Works around a bug where program parameters are only applied to the
current stage, and this one wasn't bound at the moment.
Affects all SSBO usages on GLASM.
Move code to separate files to be able to reuse it from OpenGL. This
greatly simplifies the pipeline cache logic on Vulkan.
Transform feedback state is not yet abstracted and it's still
intrusively stored inside vk_pipeline_cache. It will be moved when
needed on OpenGL.
There's an optimization bug on non-git mesa versions where not
specifying GL_CLIENT_STORAGE_BIT causes very slow reads on the CPU
side.
Add this bit for all vendors.
Removes common_sizes.h in favor of having `_KiB`, `_MiB`, `_GiB`, etc
user-literals within literals.h.
To keep the global namespace clean, users will have to use:
```
using namespace Common::Literals;
```
to access these literals.
Use its std::stop_token to abort shader cache loading.
Using std::stop_token instead of std::atomic_bool allows the usage of
other utilities like std::stop_callback.
These changes should help in reducing crashes/drivers panics that may
occur due to synchronization issues between the shader completion and
later access of the decoded texture.
* common: fs: fs_types: Create filesystem types
Contains various filesystem types used by the Common::FS library
* common: fs: fs_util: Add std::string to std::u8string conversion utility
* common: fs: path_util: Add utlity functions for paths
Contains various utility functions for getting or manipulating filesystem paths used by the Common::FS library
* common: fs: file: Rewrite the IOFile implementation
* common: fs: Reimplement Common::FS library using std::filesystem
* common: fs: fs_paths: Add fs_paths to replace common_paths
* common: fs: path_util: Add the rest of the path functions
* common: Remove the previous Common::FS implementation
* general: Remove unused fs includes
* string_util: Remove unused function and include
* nvidia_flags: Migrate to the new Common::FS library
* settings: Migrate to the new Common::FS library
* logging: backend: Migrate to the new Common::FS library
* core: Migrate to the new Common::FS library
* perf_stats: Migrate to the new Common::FS library
* reporter: Migrate to the new Common::FS library
* telemetry_session: Migrate to the new Common::FS library
* key_manager: Migrate to the new Common::FS library
* bis_factory: Migrate to the new Common::FS library
* registered_cache: Migrate to the new Common::FS library
* xts_archive: Migrate to the new Common::FS library
* service: acc: Migrate to the new Common::FS library
* applets/profile: Migrate to the new Common::FS library
* applets/web: Migrate to the new Common::FS library
* service: filesystem: Migrate to the new Common::FS library
* loader: Migrate to the new Common::FS library
* gl_shader_disk_cache: Migrate to the new Common::FS library
* nsight_aftermath_tracker: Migrate to the new Common::FS library
* vulkan_library: Migrate to the new Common::FS library
* configure_debug: Migrate to the new Common::FS library
* game_list_worker: Migrate to the new Common::FS library
* config: Migrate to the new Common::FS library
* configure_filesystem: Migrate to the new Common::FS library
* configure_per_game_addons: Migrate to the new Common::FS library
* configure_profile_manager: Migrate to the new Common::FS library
* configure_ui: Migrate to the new Common::FS library
* input_profiles: Migrate to the new Common::FS library
* yuzu_cmd: config: Migrate to the new Common::FS library
* yuzu_cmd: Migrate to the new Common::FS library
* vfs_real: Migrate to the new Common::FS library
* vfs: Migrate to the new Common::FS library
* vfs_libzip: Migrate to the new Common::FS library
* service: bcat: Migrate to the new Common::FS library
* yuzu: main: Migrate to the new Common::FS library
* vfs_real: Delete the contents of an existing file in CreateFile
Current usages of CreateFile expect to delete the contents of an existing file, retain this behavior for now.
* input_profiles: Don't iterate the input profile dir if it does not exist
Silences an error produced in the log if the directory does not exist.
* game_list_worker: Skip parsing file if the returned VfsFile is nullptr
Prevents crashes in GetLoader when the virtual file is nullptr
* common: fs: Validate paths for path length
* service: filesystem: Open the mod load directory as read only
The FPS counter was based on metrics in the nvdisp swapbuffers call. This metric would be accurate if the gpu thread/renderer were synchronous with the nvdisp service, but that's no longer the case.
