shader_recompiler: Add support for lowering geometry passthrough

Reuses most of the existing code for generating the gl_Layer passthrough. Fixes geometry in Nier: Automata on GPUs without HW passthrough support.
This commit is contained in:
Billy Laws 2022-12-31 23:19:10 +00:00
parent 9e2997c4b6
commit 625a4af73a
2 changed files with 67 additions and 40 deletions

View File

@ -171,6 +171,64 @@ std::map<IR::Attribute, IR::Attribute> GenerateLegacyToGenericMappings(
} }
return mapping; return mapping;
} }
void EmitGeometryPassthrough(IR::IREmitter& ir, const IR::Program& program, const Shader::VaryingState &passthrough_mask, bool passthrough_position, std::optional<IR::Attribute> passthrough_layer_attr) {
for (u32 i = 0; i < program.output_vertices; i++) {
// Assign generics from input
for (u32 j = 0; j < 32; j++) {
if (!passthrough_mask.Generic(j)) {
continue;
}
const IR::Attribute attr = IR::Attribute::Generic0X + (j * 4);
ir.SetAttribute(attr + 0, ir.GetAttribute(attr + 0, ir.Imm32(i)), ir.Imm32(0));
ir.SetAttribute(attr + 1, ir.GetAttribute(attr + 1, ir.Imm32(i)), ir.Imm32(0));
ir.SetAttribute(attr + 2, ir.GetAttribute(attr + 2, ir.Imm32(i)), ir.Imm32(0));
ir.SetAttribute(attr + 3, ir.GetAttribute(attr + 3, ir.Imm32(i)), ir.Imm32(0));
}
if (passthrough_position) {
// Assign position from input
const IR::Attribute attr = IR::Attribute::PositionX;
ir.SetAttribute(attr + 0, ir.GetAttribute(attr + 0, ir.Imm32(i)), ir.Imm32(0));
ir.SetAttribute(attr + 1, ir.GetAttribute(attr + 1, ir.Imm32(i)), ir.Imm32(0));
ir.SetAttribute(attr + 2, ir.GetAttribute(attr + 2, ir.Imm32(i)), ir.Imm32(0));
ir.SetAttribute(attr + 3, ir.GetAttribute(attr + 3, ir.Imm32(i)), ir.Imm32(0));
}
if (passthrough_layer_attr) {
// Assign layer
ir.SetAttribute(IR::Attribute::Layer, ir.GetAttribute(*passthrough_layer_attr), ir.Imm32(0));
}
// Emit vertex
ir.EmitVertex(ir.Imm32(0));
}
ir.EndPrimitive(ir.Imm32(0));
}
u32 GetOutputTopologyVertices(OutputTopology output_topology) {
switch (output_topology) {
case OutputTopology::PointList:
return 1;
case OutputTopology::LineStrip:
return 2;
default:
return 3;
}
}
void LowerGeometryPassthrough(const IR::Program& program, const HostTranslateInfo& host_info) {
for (IR::Block *const block : program.blocks) {
for (IR::Inst &inst : block->Instructions()) {
if (inst.GetOpcode() == IR::Opcode::Epilogue) {
IR::IREmitter ir{*block, IR::Block::InstructionList::s_iterator_to(inst)};
EmitGeometryPassthrough(ir, program, program.info.passthrough, program.info.passthrough.AnyComponent(IR::Attribute::PositionX), {});
}
}
}
}
} // Anonymous namespace } // Anonymous namespace
IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Block>& block_pool, IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Block>& block_pool,
@ -198,6 +256,11 @@ IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Blo
for (size_t i = 0; i < program.info.passthrough.mask.size(); ++i) { for (size_t i = 0; i < program.info.passthrough.mask.size(); ++i) {
program.info.passthrough.mask[i] = ((mask[i / 32] >> (i % 32)) & 1) == 0; program.info.passthrough.mask[i] = ((mask[i / 32] >> (i % 32)) & 1) == 0;
} }
if (!host_info.support_geometry_shader_passthrough) {
program.output_vertices = GetOutputTopologyVertices(program.output_topology);
