aboutsummaryrefslogtreecommitdiff
path: root/src/video_core
diff options
context:
space:
mode:
Diffstat (limited to 'src/video_core')
-rw-r--r--src/video_core/engines/maxwell_3d.h12
-rw-r--r--src/video_core/engines/maxwell_compute.cpp19
-rw-r--r--src/video_core/engines/maxwell_compute.h36
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.cpp22
-rw-r--r--src/video_core/renderer_opengl/gl_rasterizer.h6
-rw-r--r--src/video_core/textures/decoders.cpp117
6 files changed, 158 insertions, 54 deletions
diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h
index b81b0723d..16cdfc7e2 100644
--- a/src/video_core/engines/maxwell_3d.h
+++ b/src/video_core/engines/maxwell_3d.h
@@ -461,7 +461,11 @@ public:
u32 entry;
} macros;
- INSERT_PADDING_WORDS(0x1B8);
+ INSERT_PADDING_WORDS(0x189);
+
+ u32 tfb_enabled;
+
+ INSERT_PADDING_WORDS(0x2E);
RenderTargetConfig rt[NumRenderTargets];
@@ -594,7 +598,9 @@ public:
u32 depth_write_enabled;
- INSERT_PADDING_WORDS(0x7);
+ u32 alpha_test_enabled;
+
+ INSERT_PADDING_WORDS(0x6);
u32 d3d_cull_mode;
@@ -977,6 +983,7 @@ private:
"Field " #field_name " has invalid position")
ASSERT_REG_POSITION(macros, 0x45);
+ASSERT_REG_POSITION(tfb_enabled, 0x1D1);
ASSERT_REG_POSITION(rt, 0x200);
ASSERT_REG_POSITION(viewport_transform[0], 0x280);
ASSERT_REG_POSITION(viewport, 0x300);
@@ -996,6 +1003,7 @@ ASSERT_REG_POSITION(zeta_height, 0x48b);
ASSERT_REG_POSITION(depth_test_enable, 0x4B3);
ASSERT_REG_POSITION(independent_blend_enable, 0x4B9);
ASSERT_REG_POSITION(depth_write_enabled, 0x4BA);
+ASSERT_REG_POSITION(alpha_test_enabled, 0x4BB);
ASSERT_REG_POSITION(d3d_cull_mode, 0x4C2);
ASSERT_REG_POSITION(depth_test_func, 0x4C3);
ASSERT_REG_POSITION(blend, 0x4CF);
diff --git a/src/video_core/engines/maxwell_compute.cpp b/src/video_core/engines/maxwell_compute.cpp
index e4e5f9e5e..59e28b22d 100644
--- a/src/video_core/engines/maxwell_compute.cpp
+++ b/src/video_core/engines/maxwell_compute.cpp
@@ -2,12 +2,29 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
+#include "common/logging/log.h"
+#include "core/core.h"
#include "video_core/engines/maxwell_compute.h"
namespace Tegra {
namespace Engines {
-void MaxwellCompute::WriteReg(u32 method, u32 value) {}
+void MaxwellCompute::WriteReg(u32 method, u32 value) {
+ ASSERT_MSG(method < Regs::NUM_REGS,
+ "Invalid MaxwellCompute register, increase the size of the Regs structure");
+
+ regs.reg_array[method] = value;
+
+ switch (method) {
+ case MAXWELL_COMPUTE_REG_INDEX(compute): {
+ LOG_CRITICAL(HW_GPU, "Compute shaders are not implemented");
+ UNREACHABLE();
+ break;
+ }
+ default:
+ break;
+ }
+}
} // namespace Engines
} // namespace Tegra
diff --git a/src/video_core/engines/maxwell_compute.h b/src/video_core/engines/maxwell_compute.h
index 2b3e4ced6..6ea934fb9 100644
--- a/src/video_core/engines/maxwell_compute.h
+++ b/src/video_core/engines/maxwell_compute.h
@@ -4,17 +4,53 @@
#pragma once
+#include <array>
+#include "common/assert.h"
+#include "common/bit_field.h"
+#include "common/common_funcs.h"
#include "common/common_types.h"
namespace Tegra::Engines {
+#define MAXWELL_COMPUTE_REG_INDEX(field_name) \
+ (offsetof(Tegra::Engines::MaxwellCompute::Regs, field_name) / sizeof(u32))
+
class MaxwellCompute final {
public:
MaxwellCompute() = default;
~MaxwellCompute() = default;
+ struct Regs {
+ static constexpr std::size_t NUM_REGS = 0xCF8;
+
+ union {
+ struct {
+ INSERT_PADDING_WORDS(0x281);
+
+ union {
+ u32 compute_end;
+ BitField<0, 1, u32> unknown;
+ } compute;
+
+ INSERT_PADDING_WORDS(0xA76);
+ };
+ std::array<u32, NUM_REGS> reg_array;
+ };
+ } regs{};
+
+ static_assert(sizeof(Regs) == Regs::NUM_REGS * sizeof(u32),
+ "MaxwellCompute Regs has wrong size");
+
/// Write the value to the register identified by method.
