summaryrefslogtreecommitdiffstats
path: root/src
diff options
context:
space:
mode:
Diffstat (limited to 'src')
-rw-r--r--src/citra/default_ini.h4
-rw-r--r--src/citra_qt/debugger/graphics_cmdlists.cpp34
-rw-r--r--src/video_core/command_processor.cpp10
-rw-r--r--src/video_core/debug_utils/debug_utils.cpp12
-rw-r--r--src/video_core/debug_utils/debug_utils.h5
-rw-r--r--src/video_core/pica.cpp72
-rw-r--r--src/video_core/pica.h65
-rw-r--r--src/video_core/rasterizer.cpp2
-rw-r--r--src/video_core/vertex_shader.cpp59
9 files changed, 139 insertions, 124 deletions
diff --git a/src/citra/default_ini.h b/src/citra/default_ini.h
index fd5a90d56..1925bece8 100644
--- a/src/citra/default_ini.h
+++ b/src/citra/default_ini.h
@@ -33,10 +33,6 @@ pad_cleft =
pad_cright =
[Core]
-# The refresh rate for the GPU
-# Defaults to 30
-gpu_refresh_rate =
-
# The applied frameskip amount. Must be a power of two.
# 0 (default): No frameskip, 1: x2 frameskip, 2: x4 frameskip, 3: x8 frameskip, etc.
frame_skip =
diff --git a/src/citra_qt/debugger/graphics_cmdlists.cpp b/src/citra_qt/debugger/graphics_cmdlists.cpp
index de10bce1f..7ac3ea542 100644
--- a/src/citra_qt/debugger/graphics_cmdlists.cpp
+++ b/src/citra_qt/debugger/graphics_cmdlists.cpp
@@ -10,6 +10,7 @@
#include <QPushButton>
#include <QVBoxLayout>
#include <QTreeView>
+#include <QHeaderView>
#include <QSpinBox>
#include <QComboBox>
@@ -174,7 +175,7 @@ int GPUCommandListModel::rowCount(const QModelIndex& parent) const {
}
int GPUCommandListModel::columnCount(const QModelIndex& parent) const {
- return 2;
+ return 3;
}
QVariant GPUCommandListModel::data(const QModelIndex& index, int role) const {
@@ -187,14 +188,13 @@ QVariant GPUCommandListModel::data(const QModelIndex& index, int role) const {
if (role == Qt::DisplayRole) {
QString content;
- if (index.column() == 0) {
- QString content = QString::fromLatin1(Pica::Regs::GetCommandName(cmd.cmd_id).c_str());
- content.append(" ");
- return content;
- } else if (index.column() == 1) {
- QString content = QString("%1 ").arg(cmd.hex, 8, 16, QLatin1Char('0'));
- content.append(QString("%1 ").arg(val, 8, 16, QLatin1Char('0')));
- return content;
+ switch ( index.column() ) {
+ case 0:
+ return QString::fromLatin1(Pica::Regs::GetCommandName(cmd.cmd_id).c_str());
+ case 1:
+ return QString("%1").arg(cmd.cmd_id, 3, 16, QLatin1Char('0'));
+ case 2:
+ return QString("%1").arg(val, 8, 16, QLatin1Char('0'));
}
} else if (role == CommandIdRole) {
return QVariant::fromValue<int>(cmd.cmd_id.Value());
@@ -207,10 +207,13 @@ QVariant GPUCommandListModel::headerData(int section, Qt::Orientation orientatio
switch(role) {
case Qt::DisplayRole:
{
- if (section == 0) {
+ switch (section) {
+ case 0:
return tr("Command Name");
- } else if (section == 1) {
- return tr("Data");
+ case 1:
+ return tr("Register");
+ case 2:
+ return tr("New Value");
}
break;
@@ -299,6 +302,13 @@ GPUCommandListWidget::GPUCommandListWidget(QWidget* parent) : QDockWidget(tr("Pi
list_widget->setModel(model);
list_widget->setFont(QFont("monospace"));
list_widget->setRootIsDecorated(false);
+ list_widget->setUniformRowHeights(true);
+
+#if QT_VERSION >= QT_VERSION_CHECK(5,0,0)
+ list_widget->header()->setSectionResizeMode(QHeaderView::ResizeToContents);
+#else
+ list_widget->header()->setResizeMode(QHeaderView::ResizeToContents);
