Merge branch 'master' of https://github.com/yuzu-emu/yuzu
This commit is contained in:
@@ -3,7 +3,9 @@ function(copy_yuzu_Qt5_deps target_dir)
|
||||
set(DLL_DEST "${CMAKE_BINARY_DIR}/bin/$<CONFIG>/")
|
||||
set(Qt5_DLL_DIR "${Qt5_DIR}/../../../bin")
|
||||
set(Qt5_PLATFORMS_DIR "${Qt5_DIR}/../../../plugins/platforms/")
|
||||
set(Qt5_STYLES_DIR "${Qt5_DIR}/../../../plugins/styles/")
|
||||
set(PLATFORMS ${DLL_DEST}platforms/)
|
||||
set(STYLES ${DLL_DEST}styles/)
|
||||
windows_copy_files(${target_dir} ${Qt5_DLL_DIR} ${DLL_DEST}
|
||||
icudt*.dll
|
||||
icuin*.dll
|
||||
@@ -14,4 +16,5 @@ function(copy_yuzu_Qt5_deps target_dir)
|
||||
Qt5Widgets$<$<CONFIG:Debug>:d>.*
|
||||
)
|
||||
windows_copy_files(yuzu ${Qt5_PLATFORMS_DIR} ${PLATFORMS} qwindows$<$<CONFIG:Debug>:d>.*)
|
||||
windows_copy_files(yuzu ${Qt5_STYLES_DIR} ${STYLES} qwindowsvistastyle$<$<CONFIG:Debug>:d>.*)
|
||||
endfunction(copy_yuzu_Qt5_deps)
|
||||
|
||||
@@ -116,6 +116,7 @@ after_build:
|
||||
|
||||
mkdir $RELEASE_DIST
|
||||
mkdir $RELEASE_DIST/platforms
|
||||
mkdir $RELEASE_DIST/styles
|
||||
|
||||
# copy the compiled binaries and other release files to the release folder
|
||||
Get-ChildItem "$CMAKE_BINARY_DIR" -Filter "yuzu*.exe" | Copy-Item -destination $RELEASE_DIST
|
||||
@@ -136,6 +137,9 @@ after_build:
|
||||
# copy the qt windows plugin dll to platforms
|
||||
Copy-Item -path "C:/msys64/mingw64/share/qt5/plugins/platforms/qwindows.dll" -force -destination "$RELEASE_DIST/platforms"
|
||||
|
||||
# copy the qt windows vista style dll to platforms
|
||||
Copy-Item -path "C:/msys64/mingw64/share/qt5/plugins/styles/qwindowsvistastyle.dll" -force -destination "$RELEASE_DIST/styles"
|
||||
|
||||
7z a -tzip $MINGW_BUILD_ZIP $RELEASE_DIST\*
|
||||
7z a $MINGW_SEVENZIP $RELEASE_DIST
|
||||
}
|
||||
|
||||
@@ -126,6 +126,10 @@ void Maxwell3D::WriteReg(u32 method, u32 value, u32 remaining_params) {
|
||||
DrawArrays();
|
||||
break;
|
||||
}
|
||||
case MAXWELL3D_REG_INDEX(clear_buffers): {
|
||||
ProcessClearBuffers();
|
||||
break;
|
||||
}
|
||||
case MAXWELL3D_REG_INDEX(query.query_get): {
|
||||
ProcessQueryGet();
|
||||
break;
|
||||
@@ -415,5 +419,13 @@ bool Maxwell3D::IsShaderStageEnabled(Regs::ShaderStage stage) const {
|
||||
UNREACHABLE();
|
||||
}
|
||||
|
||||
void Maxwell3D::ProcessClearBuffers() {
|
||||
