl_shader_decompiler: Remove guest "position" varying
2019-06-05 18:07:23 -04:00
bunnei 0bcc305797
Merge pull request #2512 from ReinUsesLisp/comp-indexing
gl_shader_decompiler: Pessimize uniform buffer access on AMD's prorpietary driver
2019-06-05 18:02:30 -04:00
Zach Hilman 81e09bb121
Merge pull request #2545 from lioncash/timing
core/core_timing_util: Use std::chrono types for specifying time units
2019-06-05 15:52:37 -04:00
Zach Hilman dd4fe0dab1
Merge pull request #2534 from ReinUsesLisp/shader-cleanup
gl_shader_cache: Minor style changes
2019-06-05 15:28:34 -04:00
Lioncash 42f5fd0ab3 core/core_timing_util: Use std::chrono types for specifying time units
Makes the interface more type-safe and consistent in terms of return
values.
2019-06-04 20:31:24 -04:00
Fernando Sahmkow a32c52b1d8 shader_bytecode: Mark EXIT as flow instruction 2019-06-04 12:18:35 -04:00
ReinUsesLisp 0935c2d97b gl_shader_decompiler: Remove guest "position" varying
"position" was being written but not read anywhere besides geometry
shaders, where it had the same value as gl_Position.

This commit replaces "position" with gl_Position, reducing the
complexity of our code and the emitted GLSL code.
2019-06-03 01:01:34 -03:00
ReinUsesLisp e72b9044a0 gl_shader_cache: Store a system class and drop global accessors 2019-05-30 14:01:40 -03:00
ReinUsesLisp ad321564ed gl_shader_cache: Add commentaries explaining the intention in shaders creation 2019-05-30 13:58:38 -03:00
ReinUsesLisp 838b6d2ff8 gl_shader_cache: Flip if condition in GetStageProgram to reduce indentation 2019-05-30 13:56:03 -03:00
ReinUsesLisp 6ac4490751 gl_buffer_cache: Remove unused ReserveMemory method 2019-05-30 13:21:01 -03:00
ReinUsesLisp a89cc0bafc maxwell_to_gl: Use GL_CLAMP to emulate Clamp wrap mode 2019-05-30 13:21:01 -03:00
ReinUsesLisp b76df62c00 gl_rasterizer: Move alpha testing to the OpenGL pipeline
Removes the alpha testing code from each fragment shader invocation.
2019-05-30 13:21:01 -03:00
ReinUsesLisp df509486c4 gl_rasterizer: Use GL_QUADS to emulate quads rendering 2019-05-30 13:21:01 -03:00
bunnei e3608578e4
Merge pull request #2446 from ReinUsesLisp/tid
shader: Implement S2R Tid{XYZ} and CtaId{XYZ}
2019-05-29 12:21:17 -04:00
ReinUsesLisp 21c0b4dec8 gl_device: Add commentary to AOFFI unit test source code
The intention behind this commit is to hint someone inspecting an
apitrace dump to ignore this ill-formed GLSL code.
2019-05-27 00:55:57 -03:00
ReinUsesLisp 84928e6d67 gl_shader_gen: Always declare extensions after the version declaration
This addresses a bug on geometry shaders where code was being written
before all #extension declarations were done. Ref to #2523
2019-05-27 00:51:35 -03:00
ReinUsesLisp f424b46036 vk_device: Let formats array type be deduced 2019-05-26 03:09:06 -03:00
ReinUsesLisp a4c5e3e339 vk_shader_decompiler: Misc fixes
Fix missing OpSelectionMerge instruction. This caused devices loses on
most hardware, Intel didn't care.

Fix [-1;1] -> [0;1] depth conversions.

Conditionally use VK_EXT_scalar_block_layout. This allows us to use
non-std140 layouts on UBOs.

Update external Vulkan headers.
2019-05-26 01:48:04 -03:00
ReinUsesLisp dec3c981d0 vk_device: Enable features when available and misc changes
Keeps track of native ASTC support, VK_EXT_scalar_block_layout
availability and SSBO range.

Check for independentBlend and vertexPipelineStorageAndAtomics as a
required feature. Always enable it.

Use vk::to_string format to log Vulkan enums.

Style changes.
2019-05-26 01:41:34 -03:00
Lioncash 5a4564bd8e renderer_opengl/utils: Use a std::string_view with LabelGLObject()
Uses a std::string_view instead of a std::string, given the pointed to
string isn't modified and is only used in a formatting operation.

This is nice because a few usages directly supply a string literal to
the function, allowing these usages to otherwise not heap allocate,
unlike the std::string overloads.

