aboutsummaryrefslogtreecommitdiff
path: root/src/video_core/textures
diff options
context:
space:
mode:
Diffstat (limited to 'src/video_core/textures')
-rw-r--r--src/video_core/textures/convert.cpp6
-rw-r--r--src/video_core/textures/decoders.cpp249
-rw-r--r--src/video_core/textures/decoders.h47
-rw-r--r--src/video_core/textures/texture.cpp2
-rw-r--r--src/video_core/textures/texture.h49
5 files changed, 150 insertions, 203 deletions
diff --git a/src/video_core/textures/convert.cpp b/src/video_core/textures/convert.cpp
index f3efa7eb0..962921483 100644
--- a/src/video_core/textures/convert.cpp
+++ b/src/video_core/textures/convert.cpp
@@ -35,7 +35,7 @@ void SwapS8Z24ToZ24S8(u8* data, u32 width, u32 height) {
S8Z24 s8z24_pixel{};
Z24S8 z24s8_pixel{};
constexpr auto bpp{
- VideoCore::Surface::GetBytesPerPixel(VideoCore::Surface::PixelFormat::S8Z24)};
+ VideoCore::Surface::GetBytesPerPixel(VideoCore::Surface::PixelFormat::S8_UINT_D24_UNORM)};
for (std::size_t y = 0; y < height; ++y) {
for (std::size_t x = 0; x < width; ++x) {
const std::size_t offset{bpp * (y * width + x)};
@@ -73,7 +73,7 @@ void ConvertFromGuestToHost(u8* in_data, u8* out_data, PixelFormat pixel_format,
in_data, width, height, depth, block_width, block_height);
std::copy(rgba8_data.begin(), rgba8_data.end(), out_data);
- } else if (convert_s8z24 && pixel_format == PixelFormat::S8Z24) {
+ } else if (convert_s8z24 && pixel_format == PixelFormat::S8_UINT_D24_UNORM) {
Tegra::Texture::ConvertS8Z24ToZ24S8(in_data, width, height);
}
}
@@ -85,7 +85,7 @@ void ConvertFromHostToGuest(u8* data, PixelFormat pixel_format, u32 width, u32 h
static_cast<u32>(pixel_format));
UNREACHABLE();
- } else if (convert_s8z24 && pixel_format == PixelFormat::S8Z24) {
+ } else if (convert_s8z24 && pixel_format == PixelFormat::S8_UINT_D24_UNORM) {
Tegra::Texture::ConvertZ24S8ToS8Z24(data, width, height);
}
}
diff --git a/src/video_core/textures/decoders.cpp b/src/video_core/textures/decoders.cpp
index 548e4c3fe..16d46a018 100644
--- a/src/video_core/textures/decoders.cpp
+++ b/src/video_core/textures/decoders.cpp
@@ -6,6 +6,7 @@
#include <cstring>
#include "common/alignment.h"
#include "common/assert.h"
+#include "common/bit_util.h"
#include "video_core/gpu.h"
#include "video_core/textures/decoders.h"
#include "video_core/textures/texture.h"
@@ -37,20 +38,10 @@ struct alignas(64) SwizzleTable {
std::array<std::array<u16, M>, N> values{};
};
-constexpr u32 gob_size_x_shift = 6;
-constexpr u32 gob_size_y_shift = 3;
-constexpr u32 gob_size_z_shift = 0;
-constexpr u32 gob_size_shift = gob_size_x_shift + gob_size_y_shift + gob_size_z_shift;
+constexpr u32 FAST_SWIZZLE_ALIGN = 16;
-constexpr u32 gob_size_x = 1U << gob_size_x_shift;
-constexpr u32 gob_size_y = 1U << gob_size_y_shift;
-constexpr u32 gob_size_z = 1U << gob_size_z_shift;
-constexpr u32 gob_size = 1U << gob_size_shift;
-
-constexpr u32 fast_swizzle_align = 16;
-
-constexpr auto legacy_swizzle_table = SwizzleTable<gob_size_y, gob_size_x, gob_size_z>();
-constexpr auto fast_swizzle_table = SwizzleTable<gob_size_y, 4, fast_swizzle_align>();
+constexpr auto LEGACY_SWIZZLE_TABLE = SwizzleTable<GOB_SIZE_X, GOB_SIZE_X, GOB_SIZE_Z>();
+constexpr auto FAST_SWIZZLE_TABLE = SwizzleTable<GOB_SIZE_Y, 4, FAST_SWIZZLE_ALIGN>();
/**
* This function manages ALL the GOBs(Group of Bytes) Inside a single block.