ASSERT(regs.clear_buffers.R == regs.clear_buffers.G &&
|
||||
regs.clear_buffers.R == regs.clear_buffers.B &&
|
||||
regs.clear_buffers.R == regs.clear_buffers.A);
|
||||
|
||||
VideoCore::g_renderer->Rasterizer()->Clear();
|
||||
}
|
||||
|
||||
} // namespace Engines
|
||||
} // namespace Tegra
|
||||
|
||||
@@ -436,7 +436,12 @@ public:
|
||||
u32 count;
|
||||
} vertex_buffer;
|
||||
|
||||
INSERT_PADDING_WORDS(0x99);
|
||||
INSERT_PADDING_WORDS(1);
|
||||
|
||||
float clear_color[4];
|
||||
float clear_depth;
|
||||
|
||||
INSERT_PADDING_WORDS(0x93);
|
||||
|
||||
struct {
|
||||
u32 address_high;
|
||||
@@ -473,7 +478,9 @@ public:
|
||||
|
||||
u32 depth_write_enabled;
|
||||
|
||||
INSERT_PADDING_WORDS(0x8);
|
||||
INSERT_PADDING_WORDS(0x7);
|
||||
|
||||
u32 d3d_cull_mode;
|
||||
|
||||
BitField<0, 3, ComparisonOp> depth_test_func;
|
||||
|
||||
@@ -493,7 +500,13 @@ public:
|
||||
u32 enable[NumRenderTargets];
|
||||
} blend;
|
||||
|
||||
INSERT_PADDING_WORDS(0x2D);
|
||||
INSERT_PADDING_WORDS(0xB);
|
||||
|
||||
union {
|
||||
BitField<4, 1, u32> triangle_rast_flip;
|
||||
} screen_y_control;
|
||||
|
||||
INSERT_PADDING_WORDS(0x21);
|
||||
|
||||
u32 vb_element_base;
|
||||
|
||||
@@ -523,7 +536,12 @@ public:
|
||||
}
|
||||
} tic;
|
||||
|
||||
INSERT_PADDING_WORDS(0x22);
|
||||
INSERT_PADDING_WORDS(0x21);
|
||||
|
||||
union {
|
||||
BitField<2, 1, u32> coord_origin;
|
||||
BitField<3, 10, u32> enable;
|
||||
} point_coord_replace;
|
||||
|
||||
struct {
|
||||
u32 code_address_high;
|
||||
@@ -584,7 +602,21 @@ public:
|
||||
|
||||
Cull cull;
|
||||
|
||||
INSERT_PADDING_WORDS(0x77);
|
||||
INSERT_PADDING_WORDS(0x2B);
|
||||
|
||||
union {
|
||||
u32 raw;
|
||||
BitField<0, 1, u32> Z;
|
||||
BitField<1, 1, u32> S;
|
||||
BitField<2, 1, u32> R;
|
||||
BitField<3, 1, u32> G;
|
||||
BitField<4, 1, u32> B;
|
||||
BitField<5, 1, u32> A;
|
||||
BitField<6, 4, u32> RT;
|
||||
BitField<10, 11, u32> layer;
|
||||
} clear_buffers;
|
||||
|
||||
INSERT_PADDING_WORDS(0x4B);
|
||||
|
||||
struct {
|
||||
u32 query_address_high;
|
||||
@@ -766,6 +798,9 @@ private:
|
||||
/// Handles writes to the macro uploading registers.
|
||||
void ProcessMacroUpload(u32 data);
|
||||
|
||||
/// Handles a write to the CLEAR_BUFFERS register.
|
||||
void ProcessClearBuffers();
|
||||
|
||||
/// Handles a write to the QUERY_GET register.