+#endif
connect(list_widget->selectionModel(), SIGNAL(currentChanged(const QModelIndex&,const QModelIndex&)),
this, SLOT(SetCommandInfo(const QModelIndex&)));
diff --git a/src/video_core/command_processor.cpp b/src/video_core/command_processor.cpp
index 43ae06181..ef9584abd 100644
--- a/src/video_core/command_processor.cpp
+++ b/src/video_core/command_processor.cpp
@@ -116,7 +116,9 @@ static inline void WritePicaReg(u32 id, u32 value, u32 mask) {
{
Common::Profiling::ScopeTimer scope_timer(category_drawing);
+#if PICA_LOG_TEV
DebugUtils::DumpTevStageConfig(regs.GetTevStages());
+#endif
if (g_debug_context)
g_debug_context->OnEvent(DebugContext::Event::IncomingPrimitiveBatch, nullptr);
@@ -159,9 +161,11 @@ static inline void WritePicaReg(u32 id, u32 value, u32 mask) {
const u16* index_address_16 = (u16*)index_address_8;
bool index_u16 = index_info.format != 0;
+#if PICA_DUMP_GEOMETRY
DebugUtils::GeometryDumper geometry_dumper;
- PrimitiveAssembler<VertexShader::OutputVertex> primitive_assembler(regs.triangle_topology.Value());
PrimitiveAssembler<DebugUtils::GeometryDumper::Vertex> dumping_primitive_assembler(regs.triangle_topology.Value());
+#endif
+ PrimitiveAssembler<VertexShader::OutputVertex> primitive_assembler(regs.triangle_topology.Value());
if (g_debug_context) {
for (int i = 0; i < 3; ++i) {
@@ -271,6 +275,7 @@ static inline void WritePicaReg(u32 id, u32 value, u32 mask) {
if (g_debug_context)
g_debug_context->OnEvent(DebugContext::Event::VertexLoaded, (void*)&input);
+#if PICA_DUMP_GEOMETRY
// NOTE: When dumping geometry, we simply assume that the first input attribute
// corresponds to the position for now.
DebugUtils::GeometryDumper::Vertex dumped_vertex = {
@@ -280,6 +285,7 @@ static inline void WritePicaReg(u32 id, u32 value, u32 mask) {
dumping_primitive_assembler.SubmitVertex(dumped_vertex,
std::bind(&DebugUtils::GeometryDumper::AddTriangle,
&geometry_dumper, _1, _2, _3));
+#endif
// Send to vertex shader
VertexShader::OutputVertex output = VertexShader::RunShader(input, attribute_config.GetNumTotalAttributes(), g_state.regs.vs, g_state.vs);
@@ -312,7 +318,9 @@ static inline void WritePicaReg(u32 id, u32 value, u32 mask) {
VideoCore::g_renderer->hw_rasterizer->DrawTriangles();
}
+#if PICA_DUMP_GEOMETRY
geometry_dumper.Dump();
+#endif
if (g_debug_context) {
g_debug_context->OnEvent(DebugContext::Event::FinishedPrimitiveBatch, nullptr);
diff --git a/src/video_core/debug_utils/debug_utils.cpp b/src/video_core/debug_utils/debug_utils.cpp
index c3f8321c6..e9a858411 100644
--- a/src/video_core/debug_utils/debug_utils.cpp
+++ b/src/video_core/debug_utils/debug_utils.cpp
@@ -90,10 +90,6 @@ void GeometryDumper::AddTriangle(Vertex& v0, Vertex& v1, Vertex& v2) {
}
void GeometryDumper::Dump() {
- // NOTE: Permanently enabling this just trashes the hard disk for no reason.
- // Hence, this is currently disabled.
- return;
-
static int index = 0;
std::string filename = std::string("geometry_dump") + std::to_string(++index) + ".obj";
@@ -116,10 +112,6 @@ void GeometryDumper::Dump() {
void DumpShader(const u32* binary_data, u32 binary_size, const u32* swizzle_data, u32 swizzle_size,
u32 main_offset, const Regs::VSOutputAttributes* output_attributes)
{
- // NOTE: Permanently enabling this just trashes hard disks for no reason.
