Commit Graph

132 Commits

Author SHA1 Message Date
gdkchan
99445dd0a6
Add support for fragment shader interlock (#2768)
* Support coherent images

* Add support for fragment shader interlock

* Change to tree based match approach

* Refactor + check for branch targets and external registers

* Make detection more robust

* Use Intel fragment shader ordering if interlock is not available, use nothing if both are not available

* Remove unused field
2021-10-28 19:53:12 -03:00
gdkchan
d512ce122c
Initial tessellation shader support (#2534)
* Initial tessellation shader support

* Nits

* Re-arrange built-in table

* This is not needed anymore

* PR feedback
2021-10-18 18:38:04 -03:00
gdkchan
f08a280ade
Use shader subgroup extensions if shader ballot is not supported (#2627)
* Use shader subgroup extensions if shader ballot is not supported

* Shader cache version bump + cleanup

* The type is still required on the table
2021-09-19 14:38:39 +02:00
riperiperi
ec3e848d79
Add a Multithreading layer for the GAL, multi-thread shader compilation at runtime (#2501)
* Initial Implementation

About as fast as nvidia GL multithreading, can be improved with faster command queuing.

* Struct based command list

Speeds up a bit. Still a lot of time lost to resource copy.

* Do shader init while the render thread is active.

* Introduce circular span pool V1

Ideally should be able to use structs instead of references for storing these spans on commands. Will try that next.

* Refactor SpanRef some more

Use a struct to represent SpanRef, rather than a reference.

* Flush buffers on background thread

* Use a span for UpdateRenderScale.

Much faster than copying the array.

* Calculate command size using reflection

* WIP parallel shaders

* Some minor optimisation

* Only 2 max refs per command now.

The command with 3 refs is gone. 😌

* Don't cast on the GPU side

* Remove redundant casts, force sync on window present

* Fix Shader Cache

* Fix host shader save.

* Fixup to work with new renderer stuff

* Make command Run static, use array of delegates as lookup

Profile says this takes less time than the previous way.

* Bring up to date

* Add settings toggle. Fix Muiltithreading Off mode.

* Fix warning.

* Release tracking lock for flushes

* Fix Conditional Render fast path with threaded gal

* Make handle iteration safe when releasing the lock

This is mostly temporary.

* Attempt to set backend threading on driver

Only really works on nvidia before launching a game.

* Fix race condition with BufferModifiedRangeList, exceptions in tracking actions

* Update buffer set commands

* Some cleanup

* Only use stutter workaround when using opengl renderer non-threaded

* Add host-conditional reservation of counter events

There has always been the possibility that conditional rendering could use a query object just as it is disposed by the counter queue. This change makes it so that when the host decides to use host conditional rendering, the query object is reserved so that it cannot be deleted. Counter events can optionally start reserved, as the threaded implementation can reserve them before the backend creates them, and there would otherwise be a short amount of time where the counter queue could dispose the event before a call to reserve it could be made.

* Address Feedback

* Make counter flush tracked again.

Hopefully does not cause any issues this time.

* Wait for FlushTo on the main queue thread.

Currently assumes only one thread will want to FlushTo (in this case, the GPU thread)

* Add SDL2 headless integration

* Add HLE macro commands.

Co-authored-by: Mary <mary@mary.zone>
2021-08-27 00:31:29 +02:00
mpnico
8e1adb95cf
Add support for HLE macros and accelerate MultiDrawElementsIndirectCount #2 (#2557)
* Add support for HLE macros and accelerate MultiDrawElementsIndirectCount

* Add missing barrier

* Fix index buffer count

* Add support check for each macro hle before use

* Add missing xml doc

Co-authored-by: gdkchan <gab.dark.100@gmail.com>
2021-08-26 23:50:28 +02:00
gdkchan
5b8ceb9173
Swap BGR565 components by changing the format (#2577) 2021-08-26 22:47:21 +02:00
gdkchan
c702943af3
Swap BGR components for 16-bit BGR texture formats (#2567) 2021-08-20 18:26:25 -03:00
riperiperi
6c76bc3bc0
Change disabled vertex attribute value to (0, 0, 0, 1) (#2573)
This seems to be the default value when the vertex attribute is disabled, or components aren't defined.

This fixes a regression from #2307 in SMO where a plant in the Wooded Kingdom would draw slightly differently in the depth prepass, leading to depth test failing later on.

GDK has stated that the specific case in Gundam only expects x and y to be 0, and Vulkan's undefined value for z does appear to be 0 when a vertex attribute type does not have that component, hence the value (0, 0, 0, 1).