|
||||
void ProcessQueryGet();
|
||||
|
||||
@@ -788,21 +823,27 @@ ASSERT_REG_POSITION(rt, 0x200);
|
||||
ASSERT_REG_POSITION(viewport_transform[0], 0x280);
|
||||
ASSERT_REG_POSITION(viewport, 0x300);
|
||||
ASSERT_REG_POSITION(vertex_buffer, 0x35D);
|
||||
ASSERT_REG_POSITION(clear_color[0], 0x360);
|
||||
ASSERT_REG_POSITION(clear_depth, 0x364);
|
||||
ASSERT_REG_POSITION(zeta, 0x3F8);
|
||||
ASSERT_REG_POSITION(vertex_attrib_format[0], 0x458);
|
||||
ASSERT_REG_POSITION(rt_control, 0x487);
|
||||
ASSERT_REG_POSITION(depth_test_enable, 0x4B3);
|
||||
ASSERT_REG_POSITION(independent_blend_enable, 0x4B9);
|
||||
ASSERT_REG_POSITION(depth_write_enabled, 0x4BA);
|
||||
ASSERT_REG_POSITION(d3d_cull_mode, 0x4C2);
|
||||
ASSERT_REG_POSITION(depth_test_func, 0x4C3);
|
||||
ASSERT_REG_POSITION(blend, 0x4CF);
|
||||
ASSERT_REG_POSITION(screen_y_control, 0x4EB);
|
||||
ASSERT_REG_POSITION(vb_element_base, 0x50D);
|
||||
ASSERT_REG_POSITION(tsc, 0x557);
|
||||
ASSERT_REG_POSITION(tic, 0x55D);
|
||||
ASSERT_REG_POSITION(point_coord_replace, 0x581);
|
||||
ASSERT_REG_POSITION(code_address, 0x582);
|
||||
ASSERT_REG_POSITION(draw, 0x585);
|
||||
ASSERT_REG_POSITION(index_array, 0x5F2);
|
||||
ASSERT_REG_POSITION(cull, 0x646);
|
||||
ASSERT_REG_POSITION(clear_buffers, 0x674);
|
||||
ASSERT_REG_POSITION(query, 0x6C0);
|
||||
ASSERT_REG_POSITION(vertex_array[0], 0x700);
|
||||
ASSERT_REG_POSITION(independent_blend, 0x780);
|
||||
|
||||
@@ -19,6 +19,9 @@ public:
|
||||
/// Draw the current batch of vertex arrays
|
||||
virtual void DrawArrays() = 0;
|
||||
|
||||
/// Clear the current framebuffer
|
||||
virtual void Clear() = 0;
|
||||
|
||||
/// Notify rasterizer that the specified Maxwell register has been changed
|
||||
virtual void NotifyMaxwellRegisterChanged(u32 method) = 0;
|
||||
|
||||
|
||||
@@ -297,11 +297,8 @@ bool RasterizerOpenGL::AccelerateDrawBatch(bool is_indexed) {
|
||||
return true;
|
||||
}
|
||||
|
||||
void RasterizerOpenGL::DrawArrays() {
|
||||
if (accelerate_draw == AccelDraw::Disabled)
|
||||
return;
|
||||
|
||||
MICROPROFILE_SCOPE(OpenGL_Drawing);
|
||||
std::pair<Surface, Surface> RasterizerOpenGL::ConfigureFramebuffers(bool using_color_fb,
|
||||
bool using_depth_fb) {
|
||||
const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
|
||||
|
||||
// Sync the depth test state before configuring the framebuffer surfaces.
|
||||
@@ -310,9 +307,6 @@ void RasterizerOpenGL::DrawArrays() {
|
||||
// TODO(bunnei): Implement this
|
||||
const bool has_stencil = false;
|
||||
|
||||
const bool using_color_fb = true;
|
||||
const bool using_depth_fb = regs.zeta.Address() != 0;
|
||||
|
||||
const MathUtil::Rectangle<s32> viewport_rect{regs.viewport_transform[0].GetRect()};
|
||||
|
||||
const bool write_color_fb =
|
||||
@@ -344,11 +338,6 @@ void RasterizerOpenGL::DrawArrays() {
|
||||
BindFramebufferSurfaces(color_surface, depth_surface, has_stencil);
|
||||
|
||||
SyncViewport(surfaces_rect);
|
||||
SyncBlendState();
|
||||
SyncCullMode();
|
||||
|
||||
// TODO(bunnei): Sync framebuffer_scale uniform here
|
||||
// TODO(bunnei): Sync scissorbox uniform(s) here
|
||||
|
||||
// Viewport can have negative offsets or larger dimensions than our framebuffer sub-rect. Enable
|
||||
// scissor test to prevent drawing outside of the framebuffer region
|
||||
@@ -359,6 +348,66 @@ void RasterizerOpenGL::DrawArrays() {
|
||||
state.scissor.height = draw_rect.GetHeight();
|
||||
state.Apply();
|
||||
|
||||
// Only return the surface to be marked as dirty if writing to it is enabled.