This commit moves the frame counting responsibility onto the concrete renderers after their frame draw calls. Resulting in more meaningful metrics.
The displayed FPS is now made up of the average framerate between the previous and most recent update, in order to avoid distracting FPS counter updates when framerate is oscillating between close values.
The status bar update frequency was also changed from 2 seconds to 500ms.
Currently, the Windows versions of the Intel OpenGL driver and the AMD
proprietary OpenGL driver do not properly support (or in fact degrade)
when asynchronous shader compilation is enabled. This blocks
specifically those drivers from using this feature. This affects
AMDGPU-PRO on Linux, and AMD's and Intel's OpenGL drivers on Windows.
ASTC texture decoding is currently handled by a CPU decoder for GPU's without native ASTC decoding support (most desktop GPUs). This is the cause for noticeable performance degradation in titles which use the format extensively.
This commit adds support to accelerate ASTC decoding using a compute shader on OpenGL for GPUs without native support.
In order to force the BGRA8 conversion on Nvidia using OpenGL, we need to forbid texture copies and views with other formats.
This commit also adds a boolean relating to this, as this needs to be done only for the OpenGL api, Vulkan must remain unchanged.
OpenGL does not natively support BGR internal formats, which causes many BGR textures to render incorrectly, with Red and Blue channels swapped.
This commit aims to address this by swizzling the blue and red channels on texture copies when a BGR format is encountered.
Some games benefit from skipping caches (Pokémon Sword), and others
don't (Animal Crossing: New Horizons). Add an heuristic to decide this
at runtime.
The cache hit ratio has to be ~98% or better to not skip the cache.
There are 16 frames of buffer.
This creates non-sRGB texture views for sRGB texture formats to allow for interfacing with these views in compute shaders using imageLoad and imageStore.
Co-Authored-By: Rodrigo Locatti <reinuseslisp@airmail.cc>
Fix a tragic off-by-one condition that causes Vulkan's stream buffer to
think it's always full, using fallback memory. The OpenGL was also
affected by this bug to a lesser extent.
This uses a ring buffer similar to OpenGL's stream buffer for small
uploads. This stops us from allocating several small buffers, reducing
memory fragmentation and cache locality.
It uses dedicated allocations when possible.
Reimplement the buffer cache using cached bindings and page level
granularity for modification tracking. This also drops the usage of
shared pointers and virtual functions from the cache.
- Bindings are cached, allowing to skip work when the game changes few
bits between draws.
- OpenGL Assembly shaders no longer copy when a region has been modified
from the GPU to emulate constant buffers, instead GL_EXT_memory_object
is used to alias sub-buffers within the same allocation.
- OpenGL Assembly shaders stream constant buffer data using
glProgramBufferParametersIuivNV, from NV_parameter_buffer_object. In
theory this should save one hash table resolve inside the driver
compared to glBufferSubData.
- A new OpenGL stream buffer is implemented based on fences for drivers
that are not Nvidia's proprietary, due to their low performance on
partial glBufferSubData calls synchronized with 3D rendering (that
some games use a lot).
- Most optimizations are shared between APIs now, allowing Vulkan to
cache more bindings than before, skipping unnecesarry work.
This commit adds the necessary infrastructure to use Vulkan object from
OpenGL. Overall, it improves performance and fixes some bugs present on
the old cache. There are still some edge cases hit by some games that
harm performance on some vendors, this are planned to be fixed in later
commits.
Instead of using a two step initialization to report errors, initialize
the GPU renderer and rasterizer on the constructor and report errors
through std::runtime_error.
Setting __GL_SHADER_DISK_CACHE_PATH we can force the cache directory to
be in yuzu's user directory to stop commonly distributed malware from
deleting our driver shader cache. And by setting
__GL_SHADER_DISK_CACHE_SKIP_CLEANUP we can have an unbounded shader
cache size.
This has only been implemented on Windows, mostly because previous tests
didn't seem to work on Linux.
