Commit graph

6884 commits

Author SHA1 Message Date
Fernando Sahmkow
1d03a0fa75 Revert "renderer_vulkan: add locks to avoid scheduler flushes from CPU"
This reverts commit d9dde7e6f3.
2023-11-10 15:40:48 +01:00
liamwhite
40357098a2
Merge pull request #11896 from liamwhite/crop
renderer_vulkan: fix cropping for presentation
2023-11-06 12:08:03 -05:00
liamwhite
a423e0f9e0
renderer_vulkan: render on bottom of surface clip when flipped (#11894) 2023-11-05 21:47:35 +01:00
Liam
75de0cadcf renderer_null: fix 2023-11-03 20:54:38 -04:00
Liam
41701052d3 renderer_vulkan: minimize transform feedback support log 2023-11-01 20:47:08 -04:00
Ameer J
75c5be55af shader_recompiler: Align SSBO offsets in GlobalMemory functions 2023-10-31 20:14:18 -04:00
Ameer J
735612c9b3 buffer_cache: Apply storage buffer alignment only to the offset 2023-10-31 20:10:54 -04:00
Ameer J
7d34800531 shader_recompiler: Align SSBO offsets to meet host requirements
Co-Authored-By: Billy Laws <blaws05@gmail.com>
2023-10-31 20:10:54 -04:00
liamwhite
1836e62d33
Merge pull request #11910 from liamwhite/surface-lost-on-creation
renderer_vulkan: ensure exception on surface loss
2023-10-30 09:59:37 -04:00
liamwhite
ed2d77ddbc
Merge pull request #11893 from liamwhite/swizzle
renderer_vulkan: fix viewport swizzle dirty state tracking
2023-10-29 19:46:20 -04:00
Liam
8427b9d49d renderer_vulkan: ensure exception on surface loss 2023-10-29 15:31:05 -04:00
liamwhite
4da2105a32
Merge pull request #11862 from liamwhite/pascal-robust
Manually robust on Pascal and earlier
2023-10-29 11:25:15 -04:00
liamwhite
1f9684eaf9
Merge pull request #11859 from Kelebek1/compute_findbuffer
Add missing loop around compute FindBuffer calls
2023-10-29 11:25:09 -04:00
Liam
6513a356f0 renderer_vulkan: fix FSR cropping 2023-10-28 11:43:00 -04:00
Liam
65d4a16afd renderer_vulkan: fix cropping for presentation 2023-10-28 00:05:06 -04:00
Liam
21c631b33b renderer_vulkan: fix viewport swizzle dirty state tracking 2023-10-27 14:23:47 -04:00
Liam
e0834ee50b vulkan_common: use highest API version 2023-10-24 17:04:17 -04:00
Kelebek1
68f25217b8 Add missing dowhile loops around FindBuffer calls 2023-10-23 15:08:56 +01:00
Liam
0604b14263 Manually robust on Pascal and earlier 2023-10-23 09:08:57 -04:00
liamwhite
4b06bcc82c
Merge pull request #11789 from Kelebek1/spirv_shift_right
Manually robust on Maxwell and earlier
2023-10-21 18:21:53 -04:00
liamwhite
b56c7397ad
Merge pull request #11806 from liamwhite/needs-more-locking
renderer_vulkan: add locks to avoid scheduler flushes from CPU
2023-10-20 10:26:03 -04:00
Kelebek1
e02ee8e59d Manually robust on Maxwell and earlier 2023-10-19 19:54:31 +01:00
liamwhite
134ecca9b0
Merge pull request #11810 from liamwhite/clang-17
general: fix build failure on clang 17
2023-10-18 19:30:29 -04:00
liamwhite
c5f1ec8040
Merge pull request #11795 from Squall-Leonhart/D32FToOther
[Vulkan]Implement missing copy formats for D32, ARGB8_SRGB and BGRA8_Unorm/SRGB
2023-10-18 09:22:14 -04:00
Liam
c5bdc0054c general: fix build failure on clang 17 2023-10-17 22:44:21 -04:00
liamwhite
fa56518f20
Merge pull request #11747 from Kelebek1/image_alias_sample_names
Small things
2023-10-17 11:48:57 -04:00
liamwhite
b577d7a55f
Merge pull request #11349 from vonchenplus/buffer_cache_crash
video_core: Fix moltenvk crash on macos
2023-10-17 11:48:44 -04:00
Liam
d9dde7e6f3 renderer_vulkan: add locks to avoid scheduler flushes from CPU 2023-10-17 10:00:25 -04:00
Fernando S
2244b613cf
Merge pull request #11788 from Squall-Leonhart/IFREMOVED
[crash fix]brings back the removed if  statement in util.cpp and adds the  num_level test to it like previous discontinued PR
2023-10-17 14:36:36 +02:00
Squall-Leonhart
326ebbb2fa Changes based on hardware tests
Removes unnecessary d32f to bgra shader and blit functions,
update vk_texture_cache to use abgr shader for d32f to BGRA formats
updates  abgr to d32f shader to comply with hardware tests
2023-10-17 02:42:40 +11:00
Squall Leonhart
07143ce15c
Make Clang happy. 2023-10-17 00:26:19 +11:00
Squall Leonhart
dbc73c6c6c
Added missing BuildShader line
Adds `convert_abgr8_to_d32f_frag(BuildShader(device, CONVERT_ABGR8_TO_D32F_FRAG_SPV)),`
2023-10-17 00:15:31 +11:00
Squall Leonhart
90c56f5dc1
added missing trailing line. 2023-10-16 06:07:26 +11:00
Squall Leonhart
4b0291172e
meant to add the unorms as well 2023-10-16 04:29:24 +11:00
Squall-Leonhart
12e4757cf3 use texelfetch instead of texturelod 2023-10-16 04:20:45 +11:00
Squall Leonhart
144c0734f5
appease the format gods 2023-10-16 03:24:44 +11:00
Squall-Leonhart
f40f65f5d2 Another missing copy connected to Bravely Default II
adds blit_image_helper.ConvertABGR8ToD32F and fragment shader for performing ABGR and BGRA to D32F copies
2023-10-16 03:17:53 +11:00
Squall-Leonhart
03c3f936cf missed this line when editing the copypasta 2023-10-15 20:58:50 +11:00
Squall-Leonhart
66f41da365 moved line to appease the format gods 2023-10-15 20:54:25 +11:00
Squall-Leonhart
7a986d731b Implement missing formats for Bravely Default 2 2023-10-15 20:43:48 +11:00
Squall Leonhart
b57d98f847
brings back the removed If statement and adds the num_level test
This resolves the out of bounds read/writes in the linear swizzler, it brings back the scaled TOTK Recall bug however, pending further work in the block size calculation.