This worked in Vulkan despite also providing an all 0s attribute due to the vertex attribute binding being R32Float, so the other values were undefined. It should be changed there separately.
2021-08-20 18:09:30 -03:00
gdkchan
0ba4ade8f1
Ensure render scale is initialized to 1 on the backend (#2543) 2021-08-11 19:44:41 -03:00
gdkchan
d44d8f2eb6
Workaround for cubemap view data upload bug on Intel (#2539)
* Workaround for cubemap view data upload bug on Intel

* Trigger CI
2021-08-11 23:19:28 +02:00
gdkchan
c3e2646f9e
Workaround for Intel FrontFacing built-in variable bug (#2540) 2021-08-11 23:01:06 +02:00
gdkchan
0f6ec446ea
Replace BGRA and scale uniforms with a uniform block (#2496)
* Replace BGRA and scale uniforms with a uniform block

* Setting the data again on program change is no longer needed

* Optimize and resolve some warnings

* Avoid redundant support buffer updates

* Some optimizations to BindBuffers (now inlined)

* Unify render scale arrays
2021-08-11 21:33:43 +02:00
riperiperi
4b60371e64
Return mapped buffer pointer directly for flush, WriteableRegion for textures (#2494)
* Return mapped buffer pointer directly for flush, WriteableRegion for textures

A few changes here to generally improve performance, even for platforms not using the persistent buffer flush.

- Texture and buffer flush now return a ReadOnlySpan<byte>. It's guaranteed that this span is pinned in memory, but it will be overwritten on the next flush from that thread, so it is expected that the data is used before calling again.
- As a result, persistent mappings no longer copy to a new array - rather the persistent map is returned directly as a Span<>. A similar host array is used for the glGet flushes instead of allocating new arrays each time.
- Texture flushes now do their layout conversion into a WriteableRegion when the texture is not MultiRange, which allows the flush to happen directly into guest memory rather than into a temporary span, then copied over. This avoids another copy when doing layout conversion.

Overall, this saves 1 data copy for buffer flush, 1 copy for linear textures with matching source/target stride, and 2 copies for block textures or linear textures with mismatching strides.

* Fix tests

* Fix array pointer for Mesa/Intel path

* Address some feedback

* Update method for getting array pointer.
2021-07-19 19:10:54 -03:00
riperiperi
10e17ab423
Only use persistent buffers to flush on NVIDIA and Windows+AMD (#2489)
It seems like this method of flushing data is much slower on Mesa drivers, and slightly slower on Intel Windows. Have not tested Intel Mesa, but I'm assuming it is the same as AMD.

This also adds vendor detection for AMD on Unix, which counted as "Unknown" before.
2021-07-18 11:45:50 -03:00
riperiperi
ca5ac37cd6
Flush buffers and texture data through a persistent mapped buffer. (#2481)
* Use persistent buffers to flush texture data

* Flush buffers via copy to persistent buffers.

* Log error when timing out, small refactoring.
2021-07-16 18:10:20 -03:00
emmauss
28618c58d7
Add Screenshot Feature (#2354)
* Add internal screenshot  capabilities

* update version notice
2021-06-28 22:09:43 +02:00
gdkchan
fefd4619a5
Add support for custom line widths (#2406) 2021-06-25 20:11:54 -03:00
gdkchan
ed2f5ede0f
Fix texture sampling with depth compare and LOD level or bias (#2404)
* Fix texture sampling with depth compare and LOD level or bias

* Shader cache version bump

* nit: Sorting
2021-06-25 00:54:50 +02:00
riperiperi
fe29aff266
Use Quads on OpenGL host when supported. (#2331)
Improves OpenGL performance on FAST RMX and Xenoblade DE/2. Will probably only work on NVIDIA GPUs, but the emulated quads path will still be valid for other GPUs. Note: SLOW RMX gets a bit faster in handheld mode.

I'd recommend checking on platforms without supported host quads to make sure a GL error is actually thrown when attempting GL.Begin(PrimitiveType.Quads)
2021-06-02 13:27:30 +02:00
Mary
7527c5b906
Avoid clearing alpha channel by handle when presenting (#2323)
* Avoid clearning alpha channel by handle when presenting

Previous code was binding then blitting while the framebuffer was bound
and then clearing the alpha channel by its handle.

This ended up triggering a bug since AMD driver 21.4.1 ending up
clearing the whole framebuffer as a result.

New code fix this weird logic by applying the clear on the bound
framebuffer.