- // Hence, this is currently disabled.
- return;
-
struct StuffToWrite {
u8* pointer;
u32 size;
@@ -565,10 +557,6 @@ TextureInfo TextureInfo::FromPicaRegister(const Regs::TextureConfig& config,
}
void DumpTexture(const Pica::Regs::TextureConfig& texture_config, u8* data) {
- // NOTE: Permanently enabling this just trashes hard disks for no reason.
- // Hence, this is currently disabled.
- return;
-
#ifndef HAVE_PNG
return;
#else
diff --git a/src/video_core/debug_utils/debug_utils.h b/src/video_core/debug_utils/debug_utils.h
index 3f109dcb7..81eea30a9 100644
--- a/src/video_core/debug_utils/debug_utils.h
+++ b/src/video_core/debug_utils/debug_utils.h
@@ -157,6 +157,11 @@ extern std::shared_ptr<DebugContext> g_debug_context; // TODO: Get rid of this g
namespace DebugUtils {
+#define PICA_DUMP_GEOMETRY 0
+#define PICA_DUMP_SHADERS 0
+#define PICA_DUMP_TEXTURES 0
+#define PICA_LOG_TEV 0
+
// Simple utility class for dumping geometry data to an OBJ file
class GeometryDumper {
public:
diff --git a/src/video_core/pica.cpp b/src/video_core/pica.cpp
index 543d9c443..17cb66780 100644
--- a/src/video_core/pica.cpp
+++ b/src/video_core/pica.cpp
@@ -2,7 +2,8 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-#include <string.h>
+#include <cstring>
+#include <unordered_map>
#include "pica.h"
@@ -10,6 +11,75 @@ namespace Pica {
State g_state;
+std::string Regs::GetCommandName(int index) {
+ static std::unordered_map<u32, std::string> map;
+
+ if (map.empty()) {
+ #define ADD_FIELD(name) \
+ map.insert({static_cast<u32>(PICA_REG_INDEX(name)), #name}); \
+ /* TODO: change to Regs::name when VS2015 and other compilers support it */ \
+ for (u32 i = PICA_REG_INDEX(name) + 1; i < PICA_REG_INDEX(name) + sizeof(Regs().name) / 4; ++i) \
+ map.insert({i, #name + std::string("+") + std::to_string(i-PICA_REG_INDEX(name))}); \
+
+ ADD_FIELD(trigger_irq);
+ ADD_FIELD(cull_mode);
+ ADD_FIELD(viewport_size_x);
+ ADD_FIELD(viewport_size_y);
+ ADD_FIELD(viewport_depth_range);
+ ADD_FIELD(viewport_depth_far_plane);
+ ADD_FIELD(viewport_corner);
+ ADD_FIELD(texture0_enable);
+ ADD_FIELD(texture0);
+ ADD_FIELD(texture0_format);
+ ADD_FIELD(texture1);
+ ADD_FIELD(texture1_format);
+ ADD_FIELD(texture2);
+ ADD_FIELD(texture2_format);
+ ADD_FIELD(tev_stage0);
+ ADD_FIELD(tev_stage1);
+ ADD_FIELD(tev_stage2);
+ ADD_FIELD(tev_stage3);
+ ADD_FIELD(tev_combiner_buffer_input);
+ ADD_FIELD(tev_stage4);
+ ADD_FIELD(tev_stage5);
+ ADD_FIELD(tev_combiner_buffer_color);
+ ADD_FIELD(output_merger);
+ ADD_FIELD(framebuffer);
+ ADD_FIELD(vertex_attributes);
+ ADD_FIELD(index_array);
+ ADD_FIELD(num_vertices);
+ ADD_FIELD(trigger_draw);
+ ADD_FIELD(trigger_draw_indexed);
+ ADD_FIELD(vs_default_attributes_setup);
+ ADD_FIELD(command_buffer);
+ ADD_FIELD(triangle_topology);
+ ADD_FIELD(gs.bool_uniforms);
+ ADD_FIELD(gs.int_uniforms);
+ ADD_FIELD(gs.main_offset);
+ ADD_FIELD(gs.input_register_map);
+ ADD_FIELD(gs.uniform_setup);
+ ADD_FIELD(gs.program);
+ ADD_FIELD(gs.swizzle_patterns);
+ ADD_FIELD(vs.bool_uniforms);
+ ADD_FIELD(vs.int_uniforms);
+ ADD_FIELD(vs.main_offset);
+ ADD_FIELD(vs.input_register_map);
+ ADD_FIELD(vs.uniform_setup);
+ ADD_FIELD(vs.program);
+ ADD_FIELD(vs.swizzle_patterns);
+
+#undef ADD_FIELD
+ }
+
+ // Return empty string if no match is found
