TheRock/rocm-libraries/projects/hipblaslt/tensilelite/client/include/Reference.hpp

/*******************************************************************************
 *
 * MIT License
 *
 * Copyright (C) 2022-2024 Advanced Micro Devices, Inc. All rights reserved.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 *
 *******************************************************************************/

#pragma once

#include <Tensile/ContractionProblem.hpp>

#include <Tensile/DataTypes.hpp>

namespace TensileLite
{
    namespace Client
    {
        