@@ -69,17 +60,17 @@ void PreciseProcessBlock(u8* const swizzled_data, u8* const unswizzled_data, con
u32 y_address = z_address;
u32 pixel_base = layer_z * z + y_start * stride_x;
for (u32 y = y_start; y < y_end; y++) {
- const auto& table = legacy_swizzle_table[y % gob_size_y];
+ const auto& table = LEGACY_SWIZZLE_TABLE[y % GOB_SIZE_Y];
for (u32 x = x_start; x < x_end; x++) {
- const u32 swizzle_offset{y_address + table[x * bytes_per_pixel % gob_size_x]};
+ const u32 swizzle_offset{y_address + table[x * bytes_per_pixel % GOB_SIZE_X]};
const u32 pixel_index{x * out_bytes_per_pixel + pixel_base};
data_ptrs[unswizzle] = swizzled_data + swizzle_offset;
data_ptrs[!unswizzle] = unswizzled_data + pixel_index;
std::memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel);
}
pixel_base += stride_x;
- if ((y + 1) % gob_size_y == 0)
- y_address += gob_size;
+ if ((y + 1) % GOB_SIZE_Y == 0)
+ y_address += GOB_SIZE;
}
z_address += xy_block_size;
}
@@ -104,18 +95,18 @@ void FastProcessBlock(u8* const swizzled_data, u8* const unswizzled_data, const
u32 y_address = z_address;
u32 pixel_base = layer_z * z + y_start * stride_x;
for (u32 y = y_start; y < y_end; y++) {
- const auto& table = fast_swizzle_table[y % gob_size_y];
- for (u32 xb = x_startb; xb < x_endb; xb += fast_swizzle_align) {
- const u32 swizzle_offset{y_address + table[(xb / fast_swizzle_align) % 4]};
+ const auto& table = FAST_SWIZZLE_TABLE[y % GOB_SIZE_Y];
+ for (u32 xb = x_startb; xb < x_endb; xb += FAST_SWIZZLE_ALIGN) {
+ const u32 swizzle_offset{y_address + table[(xb / FAST_SWIZZLE_ALIGN) % 4]};
const u32 out_x = xb * out_bytes_per_pixel / bytes_per_pixel;
const u32 pixel_index{out_x + pixel_base};
data_ptrs[unswizzle ? 1 : 0] = swizzled_data + swizzle_offset;
data_ptrs[unswizzle ? 0 : 1] = unswizzled_data + pixel_index;
- std::memcpy(data_ptrs[0], data_ptrs[1], fast_swizzle_align);
+ std::memcpy(data_ptrs[0], data_ptrs[1], FAST_SWIZZLE_ALIGN);
}
pixel_base += stride_x;
- if ((y + 1) % gob_size_y == 0)
- y_address += gob_size;
+ if ((y + 1) % GOB_SIZE_Y == 0)
+ y_address += GOB_SIZE;
}
z_address += xy_block_size;
}
@@ -138,9 +129,9 @@ void SwizzledData(u8* const swizzled_data, u8* const unswizzled_data, const bool
auto div_ceil = [](const u32 x, const u32 y) { return ((x + y - 1) / y); };
const u32 stride_x = width * out_bytes_per_pixel;
const u32 layer_z = height * stride_x;
- const u32 gob_elements_x = gob_size_x / bytes_per_pixel;
- constexpr u32 gob_elements_y = gob_size_y;
- constexpr u32 gob_elements_z = gob_size_z;
+ const u32 gob_elements_x = GOB_SIZE_X / bytes_per_pixel;
+ constexpr u32 gob_elements_y = GOB_SIZE_Y;
+ constexpr u32 gob_elements_z = GOB_SIZE_Z;
const u32 block_x_elements = gob_elements_x;
const u32 block_y_elements = gob_elements_y * block_height;
const u32 block_z_elements = gob_elements_z * block_depth;
@@ -148,7 +139,7 @@ void SwizzledData(u8* const swizzled_data, u8* const unswizzled_data, const bool
const u32 blocks_on_x = div_ceil(aligned_width, block_x_elements);
const u32 blocks_on_y = div_ceil(height, block_y_elements);
const u32 blocks_on_z = div_ceil(depth, block_z_elements);
- const u32 xy_block_size = gob_size * block_height;