|
||||
return std::make_pair(write_color_fb ? color_surface : nullptr,
|
||||
write_depth_fb ? depth_surface : nullptr);
|
||||
}
|
||||
|
||||
void RasterizerOpenGL::Clear() {
|
||||
const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
|
||||
|
||||
bool use_color_fb = false;
|
||||
bool use_depth_fb = false;
|
||||
|
||||
GLbitfield clear_mask = 0;
|
||||
if (regs.clear_buffers.R && regs.clear_buffers.G && regs.clear_buffers.B &&
|
||||
regs.clear_buffers.A) {
|
||||
clear_mask |= GL_COLOR_BUFFER_BIT;
|
||||
use_color_fb = true;
|
||||
}
|
||||
if (regs.clear_buffers.Z) {
|
||||
clear_mask |= GL_DEPTH_BUFFER_BIT;
|
||||
use_depth_fb = true;
|
||||
}
|
||||
|
||||
if (clear_mask == 0)
|
||||
return;
|
||||
|
||||
auto [dirty_color_surface, dirty_depth_surface] =
|
||||
ConfigureFramebuffers(use_color_fb, use_depth_fb);
|
||||
|
||||
// TODO(Subv): Support clearing only partial colors.
|
||||
glClearColor(regs.clear_color[0], regs.clear_color[1], regs.clear_color[2],
|
||||
regs.clear_color[3]);
|
||||
glClearDepth(regs.clear_depth);
|
||||
|
||||
glClear(clear_mask);
|
||||
|
||||
// Mark framebuffer surfaces as dirty
|
||||
if (dirty_color_surface != nullptr) {
|
||||
res_cache.MarkSurfaceAsDirty(dirty_color_surface);
|
||||
}
|
||||
if (dirty_depth_surface != nullptr) {
|
||||
res_cache.MarkSurfaceAsDirty(dirty_depth_surface);
|
||||
}
|
||||
}
|
||||
|
||||
void RasterizerOpenGL::DrawArrays() {
|
||||
if (accelerate_draw == AccelDraw::Disabled)
|
||||
return;
|
||||
|
||||
MICROPROFILE_SCOPE(OpenGL_Drawing);
|
||||
const auto& regs = Core::System().GetInstance().GPU().Maxwell3D().regs;
|
||||
|
||||
auto [dirty_color_surface, dirty_depth_surface] =
|
||||
ConfigureFramebuffers(true, regs.zeta.Address() != 0);
|
||||
|
||||
SyncBlendState();
|
||||
SyncCullMode();
|
||||
|
||||
// TODO(bunnei): Sync framebuffer_scale uniform here
|
||||
// TODO(bunnei): Sync scissorbox uniform(s) here
|
||||
|
||||
// Draw the vertex batch
|
||||
const bool is_indexed = accelerate_draw == AccelDraw::Indexed;
|
||||
const u64 index_buffer_size{regs.index_array.count * regs.index_array.FormatSizeInBytes()};
|
||||
@@ -439,11 +488,11 @@ void RasterizerOpenGL::DrawArrays() {
|
||||
state.Apply();
|
||||
|
||||
// Mark framebuffer surfaces as dirty
|
||||
if (color_surface != nullptr && write_color_fb) {
|
||||
res_cache.MarkSurfaceAsDirty(color_surface);
|
||||
if (dirty_color_surface != nullptr) {
|
||||
res_cache.MarkSurfaceAsDirty(dirty_color_surface);
|
||||
}
|
||||
if (depth_surface != nullptr && write_depth_fb) {
|
||||
res_cache.MarkSurfaceAsDirty(depth_surface);
|
||||
if (dirty_depth_surface != nullptr) {
|
||||
res_cache.MarkSurfaceAsDirty(dirty_depth_surface);
|
||||
}
|
||||
}
|
||||
|
||||
@@ -722,6 +771,16 @@ void RasterizerOpenGL::SyncCullMode() {
|
||||
if (state.cull.enabled) {
|
||||
state.cull.front_face = MaxwellToGL::FrontFace(regs.cull.front_face);
|
||||
state.cull.mode = MaxwellToGL::CullFace(regs.cull.cull_face);
|
||||
|
||||
// If the GPU is configured to flip the rasterized triangles, then we need to flip the
|
||||
// notion of front and back. Note: We flip the triangles when the value of the register is 0
|
||||
// because OpenGL already does it for us.
