fabi/Xenia-Canary
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[D3D12] Use typed buffers for texture untiling, calculate guest address once per thread

Browse Source
This commit is contained in:
Triang3l
2020-07-03 20:03:53 +03:00
parent 1d35ba9ee7
commit 713f26b5c8
169 changed files with 49015 additions and 22864 deletions
Download Patch File Download Diff File Expand all files Collapse all files

34
src/xenia/gpu/d3d12/texture_cache.h
View File

@@ -247,9 +247,6 @@ class TextureCache {
void UseScaledResolveBufferForReading();
void UseScaledResolveBufferForWriting();
// Can't address more than 512 MB on Nvidia, so an offset is required.
void CreateScaledResolveBufferRawSRV(D3D12_CPU_DESCRIPTOR_HANDLE handle,
uint32_t first_unscaled_4kb_page,
uint32_t unscaled_4kb_page_count);
void CreateScaledResolveBufferRawUAV(D3D12_CPU_DESCRIPTOR_HANDLE handle,
uint32_t first_unscaled_4kb_page,
uint32_t unscaled_4kb_page_count);
@@ -268,10 +265,14 @@ class TextureCache {
k32bpb,
k64bpb,
k128bpb,
kR11G11B10ToRGBA16,
kR11G11B10ToRGBA16SNorm,
kR5G5B5A1ToB5G5R5A1,
kR5G6B5ToB5G6R5,
kR5G5B6ToB5G6R5WithRBGASwizzle,
kR4G4B4A4ToB4G4R4A4,
kR10G11B11ToRGBA16,
kR10G11B11ToRGBA16SNorm,
kR11G11B10ToRGBA16,
kR11G11B10ToRGBA16SNorm,
kDXT1ToRGBA8,
kDXT3ToRGBA8,
kDXT5ToRGBA8,
@@ -291,9 +292,16 @@ class TextureCache {
struct LoadModeInfo {
const void* shader;
size_t shader_size;
// Log2 of the sizes, in bytes, of the source (guest) SRV and the
// destination (host) UAV accessed by the copying shader, since the shader
// may copy multiple blocks per one invocation.
uint32_t srv_bpe_log2;
uint32_t uav_bpe_log2;
// Optional shader for loading 2x-scaled resolve targets.
const void* shader_2x;
size_t shader_2x_size;
uint32_t srv_bpe_log2_2x;
uint32_t uav_bpe_log2_2x;
};
// Tiling modes for storing textures after resolving - needed only for the
@@ -420,26 +428,22 @@ class TextureCache {
struct LoadConstants {
// vec4 0.
// Base offset in bytes.
uint32_t guest_base;
// For linear textures - row byte pitch.
uint32_t guest_pitch;
// Block-aligned and, for mipmaps, power-of-two-aligned width and height.
// In blocks - and for mipmaps, it's also power-of-two-aligned.
uint32_t guest_storage_width_height[2];
// vec4 1.
uint32_t is_3d;
uint32_t guest_format;
uint32_t endianness;
uint32_t padding_1;
uint32_t size_blocks[3];
uint32_t is_3d_endian;
// vec4 2.
uint32_t size_blocks[3];
uint32_t height_texels;
// vec4 3.
// Base offset in bytes.
uint32_t host_base;
uint32_t host_pitch;
uint32_t padding_3[2];
uint32_t height_texels;
static constexpr uint32_t kGuestPitchTiled = UINT32_MAX;
};