tacosukoneko/suyu
1
0
Fork 0
forked from suyu/suyu
Code Pull requests Activity

Merge pull request #2233 from ReinUsesLisp/morton-cleanup

Browse source 
video_core/morton: Miscellaneous changes
This commit is contained in:
bunnei 2019-03-14 21:23:12 -04:00 committed by GitHub
commit 84d3cdf7d7
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
4 changed files with 146 additions and 187 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

57
src/video_core/morton.cpp
View file

@ -16,12 +16,12 @@ namespace VideoCore {
using Surface::GetBytesPerPixel;
using Surface::PixelFormat;
using MortonCopyFn = void (*)(u32, u32, u32, u32, u32, u32, u8*, std::size_t, VAddr);
using MortonCopyFn = void (*)(u32, u32, u32, u32, u32, u32, u8*, VAddr);
using ConversionArray = std::array<MortonCopyFn, Surface::MaxPixelFormat>;
template <bool morton_to_linear, PixelFormat format>
static void MortonCopy(u32 stride, u32 block_height, u32 height, u32 block_depth, u32 depth,
u32 tile_width_spacing, u8* buffer, std::size_t buffer_size, VAddr addr) {
u32 tile_width_spacing, u8* buffer, VAddr addr) {
constexpr u32 bytes_per_pixel = GetBytesPerPixel(format);
// With the BCn formats (DXT and DXN), each 4x4 tile is swizzled instead of just individual
@ -42,7 +42,6 @@ static void MortonCopy(u32 stride, u32 block_height, u32 height, u32 block_depth
}
static constexpr ConversionArray morton_to_linear_fns = {
// clang-format off
MortonCopy<true, PixelFormat::ABGR8U>,
MortonCopy<true, PixelFormat::ABGR8S>,
MortonCopy<true, PixelFormat::ABGR8UI>,
@ -107,11 +106,9 @@ static constexpr ConversionArray morton_to_linear_fns = {
MortonCopy<true, PixelFormat::Z24S8>,
MortonCopy<true, PixelFormat::S8Z24>,
MortonCopy<true, PixelFormat::Z32FS8>,
// clang-format on
};
static constexpr ConversionArray linear_to_morton_fns = {
// clang-format off
MortonCopy<false, PixelFormat::ABGR8U>,
MortonCopy<false, PixelFormat::ABGR8S>,
MortonCopy<false, PixelFormat::ABGR8UI>,
@ -177,7 +174,6 @@ static constexpr ConversionArray linear_to_morton_fns = {
MortonCopy<false, PixelFormat::Z24S8>,
MortonCopy<false, PixelFormat::S8Z24>,
MortonCopy<false, PixelFormat::Z32FS8>,
// clang-format on
};
static MortonCopyFn GetSwizzleFunction(MortonSwizzleMode mode, Surface::PixelFormat format) {
@ -191,45 +187,6 @@ static MortonCopyFn GetSwizzleFunction(MortonSwizzleMode mode, Surface::PixelFor
return morton_to_linear_fns[static_cast<std::size_t>(format)];
}
/// 8x8 Z-Order coordinate from 2D coordinates
static u32 MortonInterleave(u32 x, u32 y) {
static const u32 xlut[] = {0x00, 0x01, 0x04, 0x05, 0x10, 0x11, 0x14, 0x15};
static const u32 ylut[] = {0x00, 0x02, 0x08, 0x0a, 0x20, 0x22, 0x28, 0x2a};
return xlut[x % 8] + ylut[y % 8];
}
/// Calculates the offset of the position of the pixel in Morton order
static u32 GetMortonOffset(u32 x, u32 y, u32 bytes_per_pixel) {
// Images are split into 8x8 tiles. Each tile is composed of four 4x4 subtiles each
// of which is composed of four 2x2 subtiles each of which is composed of four texels.
// Each structure is embedded into the next-bigger one in a diagonal pattern, e.g.
// texels are laid out in a 2x2 subtile like this:
// 2 3
// 0 1
//
// The full 8x8 tile has the texels arranged like this:
//
// 42 43 46 47 58 59 62 63
// 40 41 44 45 56 57 60 61
// 34 35 38 39 50 51 54 55
// 32 33 36 37 48 49 52 53
// 10 11 14 15 26 27 30 31
// 08 09 12 13 24 25 28 29
// 02 03 06 07 18 19 22 23
// 00 01 04 05 16 17 20 21
//
// This pattern is what's called Z-order curve, or Morton order.
const unsigned int block_height = 8;
const unsigned int coarse_x = x & ~7;
u32 i = MortonInterleave(x, y);
const unsigned int offset = coarse_x * block_height;
return (i + offset) * bytes_per_pixel;
}
static u32 MortonInterleave128(u32 x, u32 y) {
// 128x128 Z-Order coordinate from 2D coordinates
static constexpr u32 xlut[] = {
@ -325,14 +282,14 @@ static u32 GetMortonOffset128(u32 x, u32 y, u32 bytes_per_pixel) {
void MortonSwizzle(MortonSwizzleMode mode, Surface::PixelFormat format, u32 stride,
u32 block_height, u32 height, u32 block_depth, u32 depth, u32 tile_width_spacing,
u8* buffer, std::size_t buffer_size, VAddr addr) {
u8* buffer, VAddr addr) {
GetSwizzleFunction(mode, format)(stride, block_height, height, block_depth, depth,
tile_width_spacing, buffer, buffer_size, addr);
tile_width_spacing, buffer, addr);
}
void MortonCopyPixels128(u32 width, u32 height, u32 bytes_per_pixel, u32 linear_bytes_per_pixel,
u8* morton_data, u8* linear_data, bool morton_to_linear) {
void MortonCopyPixels128(MortonSwizzleMode mode, u32 width, u32 height, u32 bytes_per_pixel,
u32 linear_bytes_per_pixel, u8* morton_data, u8* linear_data) {
const bool morton_to_linear = mode == MortonSwizzleMode::MortonToLinear;
u8* data_ptrs[2];
for (u32 y = 0; y < height; ++y) {
for (u32 x = 0; x < width; ++x) {