|
||||
if (regs.screen_y_control.triangle_rast_flip == 0) {
|
||||
if (state.cull.front_face == GL_CCW)
|
||||
state.cull.front_face = GL_CW;
|
||||
else if (state.cull.front_face == GL_CW)
|
||||
state.cull.front_face = GL_CCW;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
@@ -7,6 +7,7 @@
|
||||
#include <array>
|
||||
#include <cstddef>
|
||||
#include <memory>
|
||||
#include <utility>
|
||||
#include <vector>
|
||||
#include <glad/glad.h>
|
||||
#include "common/common_types.h"
|
||||
@@ -28,6 +29,7 @@ public:
|
||||
~RasterizerOpenGL() override;
|
||||
|
||||
void DrawArrays() override;
|
||||
void Clear() override;
|
||||
void NotifyMaxwellRegisterChanged(u32 method) override;
|
||||
void FlushAll() override;
|
||||
void FlushRegion(Tegra::GPUVAddr addr, u64 size) override;
|
||||
@@ -81,6 +83,10 @@ private:
|
||||
u32 border_color_a;
|
||||
};
|
||||
|
||||
/// Configures the color and depth framebuffer states and returns the dirty <Color, Depth>
|
||||
/// surfaces if writing was enabled.
|
||||
std::pair<Surface, Surface> ConfigureFramebuffers(bool using_color_fb, bool using_depth_fb);
|
||||
|
||||
/// Binds the framebuffer color and depth surface
|
||||
void BindFramebufferSurfaces(const Surface& color_surface, const Surface& depth_surface,
|
||||
bool has_stencil);
|
||||
|
||||
@@ -65,6 +65,25 @@ struct FormatTuple {
|
||||
return params;
|
||||
}
|
||||
|
||||
/*static*/ SurfaceParams SurfaceParams::CreateForDepthBuffer(
|
||||
const Tegra::Engines::Maxwell3D::Regs::RenderTargetConfig& config, Tegra::GPUVAddr zeta_address,
|
||||
Tegra::DepthFormat format) {
|
||||
|
||||
SurfaceParams params{};
|
||||
params.addr = zeta_address;
|
||||
params.is_tiled = true;
|
||||
params.block_height = Tegra::Texture::TICEntry::DefaultBlockHeight;
|
||||
params.pixel_format = PixelFormatFromDepthFormat(format);
|
||||
params.component_type = ComponentTypeFromDepthFormat(format);
|
||||
params.type = GetFormatType(params.pixel_format);
|
||||
params.size_in_bytes = params.SizeInBytes();
|
||||
params.width = config.width;
|
||||
params.height = config.height;
|
||||
params.unaligned_height = config.height;
|
||||
params.size_in_bytes = params.SizeInBytes();
|
||||
return params;
|
||||
}
|
||||
|
||||
static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_format_tuples = {{
|
||||
{GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, ComponentType::UNorm, false}, // ABGR8
|
||||
{GL_RGB, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV, ComponentType::UNorm, false}, // B5G6R5
|
||||
@@ -88,6 +107,8 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
|
||||
// DepthStencil formats
|
||||
{GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
|
||||
false}, // Z24S8
|
||||
{GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
|
||||
false}, // S8Z24
|
||||
}};
|
||||
|
||||
static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType component_type) {
|
||||
@@ -131,13 +152,6 @@ MathUtil::Rectangle<u32> SurfaceParams::GetRect() const {
|
||||
return {0, actual_height, width, 0};
|
||||
}
|
||||
|
||||
static void ConvertASTCToRGBA8(std::vector<u8>& data, PixelFormat format, u32 width, u32 height) {
|
||||
u32 block_width{};
|
||||
u32 block_height{};
|
||||
std::tie(block_width, block_height) = GetASTCBlockSize(format);
|
||||