Disable the precompiled cache on Nvidia's driver. There's no need to
hide information the driver already has in its own cache.
Intel and AMD proprietary drivers are incapable of rendering to texture
views of different formats than the original texture. Avoid creating
these at a cache level. This will consume more memory, emulating them
with copies.
This breaks accelerated decoders trying to imageStore into images with
sRGB. The decoders are currently disabled so this won't cause issues at
runtime.
The current texture cache has several points that hurt maintainability
and performance. It's easy to break unrelated parts of the cache
when doing minor changes. The cache can easily forget valuable
information about the cached textures by CPU writes or simply by its
normal usage.The current texture cache has several points that hurt
maintainability and performance. It's easy to break unrelated parts
of the cache when doing minor changes. The cache can easily forget
valuable information about the cached textures by CPU writes or simply
by its normal usage.
This commit aims to address those issues.
fmt now automatically prints the numeric value of an enum class member
by default, so we don't need to use casts any more.
Reduces the line noise a bit.
The previous definition was:
#define NUM(field_name) (sizeof(Maxwell3D::Regs::field_name) / sizeof(u32))
In cases where `field_name` happens to refer to an array, Clang thinks
`sizeof(an array value) / sizeof(a type)` is an instance of the idiom
where `sizeof` is used to compute an array length. So it thinks the
type in the denominator ought to be the array element type, and warns if
it isn't, assuming this is a mistake.
In reality, `NUM` is not used to get array lengths at all, so there is no
mistake. Silence the warning by applying Clang's suggested workaround
of parenthesizing the denominator.
EmuWindow::PollEvents was called from the GPU thread (or the CPU thread
in sync-GPU mode) when swapping buffers. It had three implementations:
- In GRenderWindow, it didn't actually poll events, just set a flag and
emit a signal to indicate that a frame was displayed.
- In EmuWindow_SDL2_Hide, it did nothing.
- In EmuWindow_SDL2, it did call SDL_PollEvents, but this is wrong
because SDL_PollEvents is supposed to be called on the thread that set
up video - in this case, the main thread, which was sleeping in a
busyloop (regardless of whether sync-GPU was enabled). On macOS this
causes a crash.
To fix this:
- Rename EmuWindow::PollEvents to OnFrameDisplayed, and give it a
default implementation that does nothing.
- In EmuWindow_SDL2, do not override OnFrameDisplayed, but instead have
the main thread call SDL_WaitEvent in a loop.
This implements texture cube arrays with shadow comparisons but doesn't
fix the asserts related to it.
Fixes out of bounds reads on swizzle constructors and makes them use
bounds checked ::at instead of the unsafe operator[].
This emulates the behavior we get on GLSL with regular SSBOs with a
pointer + length pair. It aims to be consistent with the crashes we
might get.
Out of bounds stores are ignored. Atomics are ignored and return zero.
Reads return zero.
Allows some implementations to avoid completely zeroing out the internal
buffer of the optional, and instead only set the validity byte within
the structure.
This also makes it consistent how we return empty optionals.
This reworks how host<->device synchronization works on the Vulkan
backend. Instead of "protecting" resources with a fence and signalling
these as free when the fence is known to be signalled by the host GPU,
use timeline semaphores.
Vulkan timeline semaphores allow use to work on a subset of D3D12
fences. As far as we are concerned, timeline semaphores are a value set
by the host or the device that can be waited by either of them.
Taking advantange of this, we can have a monolithically increasing
atomic value for each submission to the graphics queue. Instead of
protecting resources with a fence, we simply store the current logical
tick (the atomic value stored in CPU memory). When we want to know if a
resource is free, it can be compared to the current GPU tick.
This greatly simplifies resource management code and the free status of
resources should have less false negatives.
To workaround bugs in validation layers, when these are attached there's
a thread waiting for timeline semaphores.
Now that the GPU is initialized when video backends are initialized,
it's no longer needed to query components once the game is running: it
can be done when yuzu is booting.
This allows us to pass components between constructors and in the
process remove all Core::System references in the video backend.