+ auto it = map.find(index);
+ if (it != map.end()) {
+ return it->second;
+ } else {
+ return std::string();
+ }
+}
+
void Init() {
}
diff --git a/src/video_core/pica.h b/src/video_core/pica.h
index 38599a7a3..34b02b2f8 100644
--- a/src/video_core/pica.h
+++ b/src/video_core/pica.h
@@ -7,7 +7,6 @@
#include <array>
#include <cmath>
#include <cstddef>
-#include <map>
#include <string>
#include "common/assert.h"
@@ -908,69 +907,7 @@ struct Regs {
// Map register indices to names readable by humans
// Used for debugging purposes, so performance is not an issue here
- static std::string GetCommandName(int index) {
- std::map<u32, std::string> map;
-
- #define ADD_FIELD(name) \
- do { \
- map.insert({static_cast<u32>(PICA_REG_INDEX(name)), #name}); \
- /* TODO: change to Regs::name when VS2015 and other compilers support it */ \
- for (u32 i = PICA_REG_INDEX(name) + 1; i < PICA_REG_INDEX(name) + sizeof(Regs().name) / 4; ++i) \
- map.insert({i, #name + std::string("+") + std::to_string(i-PICA_REG_INDEX(name))}); \
- } while(false)
-
- ADD_FIELD(trigger_irq);
- ADD_FIELD(cull_mode);
- ADD_FIELD(viewport_size_x);
- ADD_FIELD(viewport_size_y);
- ADD_FIELD(viewport_depth_range);
- ADD_FIELD(viewport_depth_far_plane);
- ADD_FIELD(viewport_corner);
- ADD_FIELD(texture0_enable);
- ADD_FIELD(texture0);
- ADD_FIELD(texture0_format);
- ADD_FIELD(texture1);
- ADD_FIELD(texture1_format);
- ADD_FIELD(texture2);
- ADD_FIELD(texture2_format);
- ADD_FIELD(tev_stage0);
- ADD_FIELD(tev_stage1);
- ADD_FIELD(tev_stage2);
- ADD_FIELD(tev_stage3);
- ADD_FIELD(tev_combiner_buffer_input);
- ADD_FIELD(tev_stage4);
- ADD_FIELD(tev_stage5);
- ADD_FIELD(tev_combiner_buffer_color);
- ADD_FIELD(output_merger);
- ADD_FIELD(framebuffer);
- ADD_FIELD(vertex_attributes);
- ADD_FIELD(index_array);
- ADD_FIELD(num_vertices);
- ADD_FIELD(trigger_draw);
- ADD_FIELD(trigger_draw_indexed);
- ADD_FIELD(vs_default_attributes_setup);
- ADD_FIELD(command_buffer);
- ADD_FIELD(triangle_topology);
- ADD_FIELD(gs.bool_uniforms);
- ADD_FIELD(gs.int_uniforms);
- ADD_FIELD(gs.main_offset);
- ADD_FIELD(gs.input_register_map);
- ADD_FIELD(gs.uniform_setup);
- ADD_FIELD(gs.program);
- ADD_FIELD(gs.swizzle_patterns);
- ADD_FIELD(vs.bool_uniforms);
- ADD_FIELD(vs.int_uniforms);
- ADD_FIELD(vs.main_offset);
- ADD_FIELD(vs.input_register_map);
- ADD_FIELD(vs.uniform_setup);
- ADD_FIELD(vs.program);
- ADD_FIELD(vs.swizzle_patterns);
-
- #undef ADD_FIELD
-
- // Return empty string if no match is found
- return map[index];
- }
+ static std::string GetCommandName(int index);
static inline size_t NumIds() {
return sizeof(Regs) / sizeof(u32);
diff --git a/src/video_core/rasterizer.cpp b/src/video_core/rasterizer.cpp
index e2b90ad1c..68b7cc05d 100644
--- a/src/video_core/rasterizer.cpp
+++ b/src/video_core/rasterizer.cpp
@@ -462,7 +462,9 @@ static void ProcessTriangleInternal(const VertexShader::OutputVertex& v0,
// TODO: Apply the min and mag filters to the texture
texture_color[i] = DebugUtils::LookupTexture(texture_data, s, t, info);
+#if PICA_DUMP_TEXTURES
DebugUtils::DumpTexture(texture.config, texture_data);
+#endif
}
}
diff --git a/src/video_core/vertex_shader.cpp b/src/video_core/vertex_shader.cpp
index b77503806..960ae5779 100644
--- a/src/video_core/vertex_shader.cpp