data = Tegra::Texture::ASTC::Decompress(data, width, height, block_width, block_height);
|
||||
}
|
||||
|
||||
template <bool morton_to_gl, PixelFormat format>
|
||||
void MortonCopy(u32 stride, u32 block_height, u32 height, u8* gl_buffer, Tegra::GPUVAddr addr) {
|
||||
constexpr u32 bytes_per_pixel = SurfaceParams::GetFormatBpp(format) / CHAR_BIT;
|
||||
@@ -177,6 +191,7 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
|
||||
MortonCopy<true, PixelFormat::DXT1>, MortonCopy<true, PixelFormat::DXT23>,
|
||||
MortonCopy<true, PixelFormat::DXT45>, MortonCopy<true, PixelFormat::DXN1>,
|
||||
MortonCopy<true, PixelFormat::ASTC_2D_4X4>, MortonCopy<true, PixelFormat::Z24S8>,
|
||||
MortonCopy<true, PixelFormat::S8Z24>,
|
||||
};
|
||||
|
||||
static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
|
||||
@@ -197,6 +212,7 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
|
||||
nullptr,
|
||||
MortonCopy<false, PixelFormat::ABGR8>,
|
||||
MortonCopy<false, PixelFormat::Z24S8>,
|
||||
MortonCopy<false, PixelFormat::S8Z24>,
|
||||
};
|
||||
|
||||
// Allocate an uninitialized texture of appropriate size and format for the surface
|
||||
@@ -234,6 +250,71 @@ CachedSurface::CachedSurface(const SurfaceParams& params) : params(params) {
|
||||
rect.GetWidth(), rect.GetHeight());
|
||||
}
|
||||
|
||||
static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
|
||||
union S8Z24 {
|
||||
BitField<0, 24, u32> z24;
|
||||
BitField<24, 8, u32> s8;
|
||||
};
|
||||
static_assert(sizeof(S8Z24) == 4, "S8Z24 is incorrect size");
|
||||
|
||||
union Z24S8 {
|
||||
BitField<0, 8, u32> s8;
|
||||
BitField<8, 24, u32> z24;
|
||||
};
|
||||
static_assert(sizeof(Z24S8) == 4, "Z24S8 is incorrect size");
|
||||
|
||||
S8Z24 input_pixel{};
|
||||
Z24S8 output_pixel{};
|
||||
for (size_t y = 0; y < height; ++y) {
|
||||
for (size_t x = 0; x < width; ++x) {
|
||||
const size_t offset{y * width + x};
|
||||
std::memcpy(&input_pixel, &data[offset], sizeof(S8Z24));
|
||||
output_pixel.s8.Assign(input_pixel.s8);
|
||||
output_pixel.z24.Assign(input_pixel.z24);
|
||||
std::memcpy(&data[offset], &output_pixel, sizeof(Z24S8));
|
||||
}
|
||||
}
|
||||
}
|
||||
/**
|
||||
* Helper function to perform software conversion (as needed) when loading a buffer from Switch
|
||||
* memory. This is for Maxwell pixel formats that cannot be represented as-is in OpenGL or with
|
||||
* typical desktop GPUs.
|
||||
*/
|
||||
static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelFormat pixel_format,
|
||||
u32 width, u32 height) {
|
||||
switch (pixel_format) {
|
||||
case PixelFormat::ASTC_2D_4X4: {
|
||||
// Convert ASTC pixel formats to RGBA8, as most desktop GPUs do not support ASTC.
|
||||
u32 block_width{};
|
||||
u32 block_height{};
|
||||
std::tie(block_width, block_height) = GetASTCBlockSize(pixel_format);
|
||||
data = Tegra::Texture::ASTC::Decompress(data, width, height, block_width, block_height);
|
||||
break;
|
||||
}
|
||||
case PixelFormat::S8Z24:
|
||||
// Convert the S8Z24 depth format to Z24S8, as OpenGL does not support S8Z24.
|
||||
ConvertS8Z24ToZ24S8(data, width, height);
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Helper function to perform software conversion (as needed) when flushing a buffer to Switch
|
||||
* memory. This is for Maxwell pixel formats that cannot be represented as-is in OpenGL or with
|
||||
* typical desktop GPUs.
