From 24fe223692af988770bcc336d5b1138a2ba91399 Mon Sep 17 00:00:00 2001 From: lizzie Date: Wed, 25 Mar 2026 10:48:53 +0100 Subject: [PATCH] [dynarmic] Remove last FPT LUT table, removing around 30kb worth of unused functions (#3718) Lets do the quick math There was 1 LUT for every fsize() instancing Now... the number of functions on each lut was (fsize + 1), multiplied by 5 (number of rounding modes) 8 = 9 * 5 = 45 16 = 17 * 5 = 85 32 = 33 * 5 = 165 64 = 65 * 5 = 325 this is just pure insanity - look at what fucking nm reported: ``` 0000000003dc39b8 0000000000000008 V guard variable for void Dynarmic::Backend::X64::EmitFPVectorToFixed<16ul, false>(Dynarmic::Backend::X64::BlockOfCode&, Dynarmic::Backend::X64::EmitContext&, Dynarmic::IR::Inst*)::lut 0000000003dc3a18 0000000000000008 V guard variable for void Dynarmic::Backend::X64::EmitFPVectorToFixed<16ul, true>(Dynarmic::Backend::X64::BlockOfCode&, Dynarmic::Backend::X64::EmitContext&, Dynarmic::IR::Inst*)::lut 0000000003dc39d8 0000000000000008 V guard variable for void Dynarmic::Backend::X64::EmitFPVectorToFixed<32ul, false>(Dynarmic::Backend::X64::BlockOfCode&, Dynarmic::Backend::X64::EmitContext&, Dynarmic::IR::Inst*)::lut 0000000003dc3a38 0000000000000008 V guard variable for void Dynarmic::Backend::X64::EmitFPVectorToFixed<32ul, true>(Dynarmic::Backend::X64::BlockOfCode&, Dynarmic::Backend::X64::EmitContext&, Dynarmic::IR::Inst*)::lut 0000000003dc39f8 0000000000000008 V guard variable for void Dynarmic::Backend::X64::EmitFPVectorToFixed<64ul, false>(Dynarmic::Backend::X64::BlockOfCode&, Dynarmic::Backend::X64::EmitContext&, Dynarmic::IR::Inst*)::lut 0000000003dc3a58 0000000000000008 V guard variable for void Dynarmic::Backend::X64::EmitFPVectorToFixed<64ul, true>(Dynarmic::Backend::X64::BlockOfCode&, Dynarmic::Backend::X64::EmitContext&, Dynarmic::IR::Inst*)::lut ``` "ah its not bad" - OH MATE ITS JUST THE GUARD VARIABLES - i attached a file with just the functions generated for each case... now with this PR only 6 * 6 functions are made (still not ideal, but way better), 36 is way better than 1156 FUCKING FUNCTIONS Signed-off-by: lizzie Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/3718 Reviewed-by: DraVee Reviewed-by: Maufeat Co-authored-by: lizzie Co-committed-by: lizzie --- src/dynarmic/src/dynarmic/CMakeLists.txt | 1 - .../emit_arm64_vector_floating_point.cpp | 3 +- .../backend/x64/emit_x64_floating_point.cpp | 1 - .../x64/emit_x64_vector_floating_point.cpp | 57 ++++++++++++------- .../src/dynarmic/common/lut_from_list.h | 55 ------------------ 5 files changed, 36 insertions(+), 81 deletions(-) delete mode 100644 src/dynarmic/src/dynarmic/common/lut_from_list.h diff --git a/src/dynarmic/src/dynarmic/CMakeLists.txt b/src/dynarmic/src/dynarmic/CMakeLists.txt index 3d2ea4d42e..50bf5a4604 100644 --- a/src/dynarmic/src/dynarmic/CMakeLists.txt +++ b/src/dynarmic/src/dynarmic/CMakeLists.txt @@ -53,7 +53,6 @@ add_library(dynarmic STATIC common/fp/util.h common/llvm_disassemble.cpp common/llvm_disassemble.h - common/lut_from_list.h common/math_util.cpp common/math_util.h common/safe_ops.h diff --git a/src/dynarmic/src/dynarmic/backend/arm64/emit_arm64_vector_floating_point.cpp b/src/dynarmic/src/dynarmic/backend/arm64/emit_arm64_vector_floating_point.cpp index 