+++ b/src/video_core/vertex_shader.cpp
@@ -2,8 +2,7 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
-#include <stack>
-
+#include <boost/container/static_vector.hpp>
#include <boost/range/algorithm.hpp>
#include <common/file_util.h>
@@ -27,7 +26,7 @@ namespace Pica {
namespace VertexShader {
struct VertexShaderState {
- const u32* program_counter;
+ u32 program_counter;
const float24* input_register_table[16];
Math::Vec4<float24> output_registers[16];
@@ -53,7 +52,7 @@ struct VertexShaderState {
};
// TODO: Is there a maximal size for this?
- std::stack<CallStackElement> call_stack;
+ boost::container::static_vector<CallStackElement, 16> call_stack;
struct {
u32 max_offset; // maximum program counter ever reached
@@ -71,15 +70,15 @@ static void ProcessShaderCode(VertexShaderState& state) {
while (true) {
if (!state.call_stack.empty()) {
- auto& top = state.call_stack.top();
- if (state.program_counter - program_code.data() == top.final_address) {
+ auto& top = state.call_stack.back();
+ if (state.program_counter == top.final_address) {
state.address_registers[2] += top.loop_increment;
if (top.repeat_counter-- == 0) {
- state.program_counter = &program_code[top.return_address];
- state.call_stack.pop();
+ state.program_counter = top.return_address;
+ state.call_stack.pop_back();
} else {
- state.program_counter = &program_code[top.loop_address];
+ state.program_counter = top.loop_address;
}
// TODO: Is "trying again" accurate to hardware?
@@ -88,17 +87,16 @@ static void ProcessShaderCode(VertexShaderState& state) {
}
bool exit_loop = false;
- const Instruction& instr = *(const Instruction*)state.program_counter;
- const SwizzlePattern& swizzle = *(SwizzlePattern*)&swizzle_data[instr.common.operand_desc_id];
+ const Instruction instr = { program_code[state.program_counter] };
+ const SwizzlePattern swizzle = { swizzle_data[instr.common.operand_desc_id] };
- static auto call = [&program_code](VertexShaderState& state, u32 offset, u32 num_instructions,
+ static auto call = [](VertexShaderState& state, u32 offset, u32 num_instructions,
u32 return_offset, u8 repeat_count, u8 loop_increment) {
- state.program_counter = &program_code[offset] - 1; // -1 to make sure when incrementing the PC we end up at the correct offset
- state.call_stack.push({ offset + num_instructions, return_offset, repeat_count, loop_increment, offset });
+ state.program_counter = offset - 1; // -1 to make sure when incrementing the PC we end up at the correct offset
+ ASSERT(state.call_stack.size() < state.call_stack.capacity());
+ state.call_stack.push_back({ offset + num_instructions, return_offset, repeat_count, loop_increment, offset });
};
- u32 binary_offset = state.program_counter - program_code.data();
-
- state.debug.max_offset = std::max<u32>(state.debug.max_offset, 1 + binary_offset);
+ state.debug.max_offset = std::max<u32>(state.debug.max_offset, 1 + state.program_counter);
auto LookupSourceRegister = [&](const SourceRegister& source_reg) -> const float24* {
switch (source_reg.GetRegisterType()) {
@@ -442,13 +440,13 @@ static void ProcessShaderCode(VertexShaderState& state) {
case OpCode::Id::JMPC:
if (evaluate_condition(state, instr.flow_control.refx, instr.flow_control.refy, instr.flow_control)) {