Close #2236.

* Address rip's comments

* Fix AMD being broken once again
2021-06-01 09:29:01 +02:00
gdkchan
3fd6b55f04
Fix value of constant vertex attributes (#2307) 2021-05-24 17:38:38 +10:00
riperiperi
c316c059ef
Assign _backgroundContext before starting its worker thread. (#2299)
Fixes a random crash when starting an embedded game.
2021-05-20 15:29:10 +02:00
EmulationFanatic
b5c72b44de
Merge pull request #2177 from riperiperi/feature/parallel-shader-cache
Allow parallel shader compilation when loading a shader cache
2021-05-19 11:39:19 -07:00
riperiperi
212e472c9f
Use copy dependencies for the Intel/AMD view format workaround (#2144)
* This might help AMD a bit

* Removal of old workaround.
2021-05-16 20:43:27 +02:00
riperiperi
ddf4b92a9c Implement parallel host shader cache compilation. 2021-04-18 17:33:58 +01:00
A-w-x
7719909397
HwCapabilities: Divide Intel into IntelWindows and IntelUnix (#2219)
as suggested by gdkchan
2021-04-18 02:27:19 +02:00
riperiperi
38be18ef2a
Fix intel view copy workaround. (#2216)
The texture target must be taken from the storage rather than the view, when using the storage handle for the copy.
2021-04-17 22:16:28 +02:00
Mary
6cb22c9d38
Miria: The Death of OpenTK 3 (#2194)
* openal: Update to OpenTK 4

* Ryujinx.Graphics.OpenGL: Update to OpenTK 4

* Entirely removed OpenTK 3, still wip

* Use SPB for context creation and handling

Still need to test on GLX and readd input support

* Start implementing a new input system

So far only gamepad are supported, no configuration possible via UI but detected via hotplug/removal

Button mapping backend is implemented

TODO: front end, configuration handling and configuration migration
TODO: keyboard support

* Enforce RGB only framebuffer on the GLWidget

Fix possible transparent window

* Implement UI gamepad frontend

Also fix bad mapping of minus button and ensure gamepad config is updated in real time

* Handle controller being disconnected and reconnected again

* Revert "Enforce RGB only framebuffer on the GLWidget"

This reverts commit 0949715d1a03ec793e35e37f7b610cbff2d63965.

* Fix first color clear

* Filter SDL2 events a bit

* Start working on the keyboard detail

- Rework configuration classes a bit to be more clean.
- Integrate fully the keyboard configuration to the front end (TODO: assigner)
- Start skeleton for the GTK3 keyboard driver

* Add KeyboardStateSnapshot and its integration

* Implement keyboard assigner and GTK3 key mapping

TODO: controller configuration mapping and IGamepad implementation for keyboard

* Add missing SR and SL definitions

* Fix copy pasta mistake on config for previous commit

* Implement IGamepad interface for GTK3 keyboard

* Fix some implementation still being commented in the controller ui for keyboard

* Port screen handle code

* Remove all configuration management code and move HidNew to Hid

* Rename InputConfigNew to InputConfig

* Add a version field to the input config

* Prepare serialization and deserialization of new input config and migrate profile loading and saving

* Support input configuration saving to config and bump config version to 23.

* Clean up in ConfigurationState

* Reference SPB via a nuget package

* Move new input system to Ryujinx.Input project and SDL2 detail to Ryujinx.Input.SDL2

* move GTK3 input to the right directory

* Fix triggers on SDL2

* Update to SDL2 2.0.14 via our own fork

* Update buttons definition for SDL2 2.0.14 and report gamepad features

* Implement motion support again with SDL2

TODO: cemu hooks integration

* Switch to latest of nightly SDL2

* SDL2: Fix bugs in gamepad id matching allowing different gamepad to match on the same device index

* Ensure values are set in UI when the gamepad get hot plugged

* Avoid trying to add controllers in the Update method and don't open SDL2 gamepad instance before checking ids

This fixes permanent rumble of pro controller in some hotplug scenario

* Fix more UI bugs

* Move legcay motion code around before reintegration

* gamecontroller UI tweaks here and there

* Hide Motion on non motion configurations

* Update the TODO grave

Some TODO were fixed long time ago or are quite oudated...