While we're at it, we can combine the address formatting into a single
formatting call.
2019-05-24 23:50:10 -04:00
bunnei 68c9c9222d
Merge pull request #2358 from ReinUsesLisp/parallel-shader
gl_shader_cache: Use shared contexts to build shaders in parallel at boot
2019-05-24 22:42:08 -04:00
bunnei 1a2d90ab09
Merge pull request #2485 from ReinUsesLisp/generic-memory
shader/memory: Implement generic memory stores and loads (ST and LD)
2019-05-24 18:24:26 -04:00
ReinUsesLisp d8827b07b5 gl_shader_decompiler: Use an if based cbuf indexing for broken drivers
The following code is broken on AMD's proprietary GLSL compiler:
```glsl
uint idx = ...;
vec4 values = ...;
float some_value = values[idx & 3];
```

It index the wrong components, to fix this the following pessimized code
is emitted when that bug is present:
```glsl
uint idx = ...;
vec4 values = ...;
float some_value;
if ((idx & 3) == 0) some_value = values.x;
if ((idx & 3) == 1) some_value = values.y;
if ((idx & 3) == 2) some_value = values.z;
if ((idx & 3) == 3) some_value = values.w;
```
2019-05-24 02:47:56 -03:00
ReinUsesLisp 46177901b8 gl_device: Add test to detect broken component indexing
Component indexing on AMD's proprietary driver is broken. This commit adds
a test to detect when we are on a driver that can't successfully manage
component indexing.

It dispatches a dummy draw with just one vertex shader that writes to an
indexed SSBO from the GPU with data sent through uniforms, it then reads
that data from the CPU and compares the expected output.
2019-05-24 02:47:56 -03:00
Lioncash b6dcb1ae4d shader/shader_ir: Make Comment() take a std::string by value
This allows for forming comment nodes without making unnecessary copies
of the std::string instance.

e.g. previously:

Comment(fmt::format("Base address is c[0x{:x}][0x{:x}]",
        cbuf->GetIndex(), cbuf_offset));

Would result in a copy of the string being created, as CommentNode()
takes a std::string by value (a const ref passed to a value parameter
results in a copy).

Now, only one instance of the string is ever moved around. (fmt::format
returns a std::string, and since it's returned from a function by value,
this is a prvalue (which can be treated like an rvalue), so it's moved
into Comment's string parameter), we then move it into the CommentNode
constructor, which then moves the string into its member variable).
2019-05-23 03:01:55 -03:00
Lioncash 228e58d0a5 shader/decode/*: Add missing newline to files lacking them
Keeps the shader code file endings consistent.
2019-05-23 02:55:52 -03:00
Lioncash 87b4c1ac5e shader/decode/*: Eliminate indirect inclusions
Amends cases where we were using things that were indirectly being
satisfied through other headers. This way, if those headers change and
eliminate dependencies on other headers in the future, we don't have
cascading compilation errors.
2019-05-23 02:55:52 -03:00
Lioncash 195b54602f shader/decode/memory: Remove left in debug pragma 2019-05-22 17:08:50 -04:00
Lioncash de23847184 renderer_opengl/gl_shader_decompiler: Remove redundant name specification in format string
This accidentally slipped through a rebase.
2019-05-21 09:47:21 -04:00
ReinUsesLisp 69215b5a55 gl_shader_cache: Fix clang strict standard build issues 2019-05-20 22:46:05 -03:00
ReinUsesLisp c03b8c4c19 gl_shader_cache: Use shared contexts to build shaders in parallel 2019-05-20 22:45:55 -03:00
ReinUsesLisp 75e7b45d69 shader/memory: Implement ST (generic memory) 2019-05-20 22:41:53 -03:00
ReinUsesLisp f78ef617b6 shader/memory: Implement LD (generic memory) 2019-05-20 22:38:59 -03:00
bunnei 9a17b20896
Merge pull request #2494 from lioncash/shader-text
gl_shader_decompiler: Add AddLine() overloads with single function that forwards to libfmt
2019-05-20 20:42:40 -04:00
ReinUsesLisp 9c3461604c shader: Implement S2R Tid{XYZ} and CtaId{XYZ} 2019-05-20 16:36:49 -03:00
ReinUsesLisp ada79fa8ad gl_shader_decompiler: Make GetSwizzle constexpr 2019-05-20 16:36:48 -03:00
Lioncash 58a0c13e34 gl_shader_decompiler: Tidy up minor remaining cases of unnecessary std::string concatenation 2019-05-20 14:14:48 -04:00
Lioncash 6fb29764d6 gl_shader_decompiler: Replace individual overloads with the fmt-based one
Gets rid of the need to special-case brace handling depending on the
overload used, and makes it consistent across the board with how fmt
handles them.