Add the necessary CMake code to copy the contents in a string source
shader (GLSL or GLASM) to a header file then consumed by video_core
files.
This allows editting GLSL in its own files without having to maintain
them in source files.
For now, only OpenGL presentation shaders are moved, but we can add
GLASM presentation shaders and static SPIR-V generation through
glslangValidator in the future.
Migrates a remaining common file over to the Common namespace, making it
consistent with the rest of common files.
This also allows for high-traffic FS related code to alias the
filesystem function namespace as
namespace FS = Common::FS;
for more concise typing.
Does not allocate more threads than available in the host system for boot-time shader compilation and always allocates at least 1 thread if hardware_concurrency() returns 0.
NV_shader_buffer_{load,store} is a 2010 extension that allows GL applications
to use what in Vulkan is known as physical pointers, this is basically C
pointers. On GLASM these is exposed through the LOAD/STORE/ATOM
instructions.
Up until now, assembly shaders were using NV_shader_storage_buffer_object.
These work fine, but have a (probably unintended) limitation that forces
us to have the limit of a single stage for all shader stages. In contrast,
with NV_shader_buffer_{load,store} we can pass GPU addresses to the
shader through local parameters (GLASM equivalent uniform constants, or
push constants on Vulkan). Local parameters have the advantage of being
per stage, allowing us to generate code without worrying about binding
overlaps.
* Switch game settings to use a pointer
In order to add full per-game settings, we need to be able to tell yuzu to switch
to using either the global or game configuration. Using a pointer makes it easier
to switch.
* configuration: add new UI without changing existing funcitonality
The new UI also adds General, System, Graphics, Advanced Graphics,
and Audio tabs, but as yet they do nothing. This commit keeps yuzu
to the same functionality as originally branched.
* configuration: Rename files
These weren't included in the last commit. Now they are.
* configuration: setup global configuration checkbox
Global config checkbox now enables/disables the appropriate tabs in the game
properties dialog. The use global configuration setting is now saved to the
config, defaulting to true. This also addresses some changes requested in the PR.
* configuration: swap to per-game config memory for properties dialog
Does not set memory going in-game. Swaps to game values when opening the
properties dialog, then swaps back when closing it. Uses a `memcpy` to swap.
Also implements saving config files, limited to certain groups of configurations
so as to not risk setting unsafe configurations.
* configuration: change config interfaces to use config-specific pointers
When a game is booted, we need to be able to open the configuration dialogs
without changing the settings pointer in the game's emualtion. A new pointer
specific to just the configuration dialogs can be used to separate changes
to just those config dialogs without affecting the emulation.
* configuration: boot a game using per-game settings
Swaps values where needed to boot a game.
* configuration: user correct config during emulation
Creates a new pointer specifically for modifying the configuration while
emulation is in progress. Both the regular configuration dialog and the game
properties dialog now use the pointer Settings::config_values to focus edits to
the correct struct.
* settings: split Settings::values into two different structs
By splitting the settings into two mutually exclusive structs, it becomes easier,
as a developer, to determine how to use the Settings structs after per-game
configurations is merged. Other benefits include only duplicating the required
settings in memory.
* settings: move use_docked_mode to Controls group
`use_docked_mode` is set in the input settings and cannot be accessed from the
system settings. Grouping it with system settings causes it to be saved with
per-game settings, which may make transferring configs more difficult later on,
especially since docked mode cannot be set from within the game properties
dialog.
* configuration: Fix the other yuzu executables and a regression
In main.cpp, we have to get the title ID before the ROM is loaded, else the
renderer will reflect only the global settings and now the user's game specific
settings.
* settings: use a template to duplicate memory for each setting
Replaces the type of each variable in the Settings::Values struct with a new
class that allows basic data reading and writing. The new struct
Settings::Setting duplicates the data in memory and can manage global overrides
per each setting.
* configuration: correct add-ons config and swap settings when apropriate
Any add-ons interaction happens directly through the global values struct.
Swapping bewteen structs now also includes copying the necessary global configs
that cannot be changed nor saved in per-game settings. General and System config
menus now update based on whether it is viewing the global or per-game settings.