Recall is not glitched in the Dynamic FPS resolution mod to the degree that it is in the native yuzu scaler, this can be a workaround for the time being.


The recall effect is constructed from multiple 320x180 texture slices, it breaking may have a similar origin to https://github.com/Ryujinx/Ryujinx/pull/5640

but it may also be connected to the other deficiencies identified in the Yuzu size calculations, such as no apparent implementation of slice testing for end of slce depth as opposed to full aligned size as implemented in https://github.com/Ryujinx/Ryujinx/pull/5220
2023-10-15 02:09:28 +11:00
Valeri
762ac5aa9f
host1x/codecs: enable CUDA on Linux 2023-10-14 17:35:45 +03:00
Kelebek1
32ad99701d Implement vertex array first and subsequent draws 2023-10-14 12:09:35 +01:00
liamwhite
7b2ac196d2
Merge pull request #11751 from Kelebek1/transition_msaa_image
Transition MSAA images to general layout without uploading data
2023-10-12 11:17:20 -04:00
Kelebek1
6f4a080b98 Transition MSAA images to general layout without uploading data 2023-10-11 23:27:23 +01:00
liamwhite
07ae6659e7
Merge pull request #11744 from Kelebek1/no_res_no_rescaled
Do not set rescaled flag when rescaling is disabled
2023-10-11 17:54:59 -04:00
Kelebek1
98cac9410c Get out of render pass before query barriers, fix image names with samples > 1, remove image alias bit 2023-10-11 17:15:35 +01:00
liamwhite
84b0e29b56
Merge pull request #11734 from Kelebek1/device_local_buffer_alloc
Do not allocate DeviceLocal buffers as mapped
2023-10-11 09:24:28 -04:00
liamwhite
5f4857691e
Merge pull request #11683 from Kelebek1/do_not_sync_on_written_buffer
Do not double sync written buffers, move mark written to binding
2023-10-11 09:24:05 -04:00
Fernando S
b6d19329ac
Merge pull request #11743 from Squall-Leonhart/IFREMOVED
Fix mistaken usage of info.block instead of level_info.block
2023-10-11 11:56:47 +02:00
Kelebek1
8c769b71a1 Do not set rescaled flag when rescaling is disabled 2023-10-11 10:29:19 +01:00
Squall-Leonhart
9512992fe2 Fix mistaken usage of info.block instead of level_info.block
Fixed an error on my part, in the last change I had mistakenly passed unadjusted block info into FullUploadSwizzles and UnswizzleImage