Strings with compile-time deducible strings are directly forwarded to
std::string's constructor, so we don't need to worry about the
performance difference here, as it'll be identical.
2019-05-20 14:14:48 -04:00
Lioncash 784d2b6c3d gl_shader_decompiler: Utilize fmt overload of AddLine() where applicable 2019-05-20 14:14:44 -04:00
Fernando Sahmkow 911fafb967 Revert #2466
This reverts a tested behavior on delay slots not exiting if the exit 
flag is set. Currently new tests are required in order to ensure this 
behavior.
2019-05-19 16:04:44 -04:00
Lioncash 91ec251c4a gl_shader_decompiler: Add AddLine() overload that forwards to fmt
In a lot of places throughout the decompiler, string concatenation via
operator+ is used quite heavily. This is usually fine, when not heavily
used, but when used extensively, can be a problem. operator+ creates an
entirely new heap allocated temporary string and given we perform
expressions like:

std::string thing = a + b + c + d;

this ends up with a lot of unnecessary temporary strings being created
and discarded, which kind of thrashes the heap more than we need to.
Given we utilize fmt in some AddLine calls, we can make this a part of
the ShaderWriter's API. We can make an overload that simply acts as a
passthrough to fmt.

This way, whenever things need to be appended to a string, the operation
can be done via a single string formatting operation instead of
discarding numerous temporary strings. This also has the benefit of
making the strings themselves look nicer and makes it easier to spot
errors in them.
2019-05-19 14:12:20 -04:00
bunnei d49efbfb4a
Merge pull request #2441 from ReinUsesLisp/al2p
shader: Implement AL2P and ALD.PHYS
2019-05-19 14:02:58 -04:00
Hexagon12 b94b08fa6f
Merge pull request #2491 from FernandoS27/dma-fix
Dma_pusher: ASSERT on empty command_list
2019-05-19 16:27:15 +01:00
Hexagon12 f8b1e53369
Merge pull request #2452 from FernandoS27/raster-cache-fix
Correct possible error on Rasterizer Caches
2019-05-19 16:00:44 +01:00
Hexagon12 2aebbe9bf9
Merge pull request #2497 from lioncash/shader-ir
shader/shader_ir: Minor changes
2019-05-19 15:51:06 +01:00
Hexagon12 fadf66993c
Merge pull request #2495 from lioncash/cache
gl_shader_disk_cache: Minor cleanup
2019-05-19 15:50:23 +01:00
Fernando Sahmkow 9e98100c94 Dma_pusher: ASSERT on empty command_list
This is a measure to avoid crashes on command list reading as an empty
command_list is considered a NOP.
2019-05-19 10:48:31 -04:00
Hexagon12 18cdbdafa2
Merge pull request #2467 from lioncash/move
video_core/gpu_thread: Remove redundant copy constructor for CommandDataContainer
2019-05-19 15:20:37 +01:00
Hexagon12 9175bffbdb
Merge pull request #2466 from yuzu-emu/mme-exit-delay-slot
GPU/MMEInterpreter: Ignore the 'exit' flag when it's executed inside a delay slot.
2019-05-19 15:14:41 +01:00
Hexagon12 ac3775e6ae
Merge pull request #2468 from lioncash/deduction
yuzu: Remove explicit types from locks where applicable
2019-05-19 15:05:56 +01:00
Hexagon12 b54bd3f018
Merge pull request #2472 from FernandoS27/tic
maxwell_3d: reduce severity of different component formats assert.
2019-05-19 15:04:47 +01:00
Hexagon12 3bd5f01240
Merge pull request #2469 from lioncash/copyable
video_core/engines/maxwell_3d: Add is_trivially_copyable_v check for Regs
2019-05-19 15:02:17 +01:00
Sebastian Valle a6ed792ac4
Merge pull request #2470 from lioncash/ranged-for
video_core/engines/maxwell_3d: Simplify for loops into ranged for loops within InitializeRegisterDefaults()
2019-05-19 09:01:19 -05:00
Hexagon12 4452195d41
Merge pull request #2480 from ReinUsesLisp/fix-quads
gl_rasterizer: Pass the right number of array quad vertices count
2019-05-19 14:58:49 +01:00
Hexagon12 8e9a1e4249
Merge pull request #2483 from ReinUsesLisp/fix-point-size
gl_rasterizer: Limit OpenGL point size to a minimum of 1
2019-05-19 14:57:05 +01:00
Sebastian Valle dfddb12255
Merge pull request #2471 from lioncash/engine-upload
video_core/engines/engine_upload: Minor tidying
2019-05-19 08:54:42 -05:00
Sebastian Valle f9ad88f9d7
Merge pull request #2484 from ReinUsesLisp/triangle-fan
maxwell_to_gl: Add TriangleFan primitive topology
2019-05-19 08:53:29 -05:00
Lioncash e310d943b8 shader/shader_ir: Remove unnecessary inline specifiers
constexpr internally links by default, so the inline specifier is
unnecessary.
2019-05-19 08:23:15 -04:00
Lioncash 212b148923 shader/shader_ir: Simplify constructors for OperationNode
Many of these constructors don't even need to be templated. The only
ones that need to be templated are the ones that actually make use of
the parameter pack.

