Config.h

/*
  ==============================================================================
    DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 
    Copyright 2022- by sonible GmbH.
 
    This file is part of VCTR - Versatile Container Templates Reconceptualized.
 
    VCTR is free software: you can redistribute it and/or modify
    it under the terms of the GNU Lesser General Public License version 3
    only, as published by the Free Software Foundation.
 
    VCTR is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU Lesser General Public License version 3 for more details.
 
    You should have received a copy of the GNU Lesser General Public License
    version 3 along with VCTR.  If not, see <https://www.gnu.org/licenses/>.
  ==============================================================================
*/
 
//==============================================================================
// Platform defines
//==============================================================================
#if defined(_WIN32) || defined(_WIN64)
#define VCTR_WINDOWS 1
#elif defined(LINUX) || defined(__linux__)
#define VCTR_LINUX 1
#elif defined(__APPLE_CPP__) || defined(__APPLE_CC__)
#define VCTR_APPLE 1
#elif defined(__wasm__)
#define VCTR_WASM 1
#endif
 
#ifndef VCTR_WINDOWS
#define VCTR_WINDOWS 0
#endif
 
#ifndef VCTR_LINUX
#define VCTR_LINUX 0
#endif
 
#ifndef VCTR_APPLE
#define VCTR_APPLE 0
#endif
 
#ifndef VCTR_WASM
#define VCTR_WASM 0
#endif
 
//==============================================================================
// CPU architecture defines
//==============================================================================
#if defined(__aarch64__) || defined(_M_ARM64)
#define VCTR_ARM 1
#define VCTR_X64 0
#define VCTR_MAX_SIMD_REGISTER_SIZE 16
#else
#define VCTR_ARM 0
#define VCTR_X64 1
#define VCTR_MAX_SIMD_REGISTER_SIZE 32
#endif
 
//==============================================================================
// Compiler defines
//==============================================================================
#if defined(__clang__)
#define VCTR_CLANG 1
#define VCTR_COMPILER_NAME clang
#elif defined(__GNUC__)
#define VCTR_GCC 1
#define VCTR_COMPILER_NAME GCC
#elif defined(_MSC_VER)
#define VCTR_MSVC 1
#define VCTR_COMPILER_NAME MSVC
// Since Visual Studio 2019, there is builtin SVML support in MSVC
// https://devblogs.microsoft.com/cppblog/msvc-backend-updates-in-visual-studio-2019-preview-2/
#define VCTR_HAS_SVML (_MSC_VER >= 1921)
#endif
 
#ifndef VCTR_CLANG
#define VCTR_CLANG 0
#endif
 
#ifndef VCTR_GCC
#define VCTR_GCC 0
#endif
 
#ifndef VCTR_MSVC
#define VCTR_MSVC 0
#endif
 
#ifndef VCTR_HAS_SVML
#define VCTR_HAS_SVML 0
#endif
 
//==============================================================================
// Build type defines
//==============================================================================
#if defined(DEBUG) || defined(_DEBUG) || ! (defined(NDEBUG) || defined(_NDEBUG))
#define VCTR_DEBUG 1
#else
#define VCTR_DEBUG 0
#endif
 
//==============================================================================
// User supplied defines
//==============================================================================
 
#ifndef VCTR_USE_IPP
#if __has_include(<ipp.h>)
#define VCTR_USE_IPP 1
#else
#define VCTR_USE_IPP 0
#endif
#endif
 
#ifndef VCTR_USE_GCEM
#if __has_include(<gcem.hpp>)
#define VCTR_USE_GCEM 1
#else
#define VCTR_USE_GCEM 0
#endif
#endif
 
#ifndef VCTR_ALIGNED_ARRAY
#define VCTR_ALIGNED_ARRAY 1
#endif
 
#if VCTR_WINDOWS
// This is needed for the CPU instruction set check found below
#include <intrin.h>
#endif
 
namespace vctr
{
 
struct CPUInstructionSets
{
    uint16_t sse4_1 : 1;
 
    uint16_t avx : 1;
 
    // Note: CPUs that support AVX2 always support FMA and AVX
    uint16_t avx2 : 1;
 
    // Note: CPUs that support FMA always support AVX
    uint16_t fma : 1;
 
    uint16_t neon : 1;
};
 
#if VCTR_WINDOWS
namespace detail
{
class X64InstructionSets
{
private:
    struct CPUFeatureFinder
    {
        CPUFeatureFinder()
        {
            std::array<int, 4> cpui;
            std::vector<std::array<int, 4>> data, extdata;
 
            // Calling __cpuid with 0x0 as the function_id argument
            // gets the number of the highest valid function ID.
            __cpuid (cpui.data(), 0);
            auto nIds = cpui[0];
 
            for (int i = 0; i <= nIds; ++i)
            {
                __cpuidex (cpui.data(), i, 0);
                data.push_back (cpui);
            }
 