+ const u32 xy_block_size = GOB_SIZE * block_height;
const u32 block_size = xy_block_size * block_depth;
u32 tile_offset = 0;
for (u32 zb = 0; zb < blocks_on_z; zb++) {
@@ -182,7 +173,7 @@ void CopySwizzledData(u32 width, u32 height, u32 depth, u32 bytes_per_pixel,
bool unswizzle, u32 block_height, u32 block_depth, u32 width_spacing) {
const u32 block_height_size{1U << block_height};
const u32 block_depth_size{1U << block_depth};
- if (bytes_per_pixel % 3 != 0 && (width * bytes_per_pixel) % fast_swizzle_align == 0) {
+ if (bytes_per_pixel % 3 != 0 && (width * bytes_per_pixel) % FAST_SWIZZLE_ALIGN == 0) {
SwizzledData<true>(swizzled_data, unswizzled_data, unswizzle, width, height, depth,
bytes_per_pixel, out_bytes_per_pixel, block_height_size,
block_depth_size, width_spacing);
@@ -193,53 +184,6 @@ void CopySwizzledData(u32 width, u32 height, u32 depth, u32 bytes_per_pixel,
}
}
-u32 BytesPerPixel(TextureFormat format) {
- switch (format) {
- case TextureFormat::DXT1:
- case TextureFormat::DXN1:
- // In this case a 'pixel' actually refers to a 4x4 tile.
- return 8;
- case TextureFormat::DXT23:
- case TextureFormat::DXT45:
- case TextureFormat::DXN2:
- case TextureFormat::BC7U:
- case TextureFormat::BC6H_UF16:
- case TextureFormat::BC6H_SF16:
- // In this case a 'pixel' actually refers to a 4x4 tile.
- return 16;
- case TextureFormat::R32_G32_B32:
- return 12;
- case TextureFormat::ASTC_2D_4X4:
- case TextureFormat::ASTC_2D_5X4:
- case TextureFormat::ASTC_2D_8X8:
- case TextureFormat::ASTC_2D_8X5:
- case TextureFormat::ASTC_2D_10X8:
- case TextureFormat::ASTC_2D_5X5:
- case TextureFormat::A8R8G8B8:
- case TextureFormat::A2B10G10R10:
- case TextureFormat::BF10GF11RF11:
- case TextureFormat::R32:
- case TextureFormat::R16_G16:
- return 4;
- case TextureFormat::A1B5G5R5:
- case TextureFormat::B5G6R5:
- case TextureFormat::G8R8:
- case TextureFormat::R16:
- return 2;
- case TextureFormat::R8:
- return 1;
- case TextureFormat::R16_G16_B16_A16:
- return 8;
- case TextureFormat::R32_G32_B32_A32:
- return 16;
- case TextureFormat::R32_G32:
- return 8;
- default:
- UNIMPLEMENTED_MSG("Format not implemented");
- return 1;
- }
-}
-
void UnswizzleTexture(u8* const unswizzled_data, u8* address, u32 tile_size_x, u32 tile_size_y,
u32 bytes_per_pixel, u32 width, u32 height, u32 depth, u32 block_height,
u32 block_depth, u32 width_spacing) {
@@ -259,47 +203,82 @@ std::vector<u8> UnswizzleTexture(u8* address, u32 tile_size_x, u32 tile_size_y,
}
void SwizzleSubrect(u32 subrect_width, u32 subrect_height, u32 source_pitch, u32 swizzled_width,
- u32 bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data,
+ u32 bytes_per_pixel, u8* swizzled_data, const u8* unswizzled_data,
u32 block_height_bit, u32 offset_x, u32 offset_y) {
const u32 block_height = 1U << block_height_bit;
- const u32 image_width_in_gobs{(swizzled_width * bytes_per_pixel + (gob_size_x - 1)) /
- gob_size_x};
+ const u32 image_width_in_gobs =
+ (swizzled_width * bytes_per_pixel + (GOB_SIZE_X - 1)) / GOB_SIZE_X;
for (u32 line = 0; line < subrect_height; ++line) {
const u32 dst_y = line + offset_y;
const u32 gob_address_y =
- (dst_y / (gob_size_y * block_height)) * gob_size * block_height * image_width_in_gobs +