* settings: restore old values struct
No longer needed with the Settings::Setting class template.
* configuration: implement hierarchical game properties dialog
This sets the apropriate global or local data in each setting.
* clang format
* clang format take 2
can the docker container save this?
* address comments and style issues
* config: read and write settings with global awareness
Adds new functions to read and write settings while keeping the global state in
focus. Files now generated per-game are much smaller since often they only need
address the global state.
* settings: restore global state when necessary
Upon closing a game or the game properties dialog, we need to restore all global
settings to the original global state so that we can properly open the
configuration dialog or boot a different game.
* configuration: guard setting values incorrectly
This disables setting values while a game is running if the setting is
overwritten by a per game setting.
* config: don't write local settings in the global config
Simple guards to prevent writing the wrong settings in the wrong files.
* configuration: add comments, assume less, and clang format
No longer assumes that a disabled UI element means the global state is turned
off, instead opting to directly answer that question. Still however assumes a
game is running if it is in that state.
* configuration: fix a logic error
Should not be negated
* restore settings' global state regardless of accept/cancel
Fixes loading a properties dialog and causing the global config dialog to show
local settings.
* fix more logic errors
Fixed the frame limit would set the global setting from the game properties
dialog. Also strengthened the Settings::Setting member variables and simplified
the logic in config reading (ReadSettingGlobal).
* fix another logic error
In my efforts to guard RestoreGlobalState, I accidentally negated the IsPowered
condition.
* configure_audio: set toggle_stretched_audio to tristate
* fixed custom rtc and rng seed overwriting the global value
* clang format
* rebased
* clang format take 4
* address my own review
Basically revert unintended changes
* settings: literal instead of casting
"No need to cast, use 1U instead"
Thanks, Morph!
Co-authored-by: Morph <39850852+Morph1984@users.noreply.github.com>
* Revert "settings: literal instead of casting
"
This reverts commit 95e992a87c898f3e882ffdb415bb0ef9f80f613f.
* main: fix status buttons reporting wrong settings after stop emulation
* settings: Log UseDockedMode in the Controls group
This should have happened when use_docked_mode was moved over to the controls group
internally. This just reflects this in the log.
* main: load settings if the file has a title id
In other words, don't exit if the loader has trouble getting a title id.
* use a zero
* settings: initalize resolution factor with constructor instead of casting
* Revert "settings: initalize resolution factor with constructor instead of casting"
This reverts commit 54c35ecb46a29953842614620f9b7de1aa9d5dc8.
* configure_graphics: guard device selector when Vulkan is global
Prevents the user from editing the device selector if Vulkan is the global
renderer backend. Also resets the vulkan_device variable when the users
switches back-and-forth between global and Vulkan.
* address reviewer concerns
Changes function variables to const wherever they don't need to be changed. Sets Settings::Setting to final as it should not be inherited from. Sets ConfigurationShared::use_global_text to static.
Co-Authored-By: VolcaEM <volcaem@users.noreply.github.com>
* main: load per-game settings after LoadROM
This prevents `Restart Emulation` from restoring the global settings *after* the per-game settings were applied. Thanks to BSoDGamingYT for finding this bug.
* Revert "main: load per-game settings after LoadROM"
This reverts commit 9d0d48c52d2dcf3bfb1806cc8fa7d5a271a8a804.
* main: only restore global settings when necessary
Loading the per-game settings cannot happen after the ROM is loaded, so we have to specify when to restore the global state. Again thanks to BSoD for finding the bug.
* configuration_shared: address reviewer concerns except operator overrides
Dropping operator override usage in next commit.
Co-Authored-By: LC <lioncash@users.noreply.github.com>
* settings: Drop operator overrides from Setting template
Requires using GetValue and SetValue explicitly. Also reverts a change that broke title ID formatting in the game properties dialog.
* complete rebase
* configuration_shared: translate "Use global configuration"
Uses ConfigurePerGame to do so, since its usage, at least as of now, corresponds with ConfigurationShared.
* configure_per_game: address reviewer concern
As far as I understand, it prevents the program from unnecessarily copying strings.