- state.program_counter = &program_code[instr.flow_control.dest_offset] - 1;
+ state.program_counter = instr.flow_control.dest_offset - 1;
}
break;
case OpCode::Id::JMPU:
if (uniforms.b[instr.flow_control.bool_uniform_id]) {
- state.program_counter = &program_code[instr.flow_control.dest_offset] - 1;
+ state.program_counter = instr.flow_control.dest_offset - 1;
}
break;
@@ -456,7 +454,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
call(state,
instr.flow_control.dest_offset,
instr.flow_control.num_instructions,
- binary_offset + 1, 0, 0);
+ state.program_counter + 1, 0, 0);
break;
case OpCode::Id::CALLU:
@@ -464,7 +462,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
call(state,
instr.flow_control.dest_offset,
instr.flow_control.num_instructions,
- binary_offset + 1, 0, 0);
+ state.program_counter + 1, 0, 0);
}
break;
@@ -473,7 +471,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
call(state,
instr.flow_control.dest_offset,
instr.flow_control.num_instructions,
- binary_offset + 1, 0, 0);
+ state.program_counter + 1, 0, 0);
}
break;
@@ -483,8 +481,8 @@ static void ProcessShaderCode(VertexShaderState& state) {
case OpCode::Id::IFU:
if (uniforms.b[instr.flow_control.bool_uniform_id]) {
call(state,
- binary_offset + 1,
- instr.flow_control.dest_offset - binary_offset - 1,
+ state.program_counter + 1,
+ instr.flow_control.dest_offset - state.program_counter - 1,
instr.flow_control.dest_offset + instr.flow_control.num_instructions, 0, 0);
} else {
call(state,
@@ -501,8 +499,8 @@ static void ProcessShaderCode(VertexShaderState& state) {
if (evaluate_condition(state, instr.flow_control.refx, instr.flow_control.refy, instr.flow_control)) {
call(state,
- binary_offset + 1,
- instr.flow_control.dest_offset - binary_offset - 1,
+ state.program_counter + 1,
+ instr.flow_control.dest_offset - state.program_counter - 1,
instr.flow_control.dest_offset + instr.flow_control.num_instructions, 0, 0);
} else {
call(state,
@@ -519,8 +517,8 @@ static void ProcessShaderCode(VertexShaderState& state) {
state.address_registers[2] = uniforms.i[instr.flow_control.int_uniform_id].y;
call(state,
- binary_offset + 1,
- instr.flow_control.dest_offset - binary_offset + 1,
+ state.program_counter + 1,
+ instr.flow_control.dest_offset - state.program_counter + 1,
instr.flow_control.dest_offset + 1,
uniforms.i[instr.flow_control.int_uniform_id].x,
uniforms.i[instr.flow_control.int_uniform_id].z);
@@ -551,8 +549,7 @@ OutputVertex RunShader(const InputVertex& input, int num_attributes, const Regs:
VertexShaderState state;
- const u32* main = &setup.program_code[config.main_offset];
- state.program_counter = (u32*)main;
+ state.program_counter = config.main_offset;
state.debug.max_offset = 0;
state.debug.max_opdesc_id = 0;
@@ -582,9 +579,11 @@ OutputVertex RunShader(const InputVertex& input, int num_attributes, const Regs:
state.conditional_code[1] = false;
ProcessShaderCode(state);
+#if PICA_DUMP_SHADERS
DebugUtils::DumpShader(setup.program_code.data(), state.debug.max_offset, setup.swizzle_data.data(),
state.debug.max_opdesc_id, config.main_offset,
g_state.regs.vs_output_attributes); // TODO: Don't hardcode VS here
+#endif
// Setup output data
OutputVertex ret;
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-13 05:14:22 +0100