            // load bitset with flags for function 0x00000001
            if (nIds >= 1)
            {
                f_1_ECX_ = data[1][2];
                f_1_EDX_ = data[1][3];
            }
 
            // load bitset with flags for function 0x00000007
            if (nIds >= 7)
            {
                f_7_EBX_ = data[7][1];
                f_7_ECX_ = data[7][2];
            }
 
            // Calling __cpuid with 0x80000000 as the function_id argument
            // gets the number of the highest valid extended ID.
            __cpuid (cpui.data(), 0x80000000);
            auto nExIds = cpui[0];
 
            for (int i = 0x80000000; i <= nExIds; ++i)
            {
                __cpuidex (cpui.data(), i, 0);
                extdata.push_back (cpui);
            }
 
            // load bitset with flags for function 0x80000001
            if (nExIds >= 0x80000001)
            {
                f_81_ECX_ = extdata[1][2];
                f_81_EDX_ = extdata[1][3];
            }
        };
 
        std::bitset<32> f_1_ECX_;
        std::bitset<32> f_1_EDX_;
        std::bitset<32> f_7_EBX_;
        std::bitset<32> f_7_ECX_;
        std::bitset<32> f_81_ECX_;
        std::bitset<32> f_81_EDX_;
    };
 
    inline static const CPUFeatureFinder cpuFeatures;
 
public:
    static bool hasFMA() { return cpuFeatures.f_1_ECX_[12]; }
    static bool hasSSE41() { return cpuFeatures.f_1_ECX_[19]; }
    static bool hasSSE42() { return cpuFeatures.f_1_ECX_[20]; }
    static bool hasAVX() { return cpuFeatures.f_1_ECX_[28]; }
    static bool hasAVX2() { return cpuFeatures.f_7_EBX_[5]; }
    static bool hasAVX512F() { return cpuFeatures.f_7_EBX_[16]; }
 
private:
};
} // namespace detail
 
inline CPUInstructionSets getSupportedCPUInstructionSets()
{
    return {
        .sse4_1 = detail::X64InstructionSets::hasSSE41(),
        .avx = detail::X64InstructionSets::hasAVX(),
        .avx2 = detail::X64InstructionSets::hasAVX2(),
        .fma = detail::X64InstructionSets::hasFMA(),
        .neon = false
    };
}
 
#elif VCTR_ARM
 
constexpr CPUInstructionSets getSupportedCPUInstructionSets()
{
    return {
        .sse4_1 = false,
        .avx = false,
        .avx2 = false,
        .fma = false,
        .neon = true
    };
}
 
#else
 
inline CPUInstructionSets getSupportedCPUInstructionSets()
{
    __builtin_cpu_init();
 
    return {
        .sse4_1 = __builtin_cpu_supports ("sse4.1"),
        .avx = __builtin_cpu_supports ("avx"),
        .avx2 = __builtin_cpu_supports ("avx2"),
        .fma = __builtin_cpu_supports ("fma"),
        .neon = false
    };
}
 
#endif
 
namespace detail
{
 
template <bool... settings>
consteval size_t trueCount()
{
    return (size_t (settings) + ...);
}
 
} // namespace detail
 
struct Config
{
    static const inline auto supportedCPUInstructionSets = getSupportedCPUInstructionSets();
 
    //==============================================================================
    // Platform config
    //==============================================================================
    static constexpr bool platformWindows = VCTR_WINDOWS;
 
    static constexpr bool platformApple = VCTR_APPLE;
 
    static constexpr bool platformLinux = VCTR_LINUX;
 
    static constexpr bool platformWasm = VCTR_WASM;
 
    static_assert (detail::trueCount<platformWindows, platformApple, platformLinux, platformWasm>() == 1, "Unsupported platform or platform detection error");
 
    //==============================================================================
    // CPU architecture config
    //==============================================================================
    static constexpr bool archARM = VCTR_ARM;
 
    static constexpr bool archX64 = VCTR_X64;
 
    static_assert (detail::trueCount<archARM, archX64>() == 1, "Unsupported architecture or architecture detection error");
 
    //==============================================================================
    // Compiler config
    //==============================================================================
    static constexpr bool compilerClang = VCTR_CLANG;
 
    static constexpr bool compilerGCC = VCTR_GCC;
 
    static constexpr bool compilerMSVC = VCTR_MSVC;
 
    static_assert (detail::trueCount<compilerClang, compilerGCC, compilerMSVC>() == 1, "Unsupported compiler or compiler detection error");
 
    static constexpr bool hasSVML = VCTR_HAS_SVML;
 
    //==============================================================================
    // User supplied config
    //==============================================================================
    static constexpr bool hasIPP = VCTR_USE_IPP && archX64;
 
    static constexpr bool alignedArray = VCTR_ALIGNED_ARRAY;
 
    //==============================================================================
    // Auto generated config
    //==============================================================================
    static constexpr size_t maxSIMDRegisterSize = VCTR_MAX_SIMD_REGISTER_SIZE;
};
 
} // namespace vctr