summaryrefslogtreecommitdiffstats
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.cpp14
-rw-r--r--src/video_core/textures/convert.h7
-rw-r--r--src/video_core/textures/decoders.cpp54
-rw-r--r--src/video_core/textures/decoders.h7
-rw-r--r--src/video_core/textures/texture.h31
5 files changed, 71 insertions, 42 deletions
diff --git a/src/video_core/textures/convert.cpp b/src/video_core/textures/convert.cpp
index 82050bd51..f3efa7eb0 100644
--- a/src/video_core/textures/convert.cpp
+++ b/src/video_core/textures/convert.cpp
@@ -62,19 +62,19 @@ static void ConvertZ24S8ToS8Z24(u8* data, u32 width, u32 height) {
SwapS8Z24ToZ24S8<true>(data, width, height);
}
-void ConvertFromGuestToHost(u8* data, PixelFormat pixel_format, u32 width, u32 height, u32 depth,
- bool convert_astc, bool convert_s8z24) {
+void ConvertFromGuestToHost(u8* in_data, u8* out_data, PixelFormat pixel_format, u32 width,
+ u32 height, u32 depth, bool convert_astc, bool convert_s8z24) {
if (convert_astc && IsPixelFormatASTC(pixel_format)) {
// 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);
- const std::vector<u8> rgba8_data =
- Tegra::Texture::ASTC::Decompress(data, width, height, depth, block_width, block_height);
- std::copy(rgba8_data.begin(), rgba8_data.end(), data);
+ const std::vector<u8> rgba8_data = Tegra::Texture::ASTC::Decompress(
+ 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) {
- Tegra::Texture::ConvertS8Z24ToZ24S8(data, width, height);
+ Tegra::Texture::ConvertS8Z24ToZ24S8(in_data, width, height);
}
}
@@ -90,4 +90,4 @@ void ConvertFromHostToGuest(u8* data, PixelFormat pixel_format, u32 width, u32 h
}
}
-} // namespace Tegra::Texture \ No newline at end of file
+} // namespace Tegra::Texture
diff --git a/src/video_core/textures/convert.h b/src/video_core/textures/convert.h
index 12542e71c..d5d6c77bb 100644
--- a/src/video_core/textures/convert.h
+++ b/src/video_core/textures/convert.h
@@ -12,10 +12,11 @@ enum class PixelFormat;
namespace Tegra::Texture {
-void ConvertFromGuestToHost(u8* data, VideoCore::Surface::PixelFormat pixel_format, u32 width,
- u32 height, u32 depth, bool convert_astc, bool convert_s8z24);
+void ConvertFromGuestToHost(u8* in_data, u8* out_data, VideoCore::Surface::PixelFormat pixel_format,
+ u32 width, u32 height, u32 depth, bool convert_astc,
+ bool convert_s8z24);
void ConvertFromHostToGuest(u8* data, VideoCore::Surface::PixelFormat pixel_format, u32 width,
u32 height, u32 depth, bool convert_astc, bool convert_s8z24);
-} // namespace Tegra::Texture \ No newline at end of file
+} // namespace Tegra::Texture
diff --git a/src/video_core/textures/decoders.cpp b/src/video_core/textures/decoders.cpp
index 217805386..7df5f1452 100644
--- a/src/video_core/textures/decoders.cpp
+++ b/src/video_core/textures/decoders.cpp
@@ -36,10 +36,16 @@ struct alignas(64) SwizzleTable {
std::array<std::array<u16, M>, N> values{};
};
-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;
+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 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>();
@@ -171,14 +177,16 @@ void SwizzledData(u8* const swizzled_data, u8* const unswizzled_data, const bool
void CopySwizzledData(u32 width, u32 height, u32 depth, u32 bytes_per_pixel,
u32 out_bytes_per_pixel, u8* const swizzled_data, u8* const unswizzled_data,
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) {
SwizzledData<true>(swizzled_data, unswizzled_data, unswizzle, width, height, depth,
- bytes_per_pixel, out_bytes_per_pixel, block_height, block_depth,
- width_spacing);
+ bytes_per_pixel, out_bytes_per_pixel, block_height_size,
+ block_depth_size, width_spacing);
} else {
SwizzledData<false>(swizzled_data, unswizzled_data, unswizzle, width, height, depth,
- bytes_per_pixel, out_bytes_per_pixel, block_height, block_depth,
- width_spacing);
+ bytes_per_pixel, out_bytes_per_pixel, block_height_size,
+ block_depth_size, width_spacing);
}
}
@@ -248,18 +256,22 @@ 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 block_height) {
+ 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;
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 =
- (line / (gob_size_y * block_height)) * gob_size * block_height * image_width_in_gobs +
