From 4006929c986a2e0e52429fe21201a7ad5ca3fea9 Mon Sep 17 00:00:00 2001 From: ReinUsesLisp Date: Wed, 3 Mar 2021 03:07:19 -0300 Subject: shader: Implement HADD2 --- .../translate/impl/half_floating_point_add.cpp | 184 +++++++++++++++++++++ .../maxwell/translate/impl/load_store_memory.cpp | 2 +- .../maxwell/translate/impl/not_implemented.cpp | 16 -- 3 files changed, 185 insertions(+), 17 deletions(-) create mode 100644 src/shader_recompiler/frontend/maxwell/translate/impl/half_floating_point_add.cpp (limited to 'src/shader_recompiler/frontend/maxwell/translate') diff --git a/src/shader_recompiler/frontend/maxwell/translate/impl/half_floating_point_add.cpp b/src/shader_recompiler/frontend/maxwell/translate/impl/half_floating_point_add.cpp new file mode 100644 index 000000000..6965adfb3 --- /dev/null +++ b/src/shader_recompiler/frontend/maxwell/translate/impl/half_floating_point_add.cpp @@ -0,0 +1,184 @@ +// Copyright 2021 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include "common/common_types.h" +#include "shader_recompiler/exception.h" +#include "shader_recompiler/frontend/maxwell/translate/impl/common_encoding.h" +#include "shader_recompiler/frontend/maxwell/translate/impl/impl.h" + +namespace Shader::Maxwell { +namespace { +enum class Merge : u64 { + H1_H0, + F32, + MRG_H0, + MRG_H1, +}; + +enum class Swizzle : u64 { + H1_H0, + F32, + H0_H0, + H1_H1, +}; + +std::pair Extract(IR::IREmitter& ir, IR::U32 value, Swizzle swizzle) { + switch (swizzle) { + case Swizzle::H1_H0: { + const IR::Value vector{ir.UnpackFloat2x16(value)}; + return {IR::F16{ir.CompositeExtract(vector, 0)}, IR::F16{ir.CompositeExtract(vector, 1)}}; + } + case Swizzle::H0_H0: { + const IR::F16 scalar{ir.CompositeExtract(ir.UnpackFloat2x16(value), 0)}; + return {scalar, scalar}; + } + case Swizzle::H1_H1: { + const IR::F16 scalar{ir.CompositeExtract(ir.UnpackFloat2x16(value), 1)}; + return {scalar, scalar}; + } + case Swizzle::F32: { + const IR::F32 scalar{ir.BitCast(value)}; + return {scalar, scalar}; + } + } + throw InvalidArgument("Invalid swizzle {}", swizzle); +} + +IR::U32 MergeResult(IR::IREmitter& ir, IR::Reg dest, const IR::F16& lhs, const IR::F16& rhs, + Merge merge) { + switch (merge) { + case Merge::H1_H0: + return ir.PackFloat2x16(ir.CompositeConstruct(lhs, rhs)); + case Merge::F32: + return ir.BitCast(ir.FPConvert(32, lhs)); + case Merge::MRG_H0: + case Merge::MRG_H1: { + const IR::Value vector{ir.UnpackFloat2x16(ir.GetReg(dest))}; + const bool h0{merge == Merge::MRG_H0}; + const IR::F16& insert{h0 ? lhs : rhs}; + return ir.PackFloat2x16(ir.CompositeInsert(vector, insert, h0 ? 