Even then, since std::vector accepts an initializer list, we can supply
the parameter pack directly to it instead of creating our own copy of
the list, then copying it again into the std::vector.
2019-05-19 08:23:14 -04:00
Lioncash 81e7e63080 shader/shader_ir: Remove unnecessary template parameter packs from Operation() overloads where applicable
These overloads don't actually make use of the parameter pack, so they
can be turned into regular non-template function overloads.
2019-05-19 08:23:14 -04:00
Lioncash e09ee0ff23 shader/shader_ir: Mark tracking functions as const member functions
These don't actually modify instance state, so they can be marked as
const member functions
2019-05-19 08:23:09 -04:00
Lioncash ce04ab38bb shader/shader_ir: Place implementations of constructor and destructor in cpp file
Given the class contains quite a lot of non-trivial types, place the
constructor and destructor within the cpp file to avoid inlining
construction and destruction code everywhere the class is used.
2019-05-19 04:02:02 -04:00
Lioncash 3356ea5bc2 gl_shader_gen: std::move objects where applicable
Avoids performing copies into the pair being returned. Instead, we can
just move the resources into the pair, avoiding the need to make copies
of both the std::string and ShaderEntries struct.
2019-05-19 03:46:54 -04:00
Lioncash 0a7f09a99b gl_shader_disk_cache: in-class initialize virtual file offset of ShaderDiskCacheOpenGL
Given the offset is assigned a fixed value in the constructor, we can
just assign it directly and get rid of the need to write the name of the
variable again in the constructor initializer list.
2019-05-19 02:55:18 -04:00
Lioncash 634b78a4c6 gl_shader_disk_cache: Default ShaderDiskCacheOpenGL's destructor in the cpp file
Given the disk shader cache contains non-trivial types, we should
default it in the cpp file in order to prevent inlining of the
complex destruction logic.
2019-05-19 02:50:50 -04:00
Lioncash 7fdc644c44 gl_shader_disk_cache: Make hash specializations noexcept
The standard library expects hash specializations that don't throw
exceptions. Make this explicit in the type to allow selection of better
code paths if possible in implementations.
2019-05-19 02:46:45 -04:00
Lioncash 683c4e523f gl_shader_disk_cache: Remove redundant code string construction in LoadDecompiledEntry()
We don't need to load the code into a vector and then construct a string
over the data. We can just create a string with the necessary size ahead
of time, and read the data directly into it, getting rid of an
unnecessary heap allocation.
2019-05-19 02:46:44 -04:00
Lioncash 5e4c227608 gl_shader_disk_cache: Make variable non-const in decompiled entry case
std::move does nothing when applied to a const variable. Resources can't
be moved if the object is immutable. With this change, we don't end up
making several unnecessary heap allocations and copies.
2019-05-19 02:46:44 -04:00
Lioncash f417be9d3b gl_shader_disk_cache: Special-case boolean handling
Booleans don't have a guaranteed size, but we still want to have them
integrate into the disk cache system without needing to actually use a
different type. We can do this by supplying non-template overloads for
the bool type.

Non-template overloads always have precedence during function
resolution, so this is safe to provide.

This gets rid of the need to smatter ternary conditionals, as well as
the need to use u8 types to store the value in.
2019-05-19 02:46:38 -04:00
ReinUsesLisp 21ea8b2fcb gl_rasterizer: Limit OpenGL point size to a minimum of 1 2019-05-18 03:07:29 -03:00
ReinUsesLisp 52340c3294 maxwell_to_gl: Add TriangleFan primitive topology 2019-05-17 19:58:02 -03:00
ReinUsesLisp a652e58c54 gl_rasterizer: Pass the right number of array quad vertices count 2019-05-17 17:08:34 -03:00