Co-Authored-By: LC <lioncash@users.noreply.github.com>
Co-authored-by: Morph <39850852+Morph1984@users.noreply.github.com>
Co-authored-by: VolcaEM <volcaem@users.noreply.github.com>
Co-authored-by: LC <lioncash@users.noreply.github.com>
Like MirrorOnceBorder, this requires the GL_EXT_texture_mirror_clamp extension. This extension is unfortunately not available on Intel's drivers (both Windows proprietary and Linux Mesa). Use GL_MIRROR_CLAMP_TO_EDGE as a fallback if the extension is unavailable.
After marking buffers as resident, Nvidia's driver seems to take a
slow path. To workaround this issue, copy to a STREAM_READ buffer and
then call GetNamedBufferSubData on it.
This is a temporary solution until we have asynchronous flushing.
Making the stream buffer resident increases GPU usage significantly on
some games. This seems to be addressed invalidating the stream buffer
with InvalidateBufferData instead of using a Unmap + Map (with
invalidation flags).
Switch games are allowed to bind less data than what they use in a
vertex buffer, the expected behavior here is that these values are read
as zero. At the moment of writing this only D3D12, OpenGL and NVN through
NV_vertex_buffer_unified_memory support vertex buffer with a size limit.
In theory this could be emulated on Vulkan creating a new VkBuffer for
each (handle, offset, length) tuple and binding the expected data to it.
This is likely going to be slow and memory expensive when used on the
vertex buffer and we have to do it on all draws because we can't know
without analyzing indices when a game is going to read vertex data out
of bounds.
This is not a problem on OpenGL's BufferAddressRangeNV because it takes
a length parameter, unlike Vulkan's CmdBindVertexBuffers that only takes
buffers and offsets (the length is implicit in VkBuffer). It isn't a
problem on D3D12 either, because D3D12_VERTEX_BUFFER_VIEW on
IASetVertexBuffers takes SizeInBytes as a parameter (although I am not
familiar with robustness on D3D12).
Currently this only implements buffer ranges for vertex buffers,
although indices can also be affected. A KHR_robustness profile is not
created, but Nvidia's driver reads out of bound vertex data as zero
anyway, this might have to be changed in the future.
- Fixes SMO random triangles when capturing an enemy, getting hit, or
looking at the environment on certain maps.
Make stream buffer and cached buffers as resident and query their
address. This allows us to use GPU addresses for several proprietary
Nvidia extensions.
Expose NV_vertex_buffer_unified_memory when the driver supports it.
This commit adds a function the determine if a GL_RENDERER is a Turing
GPU. This is required because on Turing GPUs Nvidia's driver crashes
when the buffer is marked as resident or on DeleteBuffers. Without a
synchronous debug output (single threaded driver), it's likely that
the driver will crash in the first blocking call.
Enable GL_EXT_texture_shadow_lod if available. If this extension is not available, such as on Intel/AMD proprietary drivers, use textureGrad as a workaround.
Variables that are marked as const cannot have the move constructor
invoked when returning from a function (the move constructor requires a
non-const variable so it can "steal" the resources from it.
maxwell_to_gl: Log unimplemented features under UNIMPLEMENTED_MSG instead of LOG_ERROR to bring into parity with maxwell_to_vk
maxwell_to_gl: Deduplicate logging in VertexType(), merging them into one.
maxwell_to_gl: Return GL_NEAREST instead of GL_LINEAR if an unknown texture filter mode is encountered.
maxwell_to_gl: Log the mipmap filter mode if an unknown value is passed in.
maxwell_to_gl: Reorder filtering modes to start with None, then Nearest, then Linear.
Due to the limitation of GL_MAX_IMAGE_UNITS being low (8) on Intel's and Nvidia's proprietary drivers, we have to reserve an appropriate amount of image bindings for each of the stages. So far games have been observed to use 4 image bindings on the fragment stage (Kirby Star Allies) and 1 on the vertex stage (TWD series).
No games thus far in my limited testing used more than 4 images concurrently and across all currently active programs.