- ((dst_y % (gob_size_y * block_height)) / gob_size_y) * gob_size;
- const auto& table = legacy_swizzle_table[dst_y % gob_size_y];
+ (dst_y / (GOB_SIZE_Y * block_height)) * GOB_SIZE * block_height * image_width_in_gobs +
+ ((dst_y % (GOB_SIZE_Y * block_height)) / GOB_SIZE_Y) * GOB_SIZE;
+ const auto& table = LEGACY_SWIZZLE_TABLE[dst_y % GOB_SIZE_Y];
for (u32 x = 0; x < subrect_width; ++x) {
const u32 dst_x = x + offset_x;
const u32 gob_address =
- gob_address_y + (dst_x * bytes_per_pixel / gob_size_x) * gob_size * block_height;
- const u32 swizzled_offset = gob_address + table[(dst_x * bytes_per_pixel) % gob_size_x];
- u8* source_line = unswizzled_data + line * source_pitch + x * bytes_per_pixel;
- u8* dest_addr = swizzled_data + swizzled_offset;
+ gob_address_y + (dst_x * bytes_per_pixel / GOB_SIZE_X) * GOB_SIZE * block_height;
+ const u32 swizzled_offset = gob_address + table[(dst_x * bytes_per_pixel) % GOB_SIZE_X];
+ const u32 unswizzled_offset = line * source_pitch + x * bytes_per_pixel;
+ const u8* const source_line = unswizzled_data + unswizzled_offset;
+ u8* const dest_addr = swizzled_data + swizzled_offset;
std::memcpy(dest_addr, source_line, bytes_per_pixel);
}
}
}
-void UnswizzleSubrect(u32 subrect_width, u32 subrect_height, u32 dest_pitch, u32 swizzled_width,
- u32 bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data,
- u32 block_height_bit, u32 offset_x, u32 offset_y) {
- const u32 block_height = 1U << block_height_bit;
- for (u32 line = 0; line < subrect_height; ++line) {
- const u32 y2 = line + offset_y;
- const u32 gob_address_y = (y2 / (gob_size_y * block_height)) * gob_size * block_height +
- ((y2 % (gob_size_y * block_height)) / gob_size_y) * gob_size;
- const auto& table = legacy_swizzle_table[y2 % gob_size_y];
- for (u32 x = 0; x < subrect_width; ++x) {
- const u32 x2 = (x + offset_x) * bytes_per_pixel;
- const u32 gob_address = gob_address_y + (x2 / gob_size_x) * gob_size * block_height;
- const u32 swizzled_offset = gob_address + table[x2 % gob_size_x];
- u8* dest_line = unswizzled_data + line * dest_pitch + x * bytes_per_pixel;
- u8* source_addr = swizzled_data + swizzled_offset;
+void UnswizzleSubrect(u32 line_length_in, u32 line_count, u32 pitch, u32 width, u32 bytes_per_pixel,
+ u32 block_height, u32 origin_x, u32 origin_y, u8* output, const u8* input) {
+ const u32 stride = width * bytes_per_pixel;
+ const u32 gobs_in_x = (stride + GOB_SIZE_X - 1) / GOB_SIZE_X;
+ const u32 block_size = gobs_in_x << (GOB_SIZE_SHIFT + block_height);
+
+ const u32 block_height_mask = (1U << block_height) - 1;
+ const u32 x_shift = static_cast<u32>(GOB_SIZE_SHIFT) + block_height;
+
+ for (u32 line = 0; line < line_count; ++line) {
+ const u32 src_y = line + origin_y;
+ const auto& table = LEGACY_SWIZZLE_TABLE[src_y % GOB_SIZE_Y];
+
+ const u32 block_y = src_y >> GOB_SIZE_Y_SHIFT;
+ const u32 src_offset_y = (block_y >> block_height) * block_size +
+ ((block_y & block_height_mask) << GOB_SIZE_SHIFT);
+ for (u32 column = 0; column < line_length_in; ++column) {
+ const u32 src_x = (column + origin_x) * bytes_per_pixel;
+ const u32 src_offset_x = (src_x >> GOB_SIZE_X_SHIFT) << x_shift;