void WriteReg(u32 method, u32 value);
};
+#define ASSERT_REG_POSITION(field_name, position) \
+ static_assert(offsetof(MaxwellCompute::Regs, field_name) == position * 4, \
+ "Field " #field_name " has invalid position")
+
+ASSERT_REG_POSITION(compute, 0x281);
+
+#undef ASSERT_REG_POSITION
+
} // namespace Tegra::Engines
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index 70fb54507..44850d193 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -450,6 +450,8 @@ void RasterizerOpenGL::DrawArrays() {
SyncBlendState();
SyncLogicOpState();
SyncCullMode();
+ SyncAlphaTest();
+ SyncTransformFeedback();
// TODO(bunnei): Sync framebuffer_scale uniform here
// TODO(bunnei): Sync scissorbox uniform(s) here
@@ -883,4 +885,24 @@ void RasterizerOpenGL::SyncLogicOpState() {
state.logic_op.operation = MaxwellToGL::LogicOp(regs.logic_op.operation);
}
+void RasterizerOpenGL::SyncAlphaTest() {
+ const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+
+ // TODO(Rodrigo): Alpha testing is a legacy OpenGL feature, but it can be
+ // implemented with a test+discard in fragment shaders.
+ if (regs.alpha_test_enabled != 0) {
+ LOG_CRITICAL(Render_OpenGL, "Alpha testing is not implemented");
+ UNREACHABLE();
+ }
+}
+
+void RasterizerOpenGL::SyncTransformFeedback() {
+ const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+
+ if (regs.tfb_enabled != 0) {
+ LOG_CRITICAL(Render_OpenGL, "Transform feedbacks are not implemented");
+ UNREACHABLE();
+ }
+}
+
} // namespace OpenGL
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h
index bf9560bdc..c3f1e14bf 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer.h
@@ -158,6 +158,12 @@ private:
/// Syncs the LogicOp state to match the guest state
void SyncLogicOpState();
+ /// Syncs the alpha test state to match the guest state
+ void SyncAlphaTest();
+
+ /// Syncs the transform feedback state to match the guest state
+ void SyncTransformFeedback();
+
bool has_ARB_direct_state_access = false;
bool has_ARB_multi_bind = false;
bool has_ARB_separate_shader_objects = false;
diff --git a/src/video_core/textures/decoders.cpp b/src/video_core/textures/decoders.cpp
index 20ba6d4f6..3d5476e5d 100644
--- a/src/video_core/textures/decoders.cpp
+++ b/src/video_core/textures/decoders.cpp
@@ -13,47 +13,20 @@
namespace Tegra::Texture {
/**
+ * This table represents the internal swizzle of a gob,
+ * in format 16 bytes x 2 sector packing.
* Calculates the offset of an (x, y) position within a swizzled texture.
- * Taken from the Tegra X1 TRM.