0 : 1)); + } + } + throw InvalidArgument("Invalid merge {}", merge); +} + +void HADD2(TranslatorVisitor& v, u64 insn, Merge merge, bool ftz, bool sat, bool abs_a, bool neg_a, + Swizzle swizzle_a, bool abs_b, bool neg_b, Swizzle swizzle_b, const IR::U32& src_b) { + union { + u64 raw; + BitField<0, 8, IR::Reg> dest_reg; + BitField<8, 8, IR::Reg> src_a; + } const hadd2{insn}; + + auto [lhs_a, rhs_a]{Extract(v.ir, v.X(hadd2.src_a), swizzle_a)}; + auto [lhs_b, rhs_b]{Extract(v.ir, src_b, swizzle_b)}; + const bool promotion{lhs_a.Type() != lhs_b.Type()}; + if (promotion) { + if (lhs_a.Type() == IR::Type::F16) { + lhs_a = v.ir.FPConvert(32, lhs_a); + rhs_a = v.ir.FPConvert(32, rhs_a); + } + if (lhs_b.Type() == IR::Type::F16) { + lhs_b = v.ir.FPConvert(32, lhs_b); + rhs_b = v.ir.FPConvert(32, rhs_b); + } + } + lhs_a = v.ir.FPAbsNeg(lhs_a, abs_a, neg_a); + rhs_a = v.ir.FPAbsNeg(rhs_a, abs_a, neg_a); + + lhs_b = v.ir.FPAbsNeg(lhs_b, abs_b, neg_b); + rhs_b = v.ir.FPAbsNeg(rhs_b, abs_b, neg_b); + + const IR::FpControl fp_control{ + .no_contraction{true}, + .rounding{IR::FpRounding::DontCare}, + .fmz_mode{ftz ? IR::FmzMode::FTZ : IR::FmzMode::None}, + }; + IR::F16F32F64 lhs{v.ir.FPAdd(lhs_a, lhs_b, fp_control)}; + IR::F16F32F64 rhs{v.ir.FPAdd(rhs_a, rhs_b, fp_control)}; + if (sat) { + lhs = v.ir.FPSaturate(lhs); + rhs = v.ir.FPSaturate(rhs); + } + if (promotion) { + lhs = v.ir.FPConvert(16, lhs); + rhs = v.ir.FPConvert(16, rhs); + } + v.X(hadd2.dest_reg, MergeResult(v.ir, hadd2.dest_reg, lhs, rhs, merge)); +} +} // Anonymous namespace + +void TranslatorVisitor::HADD2_reg(u64 insn) { + union { + u64 raw; + BitField<49, 2, Merge> merge; + BitField<39, 1, u64> ftz; + BitField<32, 1, u64> sat; + BitField<43, 1, u64> neg_a; + BitField<44, 1, u64> abs_a; + BitField<47, 2, Swizzle> swizzle_a; + BitField<31, 1, u64> neg_b; + BitField<30, 1, u64> abs_b; + BitField<28, 2, Swizzle> swizzle_b; + } const hadd2{insn}; + + HADD2(*this, insn, hadd2.merge, hadd2.ftz != 0, hadd2.sat != 0, hadd2.abs_a != 0, + hadd2.neg_a != 0, hadd2.swizzle_a, hadd2.abs_b != 0, hadd2.neg_b != 0, hadd2.swizzle_b, + GetReg20(insn)); +} + +void TranslatorVisitor::HADD2_cbuf(u64 insn) { + union { + u64 raw; + BitField<49, 2, Merge> merge; + BitField<39, 1, u64> ftz; + BitField<52, 1, u64> sat; + BitField<43, 1, u64> neg_a; + BitField<44, 1, u64> abs_a; + BitField<47, 2, Swizzle> swizzle_a; + BitField<56, 1, u64> neg_b; + BitField<54, 1, u64> abs_b; + } const hadd2{insn}; + + HADD2(*this, insn, hadd2.merge, hadd2.ftz != 0, hadd2.sat != 0, hadd2.abs_a != 0, + hadd2.neg_a != 0, hadd2.swizzle_a, hadd2.abs_b != 0, hadd2.neg_b != 0, Swizzle::F32, + GetCbuf(insn)); +} + +void TranslatorVisitor::HADD2_imm(u64 insn) { + union { + u64 raw; + BitField<49, 2, Merge> merge; + BitField<39, 1, u64> ftz; + BitField<52, 1, u64> sat; + BitField<43, 1, u64> neg_a; + BitField<44, 1, u64> abs_a; + BitField<47, 2, Swizzle> swizzle_a; + BitField<56, 1, u64> neg_high; + BitField<30, 9, u64> high; + BitField<29, 1, u64> neg_low; + BitField<20, 9, u64> low; + } const hadd2{insn}; + + const u32 imm{static_cast(hadd2.low << 6) | ((hadd2.neg_low != 0 ? 