6
src/video_core/morton.h
View file

@ -13,9 +13,9 @@ enum class MortonSwizzleMode { MortonToLinear, LinearToMorton };
void MortonSwizzle(MortonSwizzleMode mode, VideoCore::Surface::PixelFormat format, u32 stride,
u32 block_height, u32 height, u32 block_depth, u32 depth, u32 tile_width_spacing,
u8* buffer, std::size_t buffer_size, VAddr addr);
u8* buffer, VAddr addr);
void MortonCopyPixels128(u32 width, u32 height, u32 bytes_per_pixel, u32 linear_bytes_per_pixel,
u8* morton_data, u8* linear_data, bool morton_to_linear);
void MortonCopyPixels128(MortonSwizzleMode mode, u32 width, u32 height, u32 bytes_per_pixel,
u32 linear_bytes_per_pixel, u8* morton_data, u8* linear_data);
} // namespace VideoCore

4
src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
View file

@ -446,7 +446,7 @@ void SwizzleFunc(const MortonSwizzleMode& mode, const SurfaceParams& params,
MortonSwizzle(mode, params.pixel_format, params.MipWidth(mip_level),
params.MipBlockHeight(mip_level), params.MipHeight(mip_level),
params.MipBlockDepth(mip_level), 1, params.tile_width_spacing,
gl_buffer.data() + offset_gl, gl_size, params.addr + offset);
gl_buffer.data() + offset_gl, params.addr + offset);
offset += layer_size;
offset_gl += gl_size;
}
@ -455,7 +455,7 @@ void SwizzleFunc(const MortonSwizzleMode& mode, const SurfaceParams& params,
MortonSwizzle(mode, params.pixel_format, params.MipWidth(mip_level),
params.MipBlockHeight(mip_level), params.MipHeight(mip_level),
params.MipBlockDepth(mip_level), depth, params.tile_width_spacing,
gl_buffer.data(), gl_buffer.size(), params.addr + offset);
gl_buffer.data(), params.addr + offset);
}
}

8
src/video_core/renderer_opengl/renderer_opengl.cpp
View file

@ -167,9 +167,11 @@ void RendererOpenGL::LoadFBToScreenInfo(const Tegra::FramebufferConfig& framebuf
Memory::RasterizerFlushVirtualRegion(framebuffer_addr, size_in_bytes,
Memory::FlushMode::Flush);
VideoCore::MortonCopyPixels128(framebuffer.width, framebuffer.height, bytes_per_pixel, 4,
Memory::GetPointer(framebuffer_addr),
gl_framebuffer_data.data(), true);
constexpr u32 linear_bpp = 4;
VideoCore::MortonCopyPixels128(VideoCore::MortonSwizzleMode::MortonToLinear,
framebuffer.width, framebuffer.height, bytes_per_pixel,
linear_bpp, Memory::GetPointer(framebuffer_addr),
gl_framebuffer_data.data());
glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(framebuffer.stride));