This fixes shader compilation errors on Kirby Star Allies on OpenGL (GLSL/GLASM)
Emit code compatible with NV_gpu_program5.
This should emit code compatible with Fermi, but it wasn't tested on
that architecture. Pascal has some issues not present on Turing GPUs.
Instead of using as template argument a shared pointer, use the
underlying type and manage shared pointers explicitly. This can make
removing shared pointers from the cache more easy.
While we are at it, make some misc style changes and general
improvements (like insert_or_assign instead of operator[] + operator=).
Vertex buffers bindings become invalid after the stream buffer is
invalidated. We were originally doing this, but it got lost at some
point.
- Fixes Animal Crossing: New Horizons, but it affects everything.
This allows rendering to 3D textures with more than one slice.
Applications are allowed to render to more than one slice of a texture
using gl_Layer from a VTG shader.
This also requires reworking how 3D texture collisions are handled, for
now, this commit allows rendering to slices but not to miplevels. When a
render target attempts to write to a mipmap, we fallback to the previous
implementation (copying or flushing as needed).
- Fixes color correction 3D textures on UE4 games (rainbow effects).
- Allows Xenoblade games to render to 3D textures directly.
Skip fast buffer uploads on Nvidia 443.24 Vulkan beta driver on OpenGL.
This driver throws the following error when calling BufferSubData or
BufferData on buffers that are candidates for fast constant buffer
uploads. This is the equivalens to push constants on Vulkan, except that
they can access the full buffer. The error:
Unknown internal debug message. The NVIDIA OpenGL driver has encountered
an out of memory error. This application might
behave inconsistently and fail.
If this error persists on future drivers, we might have to look deeper
into this issue. For now, we can black list it and log it as a temporary
solution.
Games using D3D idioms can join images and samplers when a shader
executes, instead of baking them into a combined sampler image. This is
also possible on Vulkan.
One approach to this solution would be to use separate samplers on
Vulkan and leave this unimplemented on OpenGL, but we can't do this
because there's no consistent way of determining which constant buffer
holds a sampler and which one an image. We could in theory find the
first bit and if it's in the TIC area, it's an image; but this falls
apart when an image or sampler handle use an index of zero.
The used approach is to track for a LOP.OR operation (this is done at an
IR level, not at an ISA level), track again the constant buffers used as
source and store this pair. Then, outside of shader execution, join
the sample and image pair with a bitwise or operation.
This approach won't work on games that truly use separate samplers in a
meaningful way. For example, pooling textures in a 2D array and
determining at runtime what sampler to use.
This invalidates OpenGL's disk shader cache :)
- Used mostly by D3D ports to Switch
NV_transform_feedback, NV_transform_feedback2 and
ARB_transform_feedback3 with NV_transform_feedback interactions allows
implementing transform feedbacks as dynamic state.
Maxwell implements transform feedbacks as dynamic state, so using these
extensions with TransformFeedbackStreamAttribsNV allows us to properly
emulate transform feedbacks without having to recompile shaders when the
state changes.
On Intel's proprietary drivers, gl_Layer and gl_ViewportIndex are not allowed members of gl_PerVertex block, causing the shader to fail to compile. Fix this by declaring these variables outside of gl_PerVertex.
This avoids using Nvidia's ASTC decoder on OpenGL.
The last time it was profiled, it was slower than yuzu's decoder.
While we are at it, fix a bug in the texture cache when native ASTC is
not supported.
Previously we were disabling compute shaders on Intel's proprietary driver due to broken compute. This has been fixed in the latest Intel drivers. Re-enable compute for Intel proprietary drivers and remove the check for broken compute.
Stop ignoring image swizzles on depth and stencil images.
This doesn't fix a known issue on Xenoblade Chronicles 2 where an OpenGL
texture changes swizzles twice before being used. A proper fix would be
having a small texture view cache for this like we do on Vulkan.
While Vulkan was assuming we had no negative viewports, OpenGL code
was assuming we had them. Port the old code from Vulkan to OpenGL,
checking if the first viewport is negative before flipping faces.