- ((line % (gob_size_y * block_height)) / gob_size_y) * gob_size;
- const auto& table = legacy_swizzle_table[line % 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 + (x * bytes_per_pixel / gob_size_x) * gob_size * block_height;
- const u32 swizzled_offset = gob_address + table[(x * bytes_per_pixel) % gob_size_x];
+ 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;
@@ -269,8 +281,9 @@ void SwizzleSubrect(u32 subrect_width, u32 subrect_height, u32 source_pitch, u32
}
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) {
+ 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 +
@@ -289,8 +302,9 @@ void UnswizzleSubrect(u32 subrect_width, u32 subrect_height, u32 dest_pitch, u32
}
void SwizzleKepler(const u32 width, const u32 height, const u32 dst_x, const u32 dst_y,
- const u32 block_height, const std::size_t copy_size, const u8* source_data,
+ 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};
std::size_t count = 0;
for (std::size_t y = dst_y; y < height && count < copy_size; ++y) {
@@ -356,9 +370,9 @@ std::vector<u8> DecodeTexture(const std::vector<u8>& texture_data, TextureFormat
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::AlignUp(width * bytes_per_pixel, gob_size_x);
- const u32 aligned_height = Common::AlignUp(height, gob_size_y * block_height);
- const u32 aligned_depth = Common::AlignUp(depth, gob_size_z * 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;
diff --git a/src/video_core/textures/decoders.h b/src/video_core/textures/decoders.h
index e072d8401..f1e3952bc 100644
--- a/src/video_core/textures/decoders.h
+++ b/src/video_core/textures/decoders.h
@@ -12,8 +12,8 @@ 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;
+inline std::size_t GetGOBSizeShift() {
+ return 9;
}
/// Unswizzles a swizzled texture without changing its format.
@@ -44,7 +44,8 @@ std::size_t CalculateSize(bool tiled, u32 bytes_per_pixel, u32 width, u32 height
/// 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 bytes_per_pixel, u8* swizzled_data, u8* unswizzled_data, u32 block_height,
+ 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,
diff --git a/src/video_core/textures/texture.h b/src/video_core/textures/texture.h
index 219bfd559..e36bc2c04 100644
--- a/src/video_core/textures/texture.h
+++ b/src/video_core/textures/texture.h
@@ -52,9 +52,9 @@ enum class TextureFormat : u32 {
DXT45 = 0x26,
DXN1 = 0x27,
DXN2 = 0x28,
- Z24S8 = 0x29,
+ S8Z24 = 0x29,
X8Z24 = 0x2a,
- S8Z24 = 0x2b,
+ Z24S8 = 0x2b,
X4V4Z24__COV4R4V = 0x2c,
X4V4Z24__COV8R8V = 0x2d,
V8Z24__COV4R12V = 0x2e,
@@ -172,12 +172,16 @@ struct TICEntry {
BitField<26, 1, u32> use_header_opt_control;
BitField<27, 1, u32> depth_texture;
BitField<28, 4, u32> max_mip_level;
+
+ BitField<0, 16, u32> buffer_high_width_minus_one;
};
union {
BitField<0, 16, u32> width_minus_1;
BitField<22, 1, u32> srgb_conversion;
BitField<23, 4, TextureType> texture_type;
BitField<29, 3, u32> border_size;
+
+ BitField<0, 16, u32> buffer_low_width_minus_one;
};
union {
BitField<0, 16, u32> height_minus_1;
@@ -206,7 +210,10 @@ struct TICEntry {
}
u32 Width() const {
- return width_minus_1 + 1;
+ if (header_version != TICHeaderVersion::OneDBuffer) {
+ return width_minus_1 + 1;
+ }
+ return ((buffer_high_width_minus_one << 16) | buffer_low_width_minus_one) + 1;
}
u32 Height() const {
@@ -219,20 +226,17 @@ struct TICEntry {
u32 BlockWidth() const {
ASSERT(IsTiled());
- // The block height is stored in log2 format.
- return 1 << block_width;
+ return block_width;
}
u32 BlockHeight() const {
ASSERT(IsTiled());
- // The block height is stored in log2 format.
- return 1 << block_height;
+ return block_height;
}
u32 BlockDepth() const {
ASSERT(IsTiled());
- // The block height is stored in log2 format.
- return 1 << block_depth;
+ return block_depth;
}
bool IsTiled() const {
@@ -240,6 +244,15 @@ struct TICEntry {
header_version == TICHeaderVersion::BlockLinearColorKey;
}
+ bool IsLineal() const {
+ return header_version == TICHeaderVersion::Pitch ||
+ header_version == TICHeaderVersion::PitchColorKey;
+ }
+
+ bool IsBuffer() const {
+ return header_version == TICHeaderVersion::OneDBuffer;
+ }
+
bool IsSrgbConversionEnabled() const {
return srgb_conversion != 0;
}
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-08 00:04:45 +0100