+
+ const u32 swizzled_offset = src_offset_y + src_offset_x + table[src_x % GOB_SIZE_X];
+ const u32 unswizzled_offset = line * pitch + column * bytes_per_pixel;
- std::memcpy(dest_line, source_addr, bytes_per_pixel);
+ std::memcpy(output + unswizzled_offset, input + swizzled_offset, bytes_per_pixel);
+ }
+ }
+}
+
+void SwizzleSliceToVoxel(u32 line_length_in, u32 line_count, u32 pitch, u32 width, u32 height,
+ u32 bytes_per_pixel, u32 block_height, u32 block_depth, u32 origin_x,
+ u32 origin_y, u8* output, const u8* input) {
+ UNIMPLEMENTED_IF(origin_x > 0);
+ UNIMPLEMENTED_IF(origin_y > 0);
+
+ const u32 stride = width * bytes_per_pixel;
+ const u32 gobs_in_x = (stride + GOB_SIZE_X - 1) / GOB_SIZE_X;
+ const u32 block_size = gobs_in_x << (GOB_SIZE_SHIFT + block_height + block_depth);
+
+ const u32 block_height_mask = (1U << block_height) - 1;
+ const u32 x_shift = static_cast<u32>(GOB_SIZE_SHIFT) + block_height + block_depth;
+
+ for (u32 line = 0; line < line_count; ++line) {
+ const auto& table = LEGACY_SWIZZLE_TABLE[line % GOB_SIZE_Y];
+ const u32 block_y = line / GOB_SIZE_Y;
+ const u32 dst_offset_y =
+ (block_y >> block_height) * block_size + (block_y & block_height_mask) * GOB_SIZE;
+ for (u32 x = 0; x < line_length_in; ++x) {
+ const u32 dst_offset =
+ ((x / GOB_SIZE_X) << x_shift) + dst_offset_y + table[x % GOB_SIZE_X];
+ const u32 src_offset = x * bytes_per_pixel + line * pitch;
+ std::memcpy(output + dst_offset, input + src_offset, bytes_per_pixel);
}
}
}
@@ -308,17 +287,17 @@ void SwizzleKepler(const u32 width, const u32 height, const u32 dst_x, const u32
const u32 block_height_bit, const std::size_t copy_size, const u8* source_data,
u8* swizzle_data) {
const u32 block_height = 1U << block_height_bit;
- const u32 image_width_in_gobs{(width + gob_size_x - 1) / gob_size_x};
+ const u32 image_width_in_gobs{(width + GOB_SIZE_X - 1) / GOB_SIZE_X};
std::size_t count = 0;
for (std::size_t y = dst_y; y < height && count < copy_size; ++y) {
const std::size_t gob_address_y =
- (y / (gob_size_y * block_height)) * gob_size * block_height * image_width_in_gobs +
- ((y % (gob_size_y * block_height)) / gob_size_y) * gob_size;
- const auto& table = legacy_swizzle_table[y % gob_size_y];
+ (y / (GOB_SIZE_Y * block_height)) * GOB_SIZE * block_height * image_width_in_gobs +
+ ((y % (GOB_SIZE_Y * block_height)) / GOB_SIZE_Y) * GOB_SIZE;
+ const auto& table = LEGACY_SWIZZLE_TABLE[y % GOB_SIZE_Y];
for (std::size_t x = dst_x; x < width && count < copy_size; ++x) {
const std::size_t gob_address =
- gob_address_y + (x / gob_size_x) * gob_size * block_height;
- const std::size_t swizzled_offset = gob_address + table[x % gob_size_x];
+ gob_address_y + (x / GOB_SIZE_X) * GOB_SIZE * block_height;
+ const std::size_t swizzled_offset = gob_address + table[x % GOB_SIZE_X];
const u8* source_line = source_data + count;
u8* dest_addr = swizzle_data + swizzled_offset;
count++;
@@ -328,54 +307,12 @@ void SwizzleKepler(const u32 width, const u32 height, const u32 dst_x, const u32
}
}
-std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat format, u32 width,
- u32 height) {
- std::vector<u8> rgba_data;
-
- // TODO(Subv): Implement.