4d11c62abd..431d51c081 100644 --- a/src/dynarmic/src/dynarmic/backend/arm64/emit_arm64_vector_floating_point.cpp +++ b/src/dynarmic/src/dynarmic/backend/arm64/emit_arm64_vector_floating_point.cpp @@ -1,4 +1,4 @@ -// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project +// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project // SPDX-License-Identifier: GPL-3.0-or-later /* This file is part of the dynarmic project. @@ -31,7 +31,6 @@ #include "dynarmic/common/fp/info.h" #include "dynarmic/common/fp/op.h" #include "dynarmic/common/fp/rounding_mode.h" -#include "dynarmic/common/lut_from_list.h" #include "dynarmic/ir/basic_block.h" #include "dynarmic/ir/microinstruction.h" #include "dynarmic/ir/opcodes.h" diff --git a/src/dynarmic/src/dynarmic/backend/x64/emit_x64_floating_point.cpp b/src/dynarmic/src/dynarmic/backend/x64/emit_x64_floating_point.cpp index 76c103ec6f..abe04b53ff 100644 --- a/src/dynarmic/src/dynarmic/backend/x64/emit_x64_floating_point.cpp +++ b/src/dynarmic/src/dynarmic/backend/x64/emit_x64_floating_point.cpp @@ -31,7 +31,6 @@ #include "dynarmic/common/fp/info.h" #include "dynarmic/common/fp/op.h" #include "dynarmic/common/fp/rounding_mode.h" -#include "dynarmic/common/lut_from_list.h" #include "dynarmic/interface/optimization_flags.h" #include "dynarmic/ir/basic_block.h" #include "dynarmic/ir/microinstruction.h" diff --git a/src/dynarmic/src/dynarmic/backend/x64/emit_x64_vector_floating_point.cpp b/src/dynarmic/src/dynarmic/backend/x64/emit_x64_vector_floating_point.cpp index 2247b18fcd..ee9ec39f46 100644 --- a/src/dynarmic/src/dynarmic/backend/x64/emit_x64_vector_floating_point.cpp +++ b/src/dynarmic/src/dynarmic/backend/x64/emit_x64_vector_floating_point.cpp @@ -31,7 +31,6 @@ #include "dynarmic/common/fp/info.h" #include "dynarmic/common/fp/op.h" #include "dynarmic/common/fp/util.h" -#include "dynarmic/common/lut_from_list.h" #include "dynarmic/interface/optimization_flags.h" #include "dynarmic/ir/basic_block.h" #include "dynarmic/ir/microinstruction.h" @@ -2127,28 +2126,42 @@ void EmitFPVectorToFixed(BlockOfCode& code, EmitContext& ctx, IR::Inst* inst) { } } - using fbits_list = mp::lift_sequence>; - using rounding_list = mp::list< - mp::lift_value, - mp::lift_value, - mp::lift_value, - mp::lift_value, - mp::lift_value>; - - static const auto lut = Common::GenerateLookupTableFromList([](I) { - using FPT = mcl::unsigned_integer_of_size; // WORKAROUND: For issue 678 on MSVC - return std::pair{ - mp::lower_to_tuple_v, - Common::FptrCast([](VectorArray& output, const VectorArray& input, FP::FPCR fpcr, FP::FPSR& fpsr) { - constexpr size_t fbits = mp::get<0, I>::value; - constexpr FP::RoundingMode rounding_mode = mp::get<1, I>::value; + using FPT = mcl::unsigned_integer_of_size; // WORKAROUND: For issue 678 on MSVC + auto const func = [rounding]() -> void(*)(VectorArray& output, const VectorArray& input, FP::FPCR fpcr, FP::FPSR& fpsr) { + switch (rounding) { + case FP::RoundingMode::ToNearest_TieEven: + return [](VectorArray& output, const VectorArray& input, FP::FPCR fpcr, FP::FPSR& fpsr) { for (size_t i = 0; i < output.size(); ++i) - output[i] = FPT(FP::FPToFixed(fsize, input[i], fbits, unsigned_, fpcr, rounding_mode, fpsr)); - }) - }; - }, mp::cartesian_product{}); - - EmitTwoOpFallback<3>(code, ctx, inst, lut.at(std::make_tuple(fbits, rounding))); + output[i] = FPT(FP::FPToFixed(fsize, input[i], fsize, unsigned_, fpcr, FP::RoundingMode::ToNearest_TieEven, fpsr)); + }; + case