|
||||
*/
|
||||
static void ConvertFormatAsNeeded_FlushGLBuffer(std::vector<u8>& /*data*/, PixelFormat pixel_format,
|
||||
u32 /*width*/, u32 /*height*/) {
|
||||
switch (pixel_format) {
|
||||
case PixelFormat::ASTC_2D_4X4:
|
||||
case PixelFormat::S8Z24:
|
||||
LOG_CRITICAL(Render_OpenGL, "Unimplemented pixel_format={}",
|
||||
static_cast<u32>(pixel_format));
|
||||
UNREACHABLE();
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
MICROPROFILE_DEFINE(OpenGL_SurfaceLoad, "OpenGL", "Surface Load", MP_RGB(128, 64, 192));
|
||||
void CachedSurface::LoadGLBuffer() {
|
||||
ASSERT(params.type != SurfaceType::Fill);
|
||||
@@ -256,10 +337,7 @@ void CachedSurface::LoadGLBuffer() {
|
||||
params.width, params.block_height, params.height, gl_buffer.data(), params.addr);
|
||||
}
|
||||
|
||||
if (IsPixelFormatASTC(params.pixel_format)) {
|
||||
// ASTC formats are converted to RGBA8 in software, as most PC GPUs do not support this
|
||||
ConvertASTCToRGBA8(gl_buffer, params.pixel_format, params.width, params.height);
|
||||
}
|
||||
ConvertFormatAsNeeded_LoadGLBuffer(gl_buffer, params.pixel_format, params.width, params.height);
|
||||
}
|
||||
|
||||
MICROPROFILE_DEFINE(OpenGL_SurfaceFlush, "OpenGL", "Surface Flush", MP_RGB(128, 192, 64));
|
||||
@@ -272,6 +350,9 @@ void CachedSurface::FlushGLBuffer() {
|
||||
|
||||
MICROPROFILE_SCOPE(OpenGL_SurfaceFlush);
|
||||
|
||||
ConvertFormatAsNeeded_FlushGLBuffer(gl_buffer, params.pixel_format, params.width,
|
||||
params.height);
|
||||
|
||||
if (!params.is_tiled) {
|
||||
std::memcpy(dst_buffer, gl_buffer.data(), params.size_in_bytes);
|
||||
} else {
|
||||
@@ -399,15 +480,16 @@ SurfaceSurfaceRect_Tuple RasterizerCacheOpenGL::GetFramebufferSurfaces(
|
||||
LOG_WARNING(Render_OpenGL, "hard-coded for render target 0!");
|
||||
|
||||
// get color and depth surfaces
|
||||
const SurfaceParams color_params{SurfaceParams::CreateForFramebuffer(regs.rt[0])};
|
||||
SurfaceParams depth_params{color_params};
|
||||
SurfaceParams color_params{};
|
||||
SurfaceParams depth_params{};
|
||||
|
||||
if (using_color_fb) {
|
||||
color_params = SurfaceParams::CreateForFramebuffer(regs.rt[0]);
|
||||
}
|
||||
|
||||
if (using_depth_fb) {
|
||||
depth_params.addr = regs.zeta.Address();
|
||||
depth_params.pixel_format = SurfaceParams::PixelFormatFromDepthFormat(regs.zeta.format);
|
||||
depth_params.component_type = SurfaceParams::ComponentTypeFromDepthFormat(regs.zeta.format);
|
||||
depth_params.type = SurfaceParams::GetFormatType(depth_params.pixel_format);
|
||||
depth_params.size_in_bytes = depth_params.SizeInBytes();
|
||||
depth_params =
|
||||
SurfaceParams::CreateForDepthBuffer(regs.rt[0], regs.zeta.Address(), regs.zeta.format);
|
||||
}
|
||||
|
||||
MathUtil::Rectangle<u32> color_rect{};
|
||||
|
||||
@@ -41,6 +41,7 @@ struct SurfaceParams {
|
||||
|
||||
// DepthStencil formats
|
||||
Z24S8 = 13,
|
||||
S8Z24 = 14,
|
||||
|
||||
MaxDepthStencilFormat,
|
||||
|
||||
@@ -92,6 +93,7 @@ struct SurfaceParams {
|
||||
4, // DXN1
|
||||
4, // ASTC_2D_4X4
|
||||
1, // Z24S8