- switch (format) {
- case TextureFormat::DXT1:
- case TextureFormat::DXT23:
- case TextureFormat::DXT45:
- case TextureFormat::DXN1:
- case TextureFormat::DXN2:
- case TextureFormat::BC7U:
- case TextureFormat::BC6H_UF16:
- case TextureFormat::BC6H_SF16:
- case TextureFormat::ASTC_2D_4X4:
- case TextureFormat::ASTC_2D_8X8:
- case TextureFormat::ASTC_2D_5X5:
- case TextureFormat::ASTC_2D_10X8:
- case TextureFormat::A8R8G8B8:
- case TextureFormat::A2B10G10R10:
- case TextureFormat::A1B5G5R5:
- case TextureFormat::B5G6R5:
- case TextureFormat::R8:
- case TextureFormat::G8R8:
- case TextureFormat::BF10GF11RF11:
- case TextureFormat::R32_G32_B32_A32:
- case TextureFormat::R32_G32:
- case TextureFormat::R32:
- case TextureFormat::R16:
- case TextureFormat::R16_G16:
- case TextureFormat::R32_G32_B32:
- // TODO(Subv): For the time being just forward the same data without any decoding.
- rgba_data = texture_data;
- break;
- default:
- UNIMPLEMENTED_MSG("Format not implemented");
- break;
- }
-
- return rgba_data;
-}
-
std::size_t CalculateSize(bool tiled, u32 bytes_per_pixel, u32 width, u32 height, u32 depth,
u32 block_height, u32 block_depth) {
if (tiled) {
- const u32 aligned_width = Common::AlignBits(width * bytes_per_pixel, gob_size_x_shift);
- const u32 aligned_height = Common::AlignBits(height, gob_size_y_shift + block_height);
- const u32 aligned_depth = Common::AlignBits(depth, gob_size_z_shift + block_depth);
+ const u32 aligned_width = Common::AlignBits(width * bytes_per_pixel, GOB_SIZE_X_SHIFT);
+ const u32 aligned_height = Common::AlignBits(height, GOB_SIZE_Y_SHIFT + block_height);
+ const u32 aligned_depth = Common::AlignBits(depth, GOB_SIZE_Z_SHIFT + block_depth);
return aligned_width * aligned_height * aligned_depth;
} else {
return width * height * depth * bytes_per_pixel;
@@ -386,14 +323,14 @@ u64 GetGOBOffset(u32 width, u32 height, u32 dst_x, u32 dst_y, u32 block_height,
u32 bytes_per_pixel) {
auto div_ceil = [](const u32 x, const u32 y) { return ((x + y - 1) / y); };
const u32 gobs_in_block = 1 << block_height;
- const u32 y_blocks = gob_size_y << block_height;
- const u32 x_per_gob = gob_size_x / bytes_per_pixel;
+ const u32 y_blocks = GOB_SIZE_Y << block_height;
+ const u32 x_per_gob = GOB_SIZE_X / bytes_per_pixel;
const u32 x_blocks = div_ceil(width, x_per_gob);
- const u32 block_size = gob_size * gobs_in_block;
+ const u32 block_size = GOB_SIZE * gobs_in_block;
const u32 stride = block_size * x_blocks;
const u32 base = (dst_y / y_blocks) * stride + (dst_x / x_per_gob) * block_size;
const u32 relative_y = dst_y % y_blocks;
- return base + (relative_y / gob_size_y) * gob_size;
+ return base + (relative_y / GOB_SIZE_Y) * GOB_SIZE;
}
} // namespace Tegra::Texture
diff --git a/src/video_core/textures/decoders.h b/src/video_core/textures/decoders.h
index 06f3ebf87..01e156bc8 100644
--- a/src/video_core/textures/decoders.h
+++ b/src/video_core/textures/decoders.h
@@ -10,15 +10,15 @@
namespace Tegra::Texture {
-// GOBSize constant. Calculated by 64 bytes in x multiplied by 8 y coords, represents
-// an small rect of (64/bytes_per_pixel)X8.