Revert (my mistaken changing of) the construction of SwizzleParameters in UnswizzleImage and FullUploadSwizzles to use level_info.block instead of info.block. This ensures that the block information used in the swizzling process is correctly adjusted for each mip level.
2023-10-11 19:12:33 +11:00
liamwhite
36ea7565fa
Merge pull request #11534 from Squall-Leonhart/IFREMOVED
Partial revert of #10433 (Texture Cache Util: Fix block depth adjustment on slices)
2023-10-10 12:44:48 -04:00
Kelebek1
ed58445111 Not not allocate DeviceLocal buffers as mapped 2023-10-10 12:49:07 +01:00
Squall Leonhart
54fa1115a6
add Z32, FLOAT, UINT, UINT, UINT, LINEAR to format lookup table
Should fix and close #11711
2023-10-09 02:13:17 +11:00
liamwhite
bd42bba71c
Merge pull request #11656 from liamwhite/recreate-surface-automatically
vk_present_manager: recreate surface on any surface loss
2023-10-07 12:49:54 -04:00
liamwhite
a27f94830a
Merge pull request #11677 from Squall-Leonhart/D32FTOABGR8
Implements D32_Float to A8B8G8R8_UNORM format copy
2023-10-07 12:49:48 -04:00
liamwhite
bd6f9f1d91
Merge pull request #11630 from Kelebek1/clear_stencil_requires_depth_test
Enable depth test on depthstencil clear path
2023-10-07 12:49:37 -04:00
liamwhite
15a5bdd979
Merge pull request #11544 from Kelebek1/reduce_stream_buffer_renderdoc
Allow GPUs without rebar to open multiple RenderDoc captures
2023-10-07 12:49:19 -04:00
liamwhite
ff3859d482
Merge pull request #11688 from Kelebek1/x8d42
Implement X8_D24 pixel format
2023-10-07 10:55:14 -04:00
liamwhite
10de8f2c60
Merge pull request #11684 from Kelebek1/disable_push_descriptor_maxwell
Disable push descriptor for Pascal and older nVidia architectures
2023-10-07 10:54:52 -04:00
Squall Leonhart
51b89fddd0
update shader to confirmed format copy 2023-10-07 18:28:09 +11:00
Kelebek1
f585dec48d Allow GPUs without rebar to open multiple RenderDoc captures 2023-10-06 07:52:06 +01:00
Kelebek1
ad1a9f3d3a Implement X8_D24 format 2023-10-06 00:58:30 +01:00
Kelebek1
71044f6def Rework nvidia architecture detection, disable push descriptor for Pascal and older 2023-10-05 03:13:42 +01:00
Kelebek1
a764f49910 Mark a buffer GPU modified after the buffers are confirmed, do not double synch them 2023-10-05 00:19:11 +01:00
Squall-Leonhart
a17cde7b2c lets not convert depth to greyscale since this makes the exhaust and tire smoke light gray/white
tiresmoke should be a darker gray.
2023-10-05 03:14:53 +11:00
Squall-Leonhart
a84c928827 Fix CI Formatting check 2023-10-04 19:12:08 +11:00
Squall-Leonhart
9568d3bc60 Implements D32_Float to A8B8G8R8_UNORM format copy
Corrects some visual issues in games such as Disney SpeedStorm
2023-10-04 19:07:05 +11:00
Liam
b32940d3ea vk_present_manager: recreate surface on any surface loss 2023-10-02 19:07:18 -04:00
Liam
0448eb6f0f ci: fix new codespell errors 2023-10-02 18:03:05 -04:00
Fernando Sahmkow
9007d8c8d4 Query Cache: Fix memory leak. 2023-10-01 11:47:14 +02:00
Fernando S
926e24c642
Merge pull request #11622 from liamwhite/qcr-reg1
renderer_vulkan: fix query cache for homebrew
2023-09-29 06:01:18 +02:00
Kelebek1
1a246bf135 Enable depth test on stencil clear path 2023-09-28 21:19:51 +01:00
liamwhite
7bae22a3ca
Merge pull request #11402 from FernandoS27/depth-bias-control
Vulkan: Implement Depth Bias Control
2023-09-28 09:35:37 -04:00
Liam