+ * Taken from the Tegra X1 Technical Reference Manual. pages 1187-1188
*/
-static u32 GetSwizzleOffset(u32 x, u32 y, u32 image_width, u32 bytes_per_pixel, u32 block_height) {
- // Round up to the next gob
- const u32 image_width_in_gobs{(image_width * bytes_per_pixel + 63) / 64};
-
- u32 GOB_address = 0 + (y / (8 * block_height)) * 512 * block_height * image_width_in_gobs +
- (x * bytes_per_pixel / 64) * 512 * block_height +
- (y % (8 * block_height) / 8) * 512;
- x *= bytes_per_pixel;
- u32 address = GOB_address + ((x % 64) / 32) * 256 + ((y % 8) / 2) * 64 + ((x % 32) / 16) * 32 +
- (y % 2) * 16 + (x % 16);
-
- return address;
-}
-
-void CopySwizzledData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel,
- u8* swizzled_data, u8* unswizzled_data, bool unswizzle, u32 block_height) {
- u8* data_ptrs[2];
- for (unsigned y = 0; y < height; ++y) {
- for (unsigned x = 0; x < width; ++x) {
- u32 swizzle_offset = GetSwizzleOffset(x, y, width, bytes_per_pixel, block_height);
- u32 pixel_index = (x + y * width) * out_bytes_per_pixel;
-
- 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);
- }
- }
-}
-
-template <std::size_t N, std::size_t M>
+template <std::size_t N, std::size_t M, u32 Align>
struct alignas(64) SwizzleTable {
+ static_assert(M * Align == 64, "Swizzle Table does not align to GOB");
constexpr SwizzleTable() {
for (u32 y = 0; y < N; ++y) {
for (u32 x = 0; x < M; ++x) {
- const u32 x2 = x * 16;
+ const u32 x2 = x * Align;
values[y][x] = static_cast<u16>(((x2 % 64) / 32) * 256 + ((y % 8) / 2) * 64 +
- ((x2 % 32) / 16) * 32 + (y % 2) * 16);
+ ((x2 % 32) / 16) * 32 + (y % 2) * 16 + (x2 % 16));
}
}
}
@@ -63,24 +36,60 @@ struct alignas(64) SwizzleTable {
std::array<std::array<u16, M>, N> values{};
};
-constexpr auto swizzle_table = SwizzleTable<8, 4>();
+constexpr auto legacy_swizzle_table = SwizzleTable<8, 64, 1>();
+constexpr auto fast_swizzle_table = SwizzleTable<8, 4, 16>();
-void FastSwizzleData(u32 width, u32 height, u32 bytes_per_pixel, u8* swizzled_data,
- u8* unswizzled_data, bool unswizzle, u32 block_height) {
+static void LegacySwizzleData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel,
+ u8* swizzled_data, u8* unswizzled_data, bool unswizzle,
+ u32 block_height) {
+ std::array<u8*, 2> data_ptrs;
+ const std::size_t stride = width * bytes_per_pixel;
+ const std::size_t gobs_in_x = 64;
+ const std::size_t gobs_in_y = 8;
+ const std::size_t gobs_size = gobs_in_x * gobs_in_y;
+ const std::size_t image_width_in_gobs{(stride + gobs_in_x - 1) / gobs_in_x};
+ for (std::size_t y = 0; y < height; ++y) {
+ const std::size_t gob_y_address =
+ (y / (gobs_in_y * block_height)) * gobs_size * block_height * image_width_in_gobs +
+ (y % (gobs_in_y * block_height) / gobs_in_y) * gobs_size;
+ const auto& table = legacy_swizzle_table[y % gobs_in_y];
+ for (std::size_t x = 0; x < width; ++x) {
+ const std::size_t gob_address =
+ gob_y_address + (x * bytes_per_pixel / gobs_in_x) * gobs_size * block_height;
+ const std::size_t x2 = x * bytes_per_pixel;
+ const std::size_t swizzle_offset = gob_address + table[x2 % gobs_in_x];
+ const std::size_t pixel_index = (x + y * width) * out_bytes_per_pixel;
+
+ 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);
+ }
+ }
+}
+
+static void FastSwizzleData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel,