* Integrate cemu hooks motion configuration

* Integrate cemu hooks configuration options to the UI again

* cemuhooks => cemuhooks

* Add cemu hook support again

* Fix regression on normal motion and fix some very nasty bugs around

* Fix for XCB multithreads issue on Linux

* Enable motion by default

* Block inputs in the main view when in the controller configuration window

* Some fixes for the controller ui again

* Add joycon support and fixes other hints

* Bug fixes and clean up

- Invert default mapping if not a Nintendo controller
- Keep alive the controller being selected on the controller window (allow to avoid big delay for controller needing time to init when doing button assignment)
- Clean up hints in use
- Remove debug logs around
- Fixes potential double free with SDL2Gamepad

* Move the button assigner and motion logic to the Ryujinx.Input project

* Reimplement raw keyboard hle input

Also move out the logic of the hotkeys

* Move all remaining Input manager stuffs to the Ryujinx.Input project

* Increment configuration version yet again because of master changes

* Ensure input config isn't null when not present

* Fixes for VS not being nice

* Fix broken gamepad caching logic causing crashes on ui

* Ensure the background context is destroyed

* Update dependencies

* Readd retrocompat with old format of the config to avoid parsing and crashes on those versions

Also updated the debug Config.json

* Document new input APIs

* Isolate SDL2Driver to the project and remove external export of it

* Add support for external gamepad db mappings on SDL2

* Last clean up before PR

* Addresses first part of comments

* Address gdkchan's comments

* Do not use JsonException

* Last comment fixes
2021-04-14 12:28:43 +02:00
gdkchan
001005b3d5
Fix sub-image copies on intel GPUs (#2198) 2021-04-13 03:09:42 +02:00
riperiperi
0d27802902
Only flush commands when creating sync on Intel/AMD (windows) (#2165) 2021-04-03 11:10:37 +11:00
riperiperi
f2cdceb846
Force command flush after creating a syncpoint (#2056)
* Force command flush after creating GL sync

* This is not required, as the commands were flushed on creation.

* Move comment.
2021-03-30 01:10:43 +02:00
EmulationFanatic
2b92c10105
Fix typos (#2109) 2021-03-17 00:11:50 +01:00
riperiperi
1623ab524f
Improve Buffer Textures and flush Image Stores (#2088)
* Improve Buffer Textures and flush Image Stores

Fixes a number of issues with buffer textures:

- Reworked Buffer Textures to create their buffers in the TextureManager, then bind them with the BufferManager later.
  - Fixes an issue where a buffer texture's buffer could be invalidated after it is bound, but before use.
- Fixed width unpacking for large buffer textures. The width is now 32-bit rather than 16.
- Force buffer textures to be rebound whenever any buffer is created, as using the handle id wasn't reliable, and the cost of binding isn't too high.

Fixes vertex explosions and flickering animations in UE4 games.

* Set ImageStore flag... for ImageStore.

* Check the offset and size.
2021-03-08 18:43:39 -03:00
riperiperi
da283ff3c3
Flip component mask if target is BGRA. (#2087)
* Flip component mask if target is BGRA.

* Make mask selection less ugly.
2021-03-08 11:12:19 +11:00
riperiperi
b530f0e110
Texture Cache: "Texture Groups" and "Texture Dependencies" (#2001)
* Initial implementation (3d tex mips broken)

This works rather well for most games, just need to fix 3d texture mips.

* Cleanup

* Address feedback

* Copy Dependencies and various other fixes

* Fix layer/level offset for copy from view<->view.

* Remove dirty flag from dependency

The dirty flag behaviour is not needed - DeferredCopy is all we need.

* Fix tracking mip slices.

* Propagate granularity (fix astral chain)

* Address Feedback pt 1

* Save slice sizes as part of SizeInfo

* Fix nits

* Fix disposing multiple dependencies causing a crash

This list is obviously modified when removing dependencies, so create a copy of it.
2021-03-02 19:30:54 -03:00
gdkchan
7016d95eb1
Implement ETC2 (RGB) texture format (#2000)
* Implement ETC2 format

* Fix component counts for compressed formats
2021-02-08 10:23:56 +11:00
gdkchan
caf049ed15
Avoid some redundant GL calls (#1958) 2021-01-27 08:44:07 +11:00
riperiperi
a1f77a5b6a
Implement lazy flush-on-read for Buffers (SSBO/Copy) (#1790)
* Initial implementation of buffer flush (VERY WIP)

* Host shaders need to be rebuilt for the SSBO write flag.

* New approach with reserved regions and gl sync

* Fix a ton of buffer issues.

* Remove unused buffer unmapped behaviour

* Revert "Remove unused buffer unmapped behaviour"

This reverts commit f1700e52fb8760180ac5e0987a07d409d1e70ece.