|
||||
1, // S8Z24
|
||||
}};
|
||||
|
||||
ASSERT(static_cast<size_t>(format) < compression_factor_table.size());
|
||||
@@ -117,6 +119,7 @@ struct SurfaceParams {
|
||||
64, // DXN1
|
||||
32, // ASTC_2D_4X4
|
||||
32, // Z24S8
|
||||
32, // S8Z24
|
||||
}};
|
||||
|
||||
ASSERT(static_cast<size_t>(format) < bpp_table.size());
|
||||
@@ -128,6 +131,8 @@ struct SurfaceParams {
|
||||
|
||||
static PixelFormat PixelFormatFromDepthFormat(Tegra::DepthFormat format) {
|
||||
switch (format) {
|
||||
case Tegra::DepthFormat::S8_Z24_UNORM:
|
||||
return PixelFormat::S8Z24;
|
||||
case Tegra::DepthFormat::Z24_S8_UNORM:
|
||||
return PixelFormat::Z24S8;
|
||||
default:
|
||||
@@ -226,6 +231,8 @@ struct SurfaceParams {
|
||||
|
||||
static Tegra::DepthFormat DepthFormatFromPixelFormat(PixelFormat format) {
|
||||
switch (format) {
|
||||
case PixelFormat::S8Z24:
|
||||
return Tegra::DepthFormat::S8_Z24_UNORM;
|
||||
case PixelFormat::Z24S8:
|
||||
return Tegra::DepthFormat::Z24_S8_UNORM;
|
||||
default:
|
||||
@@ -274,6 +281,7 @@ struct SurfaceParams {
|
||||
|
||||
static ComponentType ComponentTypeFromDepthFormat(Tegra::DepthFormat format) {
|
||||
switch (format) {
|
||||
case Tegra::DepthFormat::S8_Z24_UNORM:
|
||||
case Tegra::DepthFormat::Z24_S8_UNORM:
|
||||
return ComponentType::UNorm;
|
||||
default:
|
||||
@@ -318,13 +326,18 @@ struct SurfaceParams {
|
||||
return addr <= (region_addr + region_size) && region_addr <= (addr + size_in_bytes);
|
||||
}
|
||||
|
||||
/// Creates SurfaceParams from a texture configation
|
||||
/// Creates SurfaceParams from a texture configuration
|
||||
static SurfaceParams CreateForTexture(const Tegra::Texture::FullTextureInfo& config);
|
||||
|
||||
/// Creates SurfaceParams from a framebuffer configation
|
||||
/// Creates SurfaceParams from a framebuffer configuration
|
||||
static SurfaceParams CreateForFramebuffer(
|
||||
const Tegra::Engines::Maxwell3D::Regs::RenderTargetConfig& config);
|
||||
|
||||
/// Creates SurfaceParams for a depth buffer configuration
|
||||
static SurfaceParams CreateForDepthBuffer(
|
||||
const Tegra::Engines::Maxwell3D::Regs::RenderTargetConfig& config,
|
||||
Tegra::GPUVAddr zeta_address, Tegra::DepthFormat format);
|
||||
|
||||
Tegra::GPUVAddr addr;
|
||||
bool is_tiled;
|
||||
u32 block_height;
|
||||
|
||||
@@ -76,6 +76,7 @@ u32 BytesPerPixel(TextureFormat format) {
|
||||
|
||||
static u32 DepthBytesPerPixel(DepthFormat format) {
|
||||
switch (format) {
|
||||
case DepthFormat::S8_Z24_UNORM:
|
||||
case DepthFormat::Z24_S8_UNORM:
|
||||
return 4;
|
||||
default:
|
||||
@@ -129,6 +130,7 @@ std::vector<u8> UnswizzleDepthTexture(VAddr address, DepthFormat format, u32 wid
|
||||
std::vector<u8> unswizzled_data(width * height * bytes_per_pixel);
|
||||
|
||||
switch (format) {
|
||||
case DepthFormat::S8_Z24_UNORM:
|
||||
case DepthFormat::Z24_S8_UNORM:
|
||||
CopySwizzledData(width, height, bytes_per_pixel, bytes_per_pixel, data,
|
||||
unswizzled_data.data(), true, block_height);
|
||||
|
||||
Reference in New Issue
Block a user