Commit graph

60 commits

Author SHA1 Message Date
riperiperi
17ba217940
Vulkan: Device map buffers written more than flushed (#4911) 2023-05-13 15:15:05 +02:00
riperiperi
5fda543f84
Vulkan: Partially workaround MoltenVK InvalidResource error (#4880)
* Add MVK stage flags workaround

* Actually do the workaround

* Remove GS on VS stuff

* Address feedback
2023-05-11 22:06:15 -03:00
Aaron O'Mullan
a7c6e6a8cf
fix(mvk): resumeLostDevice (#4800)
Command buffer errors currently trigger an exception "DeviceLost" crashing the process.

Looking at [MKV's code](53a4eb26f2/MoltenVK/MoltenVK/GPUObjects/MVKQueue.mm (L392-L408)) we observe that:
- It hard fails if error is:
  ```
   MTLCommandBufferErrorBlacklisted || MTLCommandBufferErrorNotPermitted || MTLCommandBufferErrorDeviceRemoved
   ```
- Otherwise fails conditionally if `config.resumeLostDevice == false` (current default)

For Ryujinx's use-case it's more graceful to resume on those errors rather than crashing the app, the error isn't totally silenced since `mvk` still logs it

Fixes #4704, #4575
2023-05-10 03:31:52 +02:00
Mary
a8950d6ac4
vulkan: Pass Vk instance to VulkanRenderer (#4859)
This will allow possible multiple driver selection without any need of
LD preload. (useful when testing custom version of mesa for example)
2023-05-08 13:05:37 +02:00
Mary
162798b026
vulkan: Avoid hardcoding features in CreateDevice (#4858)
Those shouldn't have been hardcoded.
2023-05-08 10:48:16 +00:00
riperiperi
1b28ecd63e
Vulkan: Simplify MultiFenceHolder and managing them (#4845)
* Vulkan: Simplify waitable add/remove

Removal of unnecessary hashset and dictionary

* Thread safety for GetBufferData in PersistentFlushBuffer

* Fix WaitForFencesImpl thread safety

* Proper methods for risky reference increments

* Wrong type of CB.

* Address feedback
2023-05-08 12:45:12 +02:00
riperiperi
895d9b53bc
Vulkan: Batch vertex buffer updates (#4843)
* Vulkan: Batch vertex buffer updates

Some games can bind a large number of vertex buffers for draws. This PR allows for vertex buffers to be updated with one call rather than one per buffer.

This mostly affects the AMD Mesa driver, the testing platform was Steam Deck with Super Mario Odyssey. It was taking about 12% before, should be greatly reduced now.

A small optimization has been added to avoid looking up the same buffer multiple times, as a common pattern is for the same buffer to be bound many times in a row with different ranges.

* Only rebind vertex buffers if they have changed

* Address feedback
2023-05-08 10:59:26 +02:00
riperiperi
7c327fecb3
Vulkan: Record modifications after changing the framebuffer (#4775)
Our Vulkan backend inserts image barriers when a texture is sampled after it is rendered. This is done via a "modification flag" which is set when a render target is unbound (presuming that a texture has finished drawing to it).

Imagine the following scenario:
- Game sets render target to texture A
- Game renders to texture A
- (render pass ends)
- Game binds texture A to a sampler
- Game sets render target to texture B
- Renders to texture B using texture A (barrier required)

Because of the previous behaviour, the check to add a barrier for sampling a texture actually happens before it is registered as modified, meaning no barrier was added at all. This isn't always the case, but it was definitely causing issues in Xenoblade 2.

This doesn't fix any more complicated issues where a texture is repeatedly sampled while it is currently being rendered.

Fixes visual glitches at lower resolutions in Xenoblade 2. May fix other cases.
2023-05-03 10:42:21 +02:00
riperiperi
e18d258fa0
GPU: Pre-emptively flush textures that are flushed often (to imported memory when available) (#4711)
* WIP texture pre-flush

Improve performance of TextureView GetData to buffer

Fix copy/sync ordering

Fix minor bug

Make this actually work

WIP host mapping stuff

* Fix usage flags

* message

* Cleanup 1

* Fix rebase

* Fix

* Improve pre-flush rules

* Fix pre-flush

* A lot of cleanup

* Use the host memory bits

* Select the correct memory type

* Cleanup TextureGroupHandle

* Missing comment

* Remove debugging logs

* Revert BufferHandle _value access modifier

* One interrupt action at a time.

* Support D32S8 to D24S8 conversion, safeguards

* Interrupt cannot happen in sync handle's lock

Waitable needs to be checked twice now, but this should stop it from deadlocking.

* Remove unused using

* Address some feedback

* Address feedback

* Address more feedback

* Address more feedback

* Improve sync rules

Should allow for faster sync in some cases.
2023-05-01 16:05:12 -03:00
TSR Berry
cee7121058 Move solution and projects to src 2023-04-27 23:51:14 +02:00