forked from ShuriZma/suyu
Shader: Remove OutputRegisters struct
This commit is contained in:
parent
9ea5eacf91
commit
6fa3687afc
|
@ -152,8 +152,8 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) {
|
||||||
Shader::UnitState shader_unit;
|
Shader::UnitState shader_unit;
|
||||||
shader_unit.LoadInputVertex(immediate_input, regs.vs.num_input_attributes + 1);
|
shader_unit.LoadInputVertex(immediate_input, regs.vs.num_input_attributes + 1);
|
||||||
shader_engine->Run(shader_unit, regs.vs.main_offset);
|
shader_engine->Run(shader_unit, regs.vs.main_offset);
|
||||||
Shader::OutputVertex output_vertex =
|
auto output_vertex = Shader::OutputVertex::FromRegisters(
|
||||||
shader_unit.output_registers.ToVertex(regs.vs);
|
shader_unit.registers.output, regs, regs.vs.output_mask);
|
||||||
|
|
||||||
// Send to renderer
|
// Send to renderer
|
||||||
using Pica::Shader::OutputVertex;
|
using Pica::Shader::OutputVertex;
|
||||||
|
@ -291,7 +291,8 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) {
|
||||||
shader_engine->Run(shader_unit, regs.vs.main_offset);
|
shader_engine->Run(shader_unit, regs.vs.main_offset);
|
||||||
|
|
||||||
// Retrieve vertex from register data
|
// Retrieve vertex from register data
|
||||||
output_vertex = shader_unit.output_registers.ToVertex(regs.vs);
|
output_vertex = Shader::OutputVertex::FromRegisters(shader_unit.registers.output,
|
||||||
|
regs, regs.vs.output_mask);
|
||||||
|
|
||||||
if (is_indexed) {
|
if (is_indexed) {
|
||||||
vertex_cache[vertex_cache_pos] = output_vertex;
|
vertex_cache[vertex_cache_pos] = output_vertex;
|
||||||
|
|
|
@ -19,7 +19,8 @@ namespace Pica {
|
||||||
|
|
||||||
namespace Shader {
|
namespace Shader {
|
||||||
|
|
||||||
OutputVertex OutputRegisters::ToVertex(const Regs::ShaderConfig& config) const {
|
OutputVertex OutputVertex::FromRegisters(Math::Vec4<float24> output_regs[16], const Regs& regs,
|
||||||
|
u32 output_mask) {
|
||||||
// Setup output data
|
// Setup output data
|
||||||
OutputVertex ret;
|
OutputVertex ret;
|
||||||
// TODO(neobrain): Under some circumstances, up to 16 attributes may be output. We need to
|
// TODO(neobrain): Under some circumstances, up to 16 attributes may be output. We need to
|
||||||
|
@ -27,13 +28,13 @@ OutputVertex OutputRegisters::ToVertex(const Regs::ShaderConfig& config) const {
|
||||||
unsigned index = 0;
|
unsigned index = 0;
|
||||||
for (unsigned i = 0; i < 7; ++i) {
|
for (unsigned i = 0; i < 7; ++i) {
|
||||||
|
|
||||||
if (index >= g_state.regs.vs_output_total)
|
if (index >= regs.vs_output_total)
|
||||||
break;
|
break;
|
||||||
|
|
||||||
if ((config.output_mask & (1 << i)) == 0)
|
if ((output_mask & (1 << i)) == 0)
|
||||||
continue;
|
continue;
|
||||||
|
|
||||||
const auto& output_register_map = g_state.regs.vs_output_attributes[index];
|
const auto& output_register_map = regs.vs_output_attributes[index];
|
||||||
|
|
||||||
u32 semantics[4] = {output_register_map.map_x, output_register_map.map_y,
|
u32 semantics[4] = {output_register_map.map_x, output_register_map.map_y,
|
||||||
output_register_map.map_z, output_register_map.map_w};
|
output_register_map.map_z, output_register_map.map_w};
|
||||||
|
@ -41,7 +42,7 @@ OutputVertex OutputRegisters::ToVertex(const Regs::ShaderConfig& config) const {
|
||||||
for (unsigned comp = 0; comp < 4; ++comp) {
|
for (unsigned comp = 0; comp < 4; ++comp) {
|
||||||
float24* out = ((float24*)&ret) + semantics[comp];
|
float24* out = ((float24*)&ret) + semantics[comp];
|
||||||
if (semantics[comp] != Regs::VSOutputAttributes::INVALID) {
|
if (semantics[comp] != Regs::VSOutputAttributes::INVALID) {
|
||||||
*out = value[i][comp];
|
*out = output_regs[i][comp];
|
||||||
} else {
|
} else {
|
||||||
// Zero output so that attributes which aren't output won't have denormals in them,
|
// Zero output so that attributes which aren't output won't have denormals in them,
|
||||||
// which would slow us down later.
|
// which would slow us down later.