FP::RoundingMode::TowardsPlusInfinity: + return [](VectorArray& output, const VectorArray& input, FP::FPCR fpcr, FP::FPSR& fpsr) { + for (size_t i = 0; i < output.size(); ++i) + output[i] = FPT(FP::FPToFixed(fsize, input[i], fsize, unsigned_, fpcr, FP::RoundingMode::TowardsPlusInfinity, fpsr)); + }; + case FP::RoundingMode::TowardsMinusInfinity: + return [](VectorArray& output, const VectorArray& input, FP::FPCR fpcr, FP::FPSR& fpsr) { + for (size_t i = 0; i < output.size(); ++i) + output[i] = FPT(FP::FPToFixed(fsize, input[i], fsize, unsigned_, fpcr, FP::RoundingMode::TowardsMinusInfinity, fpsr)); + }; + case FP::RoundingMode::TowardsZero: + return [](VectorArray& output, const VectorArray& input, FP::FPCR fpcr, FP::FPSR& fpsr) { + for (size_t i = 0; i < output.size(); ++i) + output[i] = FPT(FP::FPToFixed(fsize, input[i], fsize, unsigned_, fpcr, FP::RoundingMode::TowardsZero, fpsr)); + }; + case FP::RoundingMode::ToNearest_TieAwayFromZero: + return [](VectorArray& output, const VectorArray& input, FP::FPCR fpcr, FP::FPSR& fpsr) { + for (size_t i = 0; i < output.size(); ++i) + output[i] = FPT(FP::FPToFixed(fsize, input[i], fsize, unsigned_, fpcr, FP::RoundingMode::ToNearest_TieAwayFromZero, fpsr)); + }; + case FP::RoundingMode::ToOdd: + return [](VectorArray& output, const VectorArray& input, FP::FPCR fpcr, FP::FPSR& fpsr) { + for (size_t i = 0; i < output.size(); ++i) + output[i] = FPT(FP::FPToFixed(fsize, input[i], fsize, unsigned_, fpcr, FP::RoundingMode::ToOdd, fpsr)); + }; + } + }(); + EmitTwoOpFallback<3>(code, ctx, inst, func); } void EmitX64::EmitFPVectorToSignedFixed16(EmitContext& ctx, IR::Inst* inst) { diff --git a/src/dynarmic/src/dynarmic/common/lut_from_list.h b/src/dynarmic/src/dynarmic/common/lut_from_list.h deleted file mode 100644 index 633b62aeda..0000000000 --- a/src/dynarmic/src/dynarmic/common/lut_from_list.h +++ /dev/null @@ -1,55 +0,0 @@ -// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project -// SPDX-License-Identifier: GPL-3.0-or-later - -/* This file is part of the dynarmic project. - * Copyright (c) 2018 MerryMage - * SPDX-License-Identifier: 0BSD - */ - -#pragma once - -#include -#include -#include - -#include -#include -#include - -#ifdef _MSC_VER -# include -#endif - -namespace Dynarmic::Common { - -// prevents this function from printing 56,000 character warning messages -#ifdef __GNUC__ -#pragma GCC diagnostic push -#pragma GCC diagnostic ignored "-Wno-stack-usage" -#endif -#ifdef __clang__ -#pragma clang diagnostic push -#pragma clang diagnostic ignored "-Wno-stack-usage" -#endif - -template -inline auto GenerateLookupTableFromList(Function f, mcl::mp::list) { -#ifdef _MSC_VER - using PairT = std::invoke_result_t>>; -#else - using PairT = std::common_type_t...>; -#endif - using MapT = mcl::mp::apply; - static_assert(mcl::is_instance_of_template_v); - const std::initializer_list pair_array{f(Values{})...}; - return MapT(pair_array.begin(), pair_array.end()); -} - -#ifdef __GNUC__ -#pragma GCC diagnostic pop -#endif -#ifdef __clang__ -#pragma clang diagnostic pop -#endif - -} // namespace Dynarmic::Common