1 : 0) << 15) | + static_cast(hadd2.high << 22) | ((hadd2.neg_high != 0 ? 1 : 0) << 31)}; + HADD2(*this, insn, hadd2.merge, hadd2.ftz != 0, hadd2.sat != 0, hadd2.abs_a != 0, + hadd2.neg_a != 0, hadd2.swizzle_a, false, false, Swizzle::H1_H0, ir.Imm32(imm)); +} + +void TranslatorVisitor::HADD2_32I(u64 insn) { + union { + u64 raw; + BitField<55, 1, u64> ftz; + BitField<52, 1, u64> sat; + BitField<56, 1, u64> neg_a; + BitField<53, 2, Swizzle> swizzle_a; + BitField<20, 32, u64> imm32; + } const hadd2{insn}; + + const u32 imm{static_cast(hadd2.imm32)}; + HADD2(*this, insn, Merge::H1_H0, hadd2.ftz != 0, hadd2.sat != 0, false, hadd2.neg_a != 0, + hadd2.swizzle_a, false, false, Swizzle::H1_H0, ir.Imm32(imm)); +} +} // namespace Shader::Maxwell diff --git a/src/shader_recompiler/frontend/maxwell/translate/impl/load_store_memory.cpp b/src/shader_recompiler/frontend/maxwell/translate/impl/load_store_memory.cpp index 727524284..748b856c9 100644 --- a/src/shader_recompiler/frontend/maxwell/translate/impl/load_store_memory.cpp +++ b/src/shader_recompiler/frontend/maxwell/translate/impl/load_store_memory.cpp @@ -59,7 +59,7 @@ IR::U64 Address(TranslatorVisitor& v, u64 insn) { const IR::U64 address{[&]() -> IR::U64 { if (mem.e == 0) { // LDG/STG without .E uses a 32-bit pointer, zero-extend it - return v.ir.ConvertU(64, v.X(mem.addr_reg)); + return v.ir.UConvert(64, v.X(mem.addr_reg)); } if (!IR::IsAligned(mem.addr_reg, 2)) { throw NotImplementedException("Unaligned address register"); diff --git a/src/shader_recompiler/frontend/maxwell/translate/impl/not_implemented.cpp b/src/shader_recompiler/frontend/maxwell/translate/impl/not_implemented.cpp index a0535f1c2..c24f29ff7 100644 --- a/src/shader_recompiler/frontend/maxwell/translate/impl/not_implemented.cpp +++ b/src/shader_recompiler/frontend/maxwell/translate/impl/not_implemented.cpp @@ -265,22 +265,6 @@ void TranslatorVisitor::GETLMEMBASE(u64) { ThrowNotImplemented(Opcode::GETLMEMBASE); } -void TranslatorVisitor::HADD2_reg(u64) { - ThrowNotImplemented(Opcode::HADD2_reg); -} - -void TranslatorVisitor::HADD2_cbuf(u64) { - ThrowNotImplemented(Opcode::HADD2_cbuf); -} - -void TranslatorVisitor::HADD2_imm(u64) { - ThrowNotImplemented(Opcode::HADD2_imm); -} - -void TranslatorVisitor::HADD2_32I(u64) { - ThrowNotImplemented(Opcode::HADD2_32I); -} - void TranslatorVisitor::HFMA2_reg(u64) { ThrowNotImplemented(Opcode::HFMA2_reg); } -- cgit v1.2.3