|
||||||
|
|
|
@ -73,19 +73,13 @@ struct OutputVertex {
|
||||||
ret.Lerp(factor, v1);
|
ret.Lerp(factor, v1);
|
||||||
return ret;
|
return ret;
|
||||||
}
|
}
|
||||||
|
|
||||||
|
static OutputVertex FromRegisters(Math::Vec4<float24> output_regs[16], const Regs& regs,
|
||||||
|
u32 output_mask);
|
||||||
};
|
};
|
||||||
static_assert(std::is_pod<OutputVertex>::value, "Structure is not POD");
|
static_assert(std::is_pod<OutputVertex>::value, "Structure is not POD");
|
||||||
static_assert(sizeof(OutputVertex) == 32 * sizeof(float), "OutputVertex has invalid size");
|
static_assert(sizeof(OutputVertex) == 32 * sizeof(float), "OutputVertex has invalid size");
|
||||||
|
|
||||||
struct OutputRegisters {
|
|
||||||
OutputRegisters() = default;
|
|
||||||
|
|
||||||
alignas(16) Math::Vec4<float24> value[16];
|
|
||||||
|
|
||||||
OutputVertex ToVertex(const Regs::ShaderConfig& config) const;
|
|
||||||
};
|
|
||||||
static_assert(std::is_pod<OutputRegisters>::value, "Structure is not POD");
|
|
||||||
|
|
||||||
/**
|
/**
|
||||||
* This structure contains the state information that needs to be unique for a shader unit. The 3DS
|
* This structure contains the state information that needs to be unique for a shader unit. The 3DS
|
||||||
* has four shader units that process shaders in parallel. At the present, Citra only implements a
|
* has four shader units that process shaders in parallel. At the present, Citra only implements a
|
||||||
|
@ -98,11 +92,10 @@ struct UnitState {
|
||||||
// required to be 16-byte aligned.
|
// required to be 16-byte aligned.
|
||||||
alignas(16) Math::Vec4<float24> input[16];
|
alignas(16) Math::Vec4<float24> input[16];
|
||||||
alignas(16) Math::Vec4<float24> temporary[16];
|
alignas(16) Math::Vec4<float24> temporary[16];
|
||||||
|
alignas(16) Math::Vec4<float24> output[16];
|
||||||
} registers;
|
} registers;
|
||||||
static_assert(std::is_pod<Registers>::value, "Structure is not POD");
|
static_assert(std::is_pod<Registers>::value, "Structure is not POD");
|
||||||
|
|
||||||
OutputRegisters output_registers;
|
|
||||||
|
|
||||||
bool conditional_code[2];
|
bool conditional_code[2];
|
||||||
|
|
||||||
// Two Address registers and one loop counter
|
// Two Address registers and one loop counter
|
||||||
|
@ -128,7 +121,7 @@ struct UnitState {
|
||||||
static size_t OutputOffset(const DestRegister& reg) {
|
static size_t OutputOffset(const DestRegister& reg) {
|
||||||
switch (reg.GetRegisterType()) {
|
switch (reg.GetRegisterType()) {
|
||||||
case RegisterType::Output:
|
case RegisterType::Output:
|
||||||
return offsetof(UnitState, output_registers.value) +
|
return offsetof(UnitState, registers.output) +
|
||||||
reg.GetIndex() * sizeof(Math::Vec4<float24>);
|
reg.GetIndex() * sizeof(Math::Vec4<float24>);
|
||||||
|
|
||||||
case RegisterType::Temporary:
|
case RegisterType::Temporary:
|
||||||
|
|
|
@ -175,7 +175,7 @@ static void RunInterpreter(const ShaderSetup& setup, UnitState& state, DebugData
|
||||||
|
|
||||||
float24* dest =
|
float24* dest =
|
||||||
(instr.common.dest.Value() < 0x10)
|
(instr.common.dest.Value() < 0x10)
|
||||||
? &state.output_registers.value[instr.common.dest.Value().GetIndex()][0]
|
? &state.registers.output[instr.common.dest.Value().GetIndex()][0]
|
||||||
: (instr.common.dest.Value() < 0x20)
|
: (instr.common.dest.Value() < 0x20)
|
||||||
? &state.registers.temporary[instr.common.dest.Value().GetIndex()][0]
|
? &state.registers.temporary[instr.common.dest.Value().GetIndex()][0]
|
||||||
: dummy_vec4_float24;
|
: dummy_vec4_float24;
|
||||||
|
@ -518,7 +518,7 @@ static void RunInterpreter(const ShaderSetup& setup, UnitState& state, DebugData
|
||||||
|
|
||||||
float24* dest =
|
float24* dest =
|
||||||
(instr.mad.dest.Value() < 0x10)
|
(instr.mad.dest.Value() < 0x10)
|
||||||
? &state.output_registers.value[instr.mad.dest.Value().GetIndex()][0]
|
? &state.registers.output[instr.mad.dest.Value().GetIndex()][0]
|
||||||
: (instr.mad.dest.Value() < 0x20)
|
: (instr.mad.dest.Value() < 0x20)
|
||||||
? &state.registers.temporary[instr.mad.dest.Value().GetIndex()][0]
|
? &state.registers.temporary[instr.mad.dest.Value().GetIndex()][0]
|
||||||
: dummy_vec4_float24;
|
: dummy_vec4_float24;
|
||||||
|
|
Loading…
Reference in New Issue