7507a7f89f renderer_vulkan: fix query cache for homebrew 2023-09-27 19:11:47 -04:00
GPUCode
b60013b277 host_shaders: More proper handling of x2 MSAA copies 2023-09-25 09:20:32 -04:00
GPUCode
5e4938ab1a renderer_vulkan: Implement MSAA copies 2023-09-25 09:20:32 -04:00
liamwhite
854457a392
Merge pull request #11225 from FernandoS27/no-laxatives-in-santas-cookies
Y.F.C: Rework the Query Cache.
2023-09-25 09:18:29 -04:00
liamwhite
93a1cd75fe
Merge pull request #11562 from GPUCode/srgb-madness
vk_texture_cache: Limit srgb block to transcoding only
2023-09-24 10:50:28 -04:00
liamwhite
b356909212
Merge pull request #11165 from Morph1984/ds_blit
vulkan_device: Return true if either depth/stencil format supports blit
2023-09-24 10:50:04 -04:00
Fernando Sahmkow
57d8cd6c40 Query Cache: Fix Prefix Sums 2023-09-23 23:05:30 +02:00
Fernando Sahmkow
bf0d6b8806 Query Cache: Fix behavior in Normal Accuracy 2023-09-23 23:05:30 +02:00
Fernando Sahmkow
a07c88e686 Query Cache: Simplify Prefix Sum compute shader 2023-09-23 23:05:30 +02:00
Fernando Sahmkow
c8237d5c31 Query Cache: Implement host side sample counting. 2023-09-23 23:05:30 +02:00
Fernando Sahmkow
2fea1b8407 Query Cache: Fix guest side sample counting 2023-09-23 23:05:30 +02:00
Fernando Sahmkow
282ae8fa51 Query Cache: address issues 2023-09-23 23:05:30 +02:00
Fernando Sahmkow
aa6587d854 QueryCache: Implement dependant queries. 2023-09-23 23:05:29 +02:00
Fernando Sahmkow
57401589c2 Macro HLE: Add DrawIndirectByteCount 2023-09-23 23:05:29 +02:00
Fernando Sahmkow
f1a2e36711 Query Cachge: Fully rework Vulkan's query cache 2023-09-23 23:05:29 +02:00
Fernando Sahmkow
bdc01254a9 Query Cache: Setup Base rework 2023-09-23 23:05:29 +02:00
liamwhite
16f1592e50
Merge pull request #11557 from GPUCode/brr-format
renderer_vulkan: Correct component order for A4B4G4R4_UNORM
2023-09-22 09:56:04 -04:00
Kelebek1
4f69be8169 Fix DMA engine register offsets 2023-09-21 20:21:00 +01:00
GPUCode
b6ad7e263b vk_texture_cache: Limit srgb block to transcoding only 2023-09-21 21:46:35 +03:00
GPUCode
4a59dc2947 renderer_vulkan: Correct component order for A4B4G4R4_UNORM 2023-09-21 15:33:44 +03:00
Squall-Leonhart
02b897ce27 Reuse part of my previous idea to to use num_levels to check within AdjustMipBlockSize
The partial revert was not enough for Tsukihime, this might do the trick
2023-09-20 03:27:13 +10:00
liamwhite
974380fe10
Merge pull request #11258 from Squall-Leonhart/Z16_Assert_Fix
Fix a logged assert in the format lookup table for Z16
2023-09-18 09:31:05 -04:00
Squall Leonhart
79f0202045
Partial revert of #10433
The If block in this change was causing some 2D textures to be treated as if their mip 0 was a 3D Slice, this could be ascertained as the same texture viewed from different distances would render fine, but then close up would look like a decoding failure.

It also resulted in some 3D ASTC textures not being scaled appropriate leading to broken graphical effects such as the jagged TOTK recall animation being a circle, as the If block was only accepting the image based on its original info without any adjustments applied.
2023-09-18 23:28:53 +10:00
Charles Lombardo
af0c1b0cb7 android: Use 1 worker for shader compilation for all devices 2023-09-16 21:38:28 -04:00
Fernando Sahmkow
b99f94a7ff Vulkan: add temporary workaround for AMDVLK 2023-09-16 11:59:20 -04:00