LowerGeometryPassthrough(program, host_info);
}
} }
break; break;
} }
@ -342,17 +405,8 @@ IR::Program GenerateGeometryPassthrough(ObjectPool<IR::Inst>& inst_pool,
IR::Program program; IR::Program program;
program.stage = Stage::Geometry; program.stage = Stage::Geometry;
program.output_topology = output_topology; program.output_topology = output_topology;
switch (output_topology) { program.output_vertices = GetOutputTopologyVertices(output_topology);
case OutputTopology::PointList:
program.output_vertices = 1;
break;
case OutputTopology::LineStrip:
program.output_vertices = 2;
break;
default:
program.output_vertices = 3;
break;
}
program.is_geometry_passthrough = false; program.is_geometry_passthrough = false;
program.info.loads.mask = source_program.info.stores.mask; program.info.loads.mask = source_program.info.stores.mask;
@ -366,35 +420,7 @@ IR::Program GenerateGeometryPassthrough(ObjectPool<IR::Inst>& inst_pool,
node.data.block = current_block; node.data.block = current_block;
IR::IREmitter ir{*current_block}; IR::IREmitter ir{*current_block};
for (u32 i = 0; i < program.output_vertices; i++) { EmitGeometryPassthrough(ir, program, program.info.stores, true, source_program.info.emulated_layer);
// Assign generics from input
for (u32 j = 0; j < 32; j++) {
if (!program.info.stores.Generic(j)) {
continue;
}
const IR::Attribute attr = IR::Attribute::Generic0X + (j * 4);
ir.SetAttribute(attr + 0, ir.GetAttribute(attr + 0, ir.Imm32(i)), ir.Imm32(0));
ir.SetAttribute(attr + 1, ir.GetAttribute(attr + 1, ir.Imm32(i)), ir.Imm32(0));
ir.SetAttribute(attr + 2, ir.GetAttribute(attr + 2, ir.Imm32(i)), ir.Imm32(0));
ir.SetAttribute(attr + 3, ir.GetAttribute(attr + 3, ir.Imm32(i)), ir.Imm32(0));
}
// Assign position from input
const IR::Attribute attr = IR::Attribute::PositionX;
ir.SetAttribute(attr + 0, ir.GetAttribute(attr + 0, ir.Imm32(i)), ir.Imm32(0));
ir.SetAttribute(attr + 1, ir.GetAttribute(attr + 1, ir.Imm32(i)), ir.Imm32(0));
ir.SetAttribute(attr + 2, ir.GetAttribute(attr + 2, ir.Imm32(i)), ir.Imm32(0));
ir.SetAttribute(attr + 3, ir.GetAttribute(attr + 3, ir.Imm32(i)), ir.Imm32(0));
// Assign layer
ir.SetAttribute(IR::Attribute::Layer, ir.GetAttribute(source_program.info.emulated_layer),
ir.Imm32(0));
// Emit vertex
ir.EmitVertex(ir.Imm32(0));
}
ir.EndPrimitive(ir.Imm32(0));
IR::Block* return_block{block_pool.Create(inst_pool)}; IR::Block* return_block{block_pool.Create(inst_pool)};
IR::IREmitter{*return_block}.Epilogue(); IR::IREmitter{*return_block}.Epilogue();

View File

@ -16,6 +16,7 @@ struct HostTranslateInfo {
bool support_snorm_render_buffer{}; ///< True when the device supports SNORM render buffers bool support_snorm_render_buffer{}; ///< True when the device supports SNORM render buffers
bool support_viewport_index_layer{}; ///< True when the device supports gl_Layer in VS bool support_viewport_index_layer{}; ///< True when the device supports gl_Layer in VS
u32 min_ssbo_alignment{}; ///< Minimum alignment supported by the device for SSBOs u32 min_ssbo_alignment{}; ///< Minimum alignment supported by the device for SSBOs
bool support_geometry_shader_passthrough{}; ///< True when the device supports geometry passthrough shaders
}; };
} // namespace Shader } // namespace Shader