This is not a complete implementation since we only check for the first
viewport to be negative. That said, unless a game is using Vulkan,
OpenGL and NVN games should be fine here, and we can always compare with
our Vulkan backend to see if there's a difference.
Hardware S2R special registers match gl_Thread*MaskNV. We can trivially
implement these using Nvidia's extension on OpenGL or naively stubbing
them with the ARB instructions to match. This might cause issues if the
host device warp size doesn't match Nvidia's. That said, this is
unlikely on proper shaders.
Refer to the attached url for more documentation about these flags.
https://www.khronos.org/registry/OpenGL/extensions/NV/NV_shader_thread_group.txt
Some operations like atomicMin were ignored because they returned were
being stored to RZ. This operations have a side effect and it was being
ignored.
Instead of using boost::icl::interval_map for caching, use
boost::intrusive::set. interval_map is intended as a container where the
keys can overlap with one another; we don't need this for caching
buffers and a std::set-like data structure that allows us to search with
lower_bound is enough.
Add code required to use OpenGL assembly programs based on
NV_gpu_program5. Decompilation for ARB programs is intended to be added
in a follow up commit. This does **not** include ARB decompilation and
it's not in an usable state.
The intention behind assembly programs is to reduce shader stutter
significantly on drivers supporting NV_gpu_program5 (and other required
extensions). Currently only Nvidia's proprietary driver supports these
extensions.
Add a UI option hidden for now to avoid people enabling this option
accidentally.
This code path has some limitations that OpenGL compatibility doesn't
have:
- NV_shader_storage_buffer_object is limited to 16 entries for a single
OpenGL context state (I don't know if this is an intended limitation, an
specification issue or I am missing something). Currently causes issues
on The Legend of Zelda: Link's Awakening.
- NV_parameter_buffer_object can't bind buffers using an offset
different to zero. The used workaround is to copy to a temporary buffer
(this doesn't happen often so it's not an issue).
On the other hand, it has the following advantages:
- Shaders build a lot faster.
- We have control over how floating point rounding is done over
individual instructions (SPIR-V on Vulkan can't do this).
- Operations on shared memory can be unsigned and signed.
- Transform feedbacks are dynamic state (not yet implemented).
- Parameter buffers (uniform buffers) are per stage, matching NVN and
hardware's behavior.
- The API to bind and create assembly programs makes sense, unlike
ARB_separate_shader_objects.
"Not equal" operators on GLSL seem to behave as unordered when we expect
an ordered comparison.
Manually emulate this checking for LGE values (numbers, not-NaNs).
Drop MemoryBarrier from the buffer cache and use Maxwell3D's register
WaitForIdle.
To implement this on OpenGL we just call glMemoryBarrier with the
necessary bits.
Vulkan lacks this synchronization primitive, so we set an event and
immediately wait for it. This is not a pretty solution, but it's what
Vulkan can do without submitting the current command buffer to the queue
(which ends up being more expensive on the CPU).
Using deko3d as reference:
4e47ba0013/source/maxwell/gpu_3d_state.cpp (L42)
We were using bits 3 and 4 to determine depth clamping, but these are
the same both enabled and disabled:
state->depthClampEnable ? 0x101A : 0x181D
The same happens on Nvidia's OpenGL driver, where they do something like
this (default capabilities, GL 4.5 compatibility):
(state & DEPTH_CLAMP) != 0 ? 0x201a : 0x281c
There's always a difference between the first bits in this register, but
bit 11 is consistently disabled on both deko3d/NVN and OpenGL. This
commit changes yuzu's behaviour to use bit 11 to determine depth
clamping.
- Fixes depth issues on Super Mario Odyssey's intro.
This reverts commit 94b0e2e5da.
preserve_contents proved to be a meaningful optimization. This commit
reintroduces it but properly implemented on OpenGL.
We have to make sure the clear removes all the previous contents of the
image.
It's not currently implemented on Vulkan because we can do smart things
there that's preferred to be introduced in a separate commit.
Deduplicate code shared between vk_pipeline_cache and gl_shader_cache as
well as shader decoder code.
While we are at it, fix a bug in gl_shader_cache where compute shaders
had an start offset of a stage shader.