-inline std::size_t GetGOBSize() {
- return 512;
-}
+constexpr u32 GOB_SIZE_X = 64;
+constexpr u32 GOB_SIZE_Y = 8;
+constexpr u32 GOB_SIZE_Z = 1;
+constexpr u32 GOB_SIZE = GOB_SIZE_X * GOB_SIZE_Y * GOB_SIZE_Z;
-inline std::size_t GetGOBSizeShift() {
- return 9;
-}
+constexpr std::size_t GOB_SIZE_X_SHIFT = 6;
+constexpr std::size_t GOB_SIZE_Y_SHIFT = 3;
+constexpr std::size_t GOB_SIZE_Z_SHIFT = 0;
+constexpr std::size_t GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_SHIFT;
/// Unswizzles a swizzled texture without changing its format.
void UnswizzleTexture(u8* unswizzled_data, u8* address, u32 tile_size_x, u32 tile_size_y,
@@ -38,23 +38,36 @@ void CopySwizzledData(u32 width, u32 height, u32 depth, u32 bytes_per_pixel,
u32 out_bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data,
bool unswizzle, u32 block_height, u32 block_depth, u32 width_spacing);
-/// Decodes an unswizzled texture into a A8R8G8B8 texture.
-std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat format, u32 width,
- u32 height);
-
/// This function calculates the correct size of a texture depending if it's tiled or not.
std::size_t CalculateSize(bool tiled, u32 bytes_per_pixel, u32 width, u32 height, u32 depth,
u32 block_height, u32 block_depth);
/// Copies an untiled subrectangle into a tiled surface.
void SwizzleSubrect(u32 subrect_width, u32 subrect_height, u32 source_pitch, u32 swizzled_width,
- u32 bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data, u32 block_height,
- u32 offset_x, u32 offset_y);
+ u32 bytes_per_pixel, u8* swizzled_data, const u8* unswizzled_data,
+ u32 block_height_bit, u32 offset_x, u32 offset_y);
/// Copies a tiled subrectangle into a linear surface.
-void UnswizzleSubrect(u32 subrect_width, u32 subrect_height, u32 dest_pitch, u32 swizzled_width,
- u32 bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data, u32 block_height,
- u32 offset_x, u32 offset_y);
+void UnswizzleSubrect(u32 line_length_in, u32 line_count, u32 pitch, u32 width, u32 bytes_per_pixel,
+ u32 block_height, u32 origin_x, u32 origin_y, u8* output, const u8* input);
+
+/// @brief Swizzles a 2D array of pixels into a 3D texture
+/// @param line_length_in Number of pixels per line
+/// @param line_count Number of lines
+/// @param pitch Number of bytes per line
+/// @param width Width of the swizzled texture
+/// @param height Height of the swizzled texture
+/// @param bytes_per_pixel Number of bytes used per pixel
+/// @param block_height Block height shift
+/// @param block_depth Block depth shift
+/// @param origin_x Column offset in pixels of the swizzled texture
+/// @param origin_y Row offset in pixels of the swizzled texture
+/// @param output Pointer to the pixels of the swizzled texture
+/// @param input Pointer to the 2D array of pixels used as input
+/// @pre input and output points to an array large enough to hold the number of bytes used
+void SwizzleSliceToVoxel(u32 line_length_in, u32 line_count, u32 pitch, u32 width, u32 height,