+ u8* swizzled_data, u8* unswizzled_data, bool unswizzle,
+ u32 block_height) {
std::array<u8*, 2> data_ptrs;
const std::size_t stride{width * bytes_per_pixel};
- const std::size_t image_width_in_gobs{(stride + 63) / 64};
+ const std::size_t gobs_in_x = 64;
+ const std::size_t gobs_in_y = 8;
+ const std::size_t gobs_size = gobs_in_x * gobs_in_y;
+ const std::size_t image_width_in_gobs{(stride + gobs_in_x - 1) / gobs_in_x};
const std::size_t copy_size{16};
for (std::size_t y = 0; y < height; ++y) {
const std::size_t initial_gob =
- (y / (8 * block_height)) * 512 * block_height * image_width_in_gobs +
- (y % (8 * block_height) / 8) * 512;
- const std::size_t pixel_base{y * width * bytes_per_pixel};
- const auto& table = swizzle_table[y % 8];
+ (y / (gobs_in_y * block_height)) * gobs_size * block_height * image_width_in_gobs +
+ (y % (gobs_in_y * block_height) / gobs_in_y) * gobs_size;
+ const std::size_t pixel_base{y * width * out_bytes_per_pixel};
+ const auto& table = fast_swizzle_table[y % gobs_in_y];
for (std::size_t xb = 0; xb < stride; xb += copy_size) {
- const std::size_t gob_address{initial_gob + (xb / 64) * 512 * block_height};
+ const std::size_t gob_address{initial_gob +
+ (xb / gobs_in_x) * gobs_size * block_height};
const std::size_t swizzle_offset{gob_address + table[(xb / 16) % 4]};
- const std::size_t pixel_index{xb + pixel_base};
+ const std::size_t out_x = xb * out_bytes_per_pixel / bytes_per_pixel;
+ const std::size_t pixel_index{out_x + pixel_base};
data_ptrs[unswizzle] = swizzled_data + swizzle_offset;
data_ptrs[!unswizzle] = unswizzled_data + pixel_index;
std::memcpy(data_ptrs[0], data_ptrs[1], copy_size);
@@ -88,6 +97,17 @@ void FastSwizzleData(u32 width, u32 height, u32 bytes_per_pixel, u8* swizzled_da
}
}
+void CopySwizzledData(u32 width, u32 height, u32 bytes_per_pixel, u32 out_bytes_per_pixel,
+ u8* swizzled_data, u8* unswizzled_data, bool unswizzle, u32 block_height) {
+ if (bytes_per_pixel % 3 != 0 && (width * bytes_per_pixel) % 16 == 0) {
+ FastSwizzleData(width, height, bytes_per_pixel, out_bytes_per_pixel, swizzled_data,
+ unswizzled_data, unswizzle, block_height);
+ } else {
+ LegacySwizzleData(width, height, bytes_per_pixel, out_bytes_per_pixel, swizzled_data,
+ unswizzled_data, unswizzle, block_height);
+ }
+}
+
u32 BytesPerPixel(TextureFormat format) {
switch (format) {
case TextureFormat::DXT1:
@@ -134,13 +154,8 @@ u32 BytesPerPixel(TextureFormat format) {
std::vector<u8> UnswizzleTexture(VAddr address, u32 tile_size, u32 bytes_per_pixel, u32 width,
u32 height, u32 block_height) {
std::vector<u8> unswizzled_data(width * height * bytes_per_pixel);
- if (bytes_per_pixel % 3 != 0 && (width * bytes_per_pixel) % 16 == 0) {
- FastSwizzleData(width / tile_size, height / tile_size, bytes_per_pixel,
- Memory::GetPointer(address), unswizzled_data.data(), true, block_height);
- } else {
- CopySwizzledData(width / tile_size, height / tile_size, bytes_per_pixel, bytes_per_pixel,
- Memory::GetPointer(address), unswizzled_data.data(), true, block_height);
- }
+ CopySwizzledData(width / tile_size, height / tile_size, bytes_per_pixel, bytes_per_pixel,
+ Memory::GetPointer(address), unswizzled_data.data(), true, block_height);
return unswizzled_data;
}
generated by cgit v1.2.3 (git 2.25.1) at 2025-12-27 15:33:04 +0000