a7885ae1a4
[GPU] Fix CPU-side float24 conversion broken recently
2022-06-22 20:47:44 +03:00
4514050f55
[Vulkan] Truncate depth to float24 in EDRAM range ownership transfers and resolves by default
...
Doesn't ruin the "greater or equal" depth test in subsequent rendering passes if precision is lost, unlike rounding to the nearest
2022-06-22 13:25:06 +03:00
0d8bd0e0c6
Merge branch 'master' into vulkan
2022-06-22 13:15:50 +03:00
cbf0476d42
[D3D12] Don't round float24 depth when it's known to be exact
2022-06-22 13:14:38 +03:00
7869b080d3
[D3D12] Truncate depth to float24 in EDRAM range ownership transfers and resolves by default
...
Doesn't ruin the "greater or equal" depth test in subsequent rendering passes if precision is lost, unlike rounding to the nearest
2022-06-22 12:53:09 +03:00
c0703e64db
Merge branch 'master' into vulkan
2022-06-20 22:40:19 +03:00
e2f632f8fa
[D3D12] Use udiv by constant tile size + minor transfer cleanup
...
Drivers compile that to a multiplication and a shift anyway.
2022-06-20 22:39:30 +03:00
0dc480721f
[Vulkan] Render target resolving
2022-06-20 22:29:07 +03:00
c6ec6d8239
[Vulkan] Use UDiv/UMod by constant tile size + minor transfer cleanup
...
Drivers compile that to a multiplication and a shift anyway.
2022-06-20 22:24:07 +03:00
61c4c49d76
Merge branch 'master' into vulkan
2022-06-20 12:34:41 +03:00
207e11c8d2
[GPU] Separate range arguments for fixed16 RG and RGBA in GetResolveInfo
...
On Vulkan, when snorm16 in unsupported, these formats may be emulated as float16, which natively can represent a wide range of numbers including -32 to 32 with blending. However, R16G16_SNORM and R16G16B16A16_SNORM are two separate formats, which may have different support on the device.
2022-06-20 12:29:45 +03:00
3b4845511d
[Vulkan] Don't require an explicit uint64_t cast for SetDeviceObjectName
2022-06-20 12:25:52 +03:00
67ff108f53
[Vulkan] Explain why CreateShaderModule takes uint32_t* [ci skip]
2022-06-20 12:22:41 +03:00
b61953374e
[GPU] Make resolve EDRAM binding DS 0 and rename it
...
Ordering the descriptor sets by the change frequency on Vulkan, in increasing order (the opposite of D3D12 root signatures). The EDRAM binding never changes there (always one storage buffer), while the destination buffer binding may become changeable in the future (to split dispatches if exceeding `maxStorageBufferRange`, for example).
2022-06-20 12:15:52 +03:00
1200b205cf
Merge branch 'master' into vulkan
2022-06-19 17:52:28 +03:00
9b83d3d0f4
[GPU] XeSL resolve shaders + host depth store width fix
2022-06-19 17:50:21 +03:00
166be463be
[XeSL] Metal Shading Language definitions
2022-06-16 21:39:16 +03:00
127bf34264
[Vulkan] Trace dump tool
2022-06-13 13:03:02 +03:00
ac268afbe9
[Vulkan] Fix 1<< uint32_t constants
2022-06-12 19:45:12 +03:00
140ed51e9a
[GPU] Fix missing xenia-ui dependency in gpu > gpu-shader-compiler (needed for gmake2)
2022-06-12 19:44:24 +03:00
17c835b245
Merge branch 'master' into vulkan
2022-06-12 18:51:08 +03:00
820b7ba217
[GPU] Fix GetActiveTextureHostSwizzle return type
2022-06-12 18:50:38 +03:00
1a22216e44
[SPIR-V] Texture fetch instructions
2022-06-09 21:42:16 +03:00
f875a8d887
Merge branch 'master' into vulkan
2022-06-09 21:35:12 +03:00
78d1eb8bf8
[GPU] TextureCache::GetActiveTextureHostSwizzle
2022-06-09 21:34:21 +03:00
56f72da137
[GPU] More exact PWL texture/RT gamma conversion
2022-06-07 21:26:34 +03:00
a8cfe9bebb
[Vulkan] Unsubsample odd-sized 4:2:2 textures
2022-06-02 23:10:50 +03:00
1ce45ee150
Merge branch 'master' into vulkan
2022-06-02 22:50:14 +03:00
55a91afcc7
[D3D12] Don't decompress unaligned BC textures if supported
2022-06-02 22:48:03 +03:00
84fcd5defa
[GPU] Fix resolve destination offset and extent calculation
2022-06-02 21:47:30 +03:00
a9a072bf00
[GPU] Explain why a 32x32x4bpp linear texture takes 2 pages, not 1 [ci skip]
2022-06-01 13:00:23 +03:00
8bd244f277
[GPU] Better explanation for exact texture memory extent calculation [ci skip]
2022-06-01 12:55:16 +03:00
d1ad10b98c
[GPU] Primitive reset comment typo correction [ci skip]
2022-05-31 23:23:53 +03:00
efd7ef212a
[D3D12] 128 megatexel limit explanation based on the spec [ci skip]
2022-05-31 23:23:10 +03:00
25594c918c
[GPU] Fix tiled texture memory extent calculation
2022-05-31 23:17:33 +03:00
a3e5ea8575
[Base] Fix missing include in utf8.cc.
2022-05-27 17:56:14 -05:00
6c9a06b2da
[Vulkan] Texture loading
2022-05-24 22:42:22 +03:00
9c445d397b
[Vulkan] Fix single-type descriptor pool reuse
2022-05-24 22:37:49 +03:00
aac28f19d1
Merge branch 'master' into vulkan
2022-05-24 22:34:40 +03:00
a4840e1992
[GPU] FIXME comment for 1bpb/2bpb texture tiled extent
2022-05-24 22:33:27 +03:00
8701c9f24e
[D3D12] Texture load code cleanup and resolution scaling fixes
...
The resolution scale is now taken into account when copying from the mip tail.
2022-05-24 22:28:42 +03:00
75c185e759
[GPU] Move texture load shader info to common
2022-05-24 22:24:33 +03:00
f994d3ebb3
[Vulkan] Single block-compressed flag for host texture formats, not block sizes
2022-05-23 13:27:43 +03:00
f7b0edee6b
[Vulkan] GBGR/BGRG decompression
2022-05-23 13:18:47 +03:00
4c2f8764d6
Merge branch 'master' into vulkan
2022-05-23 12:36:35 +03:00
c1f15c86a3
[GPU] Decompress GBGR/BGRG into RGBB, not RGB1
...
While the alpha of the texture data is not used at all (replaced with blue using the view swizzle), still make the shader code state the intention more explicitly if the format is decompressed for use as signed. Unsigned 1.0 is 0xFF, while signed 1.0 is 0x7F.
2022-05-23 12:31:45 +03:00
cf3069eb13
[GPU] Signedness in Cr_Y1_Cb_Y0_REP/Y1_Cr_Y0_Cb_REP comment [ci skip]
2022-05-22 22:11:59 +03:00
ef808e9def
[GPU] _REP explanation in Cr_Y1_Cb_Y0_REP/Y1_Cr_Y0_Cb_REP comment [ci skip]
2022-05-22 21:46:11 +03:00
6735dbd941
[GPU] Calculate, not store, texture load host X blocks per thread
2022-05-22 21:21:54 +03:00
888d5044e0
[GPU] 2x1-subsampled texture RGBA8 conversion shader
2022-05-22 21:07:38 +03:00
Triang3l