* Delete modified ranges on unmap

Fixes potential crashes in Super Smash Bros, where a previously modified range could lie on either side of an unmap.

* Cache some more delegates.

* Dispose Sync on Close

* Also create host sync for GPFifo syncpoint increment.

* Copy buffer optimization, add docs

* Fix race condition with OpenGL Sync

* Enable read tracking on CommandBuffer, insert syncpoint on WaitForIdle

* Performance: Only flush individual pages of SSBO at a time

This avoids flushing large amounts of data when only a small amount is actually used.

* Signal Modified rather than flushing after clear

* Fix some docs and code style.

* Introduce a new test for tracking memory protection.

Sucessfully demonstrates that the bug causing write protection to be cleared by a read action has been fixed. (these tests fail on master)

* Address Comments

* Add host sync for SetReference

This ensures that any indirect draws will correctly flush any related buffer data written before them. Fixes some flashing and misplaced world geometry in MH rise.

* Make PageAlign static

* Re-enable read tracking, for reads.
2021-01-17 17:08:06 -03:00
gdkchan
df820a72de
Implement clear buffer (fast path) (#1902)
* Implement clear buffer (fast path)

* Remove blank line
2021-01-13 08:50:54 +11:00
Ac_K
73f6149bd6
gpu: Implement missing texture formats (#1867)
* gpu: Implement Etc2Rgba texture format

* Add more format

* Fix wrong pixel format
2021-01-05 06:02:49 +01:00
Ac_K
11222516c4
gui/gpu: Implement setting and toggle for Aspect Ratio (#1777)
* gui/gpu: Implement setting and toggle for Aspect Ratio

* address gdkchan feedback and add 16:10

* fix config.json file

* Fix rebase

* Address gdkchan feedback

* Address rip feedback

* Fix aspectWidth
2020-12-15 23:19:07 -03:00
riperiperi
06057a99a6
End empty gl queries before returning them to the pool. (#1806) 2020-12-12 10:31:39 +11:00
riperiperi
c00d39b675
Dummy out gl queries with 0 draws, remove glFlush call (#1773) 2020-12-03 19:42:59 +01:00
riperiperi
9493cdfe55
Allow copy destination to have a different scale from source (#1711)
* Allow copy destination to have a different scale from source

Will result in more scaled copy destinations, but allows scaling in some games that copy textures to the output framebuffer.

* Support copying multiple levels/layers

Uses glFramebufferTextureLayer to copy multiple layers, copies levels individually (and scales the regions).

Remove CopyArrayScaled, since the backend copy handles it now.
2020-11-20 17:14:45 -03:00
riperiperi
cf7044e37b
Perform Compressed<->Uncompressed copies using Pixel Buffer Objects (#1732)
* PBO single layer copy, part 1

Still needs ability to take and set width/height slices. (using pack paramaters)

* PBO Copies pt 2

* Some fixes and cleanup.

* Misc Cleanup

* Move handle into the TextureInfo interface.

This interface is shared between texture storages and views.

* Move unscaled copy to the TextureCopy class.

* Address feedback.
2020-11-20 13:30:59 -03:00
gdkchan
9435d62206
Simplify depth test state updates (#1695) 2020-11-17 23:20:17 +01:00
Mary
aa129fdbdf
infra: Migrate to .NET 5 (#1694)
* infra: Migrate to .NET 5

This migrate projects and CI to .NET 5

* Remove language version restrictions (now on 9.0 by default)

* infra: pin .NET 5 to avoid later issues

* infra: Cleanup csproj files

* infra: update dependencies

* infra: Add temporary workaround for a bug in Vector128.Create

see https://github.com/dotnet/runtime/issues/44704 for more informations
2020-11-15 19:27:15 +01:00
Mary
48f6570557
Salieri: shader cache (#1701)
Here come Salieri, my implementation of a disk shader cache!

"I'm sure you know why I named it that."
"It doesn't really mean anything."

This implementation collects shaders at runtime and cache them to be later compiled when starting a game.
2020-11-13 00:15:34 +01:00
gdkchan
8d168574eb
Use explicit buffer and texture bindings on shaders (#1666)
* Use explicit buffer and texture bindings on shaders

* More XML docs and other nits
2020-11-08 12:10:00 +01:00
riperiperi
4c6feb652f
Add seamless cubemap flag in sampler parameters. (#1658)
* Add seamless cubemap flag in sampler parameters.

* Check for the extension
2020-11-02 17:03:06 -03:00