+ u32 bytes_per_pixel, u32 block_height, u32 block_depth, u32 origin_x,
+ u32 origin_y, u8* output, const u8* input);
void SwizzleKepler(u32 width, u32 height, u32 dst_x, u32 dst_y, u32 block_height,
std::size_t copy_size, const u8* source_data, u8* swizzle_data);
diff --git a/src/video_core/textures/texture.cpp b/src/video_core/textures/texture.cpp
index d1939d744..4171e3ef2 100644
--- a/src/video_core/textures/texture.cpp
+++ b/src/video_core/textures/texture.cpp
@@ -48,7 +48,7 @@ constexpr std::array<float, 256> SRGB_CONVERSION_LUT = {
};
unsigned SettingsMinimumAnisotropy() noexcept {
- switch (static_cast<Anisotropy>(Settings::values.max_anisotropy)) {
+ switch (static_cast<Anisotropy>(Settings::values.max_anisotropy.GetValue())) {
default:
case Anisotropy::Default:
return 1U;
diff --git a/src/video_core/textures/texture.h b/src/video_core/textures/texture.h
index eba05aced..0574fef12 100644
--- a/src/video_core/textures/texture.h
+++ b/src/video_core/textures/texture.h
@@ -12,10 +12,10 @@
namespace Tegra::Texture {
enum class TextureFormat : u32 {
- R32_G32_B32_A32 = 0x01,
- R32_G32_B32 = 0x02,
- R16_G16_B16_A16 = 0x03,
- R32_G32 = 0x04,
+ R32G32B32A32 = 0x01,
+ R32G32B32 = 0x02,
+ R16G16B16A16 = 0x03,
+ R32G32 = 0x04,
R32_B24G8 = 0x05,
ETC2_RGB = 0x06,
X8B8G8R8 = 0x07,
@@ -23,19 +23,19 @@ enum class TextureFormat : u32 {
A2B10G10R10 = 0x09,
ETC2_RGB_PTA = 0x0a,
ETC2_RGBA = 0x0b,
- R16_G16 = 0x0c,
- G8R24 = 0x0d,
- G24R8 = 0x0e,
+ R16G16 = 0x0c,
+ R24G8 = 0x0d,
+ R8G24 = 0x0e,
R32 = 0x0f,
- BC6H_SF16 = 0x10,
- BC6H_UF16 = 0x11,
+ BC6H_SFLOAT = 0x10,
+ BC6H_UFLOAT = 0x11,
A4B4G4R4 = 0x12,
A5B5G5R1 = 0x13,
A1B5G5R5 = 0x14,
B5G6R5 = 0x15,
B6G5R5 = 0x16,
- BC7U = 0x17,
- G8R8 = 0x18,
+ BC7 = 0x17,
+ R8G8 = 0x18,
EAC = 0x19,
EACX2 = 0x1a,
R16 = 0x1b,
@@ -43,23 +43,23 @@ enum class TextureFormat : u32 {
R8 = 0x1d,
G4R4 = 0x1e,
R1 = 0x1f,
- E5B9G9R9_SHAREDEXP = 0x20,
- BF10GF11RF11 = 0x21,
+ E5B9G9R9 = 0x20,
+ B10G11R11 = 0x21,
G8B8G8R8 = 0x22,
B8G8R8G8 = 0x23,
- DXT1 = 0x24,
- DXT23 = 0x25,
- DXT45 = 0x26,
- DXN1 = 0x27,
- DXN2 = 0x28,
- S8Z24 = 0x29,
+ BC1_RGBA = 0x24,
+ BC2 = 0x25,
+ BC3 = 0x26,
+ BC4 = 0x27,
+ BC5 = 0x28,
+ S8D24 = 0x29,
X8Z24 = 0x2a,
- Z24S8 = 0x2b,
+ D24S8 = 0x2b,
X4V4Z24__COV4R4V = 0x2c,
X4V4Z24__COV8R8V = 0x2d,
V8Z24__COV4R12V = 0x2e,
- ZF32 = 0x2f,
- ZF32_X24S8 = 0x30,
+ D32 = 0x2f,
+ D32S8 = 0x30,
X8Z24_X20V4S8__COV4R4V = 0x31,
X8Z24_X20V4S8__COV8R8V = 0x32,
ZF32_X20V4X8__COV4R4V = 0x33,
@@ -69,7 +69,7 @@ enum class TextureFormat : u32 {
X8Z24_X16V8S8__COV4R12V = 0x37,
ZF32_X16V8X8__COV4R12V = 0x38,
ZF32_X16V8S8__COV4R12V = 0x39,
- Z16 = 0x3a,
+ D16 = 0x3a,
V8Z24__COV8R24V = 0x3b,
X8Z24_X16V8S8__COV8R24V = 0x3c,
ZF32_X16V8X8__COV8R24V = 0x3d,
@@ -375,7 +375,4 @@ struct FullTextureInfo {
TSCEntry tsc;
};
-/// Returns the number of bytes per pixel of the input texture format.
-u32 BytesPerPixel(TextureFormat format);
-
} // namespace Tegra::Texture
generated by cgit v1.2.3 (git 2.25.1) at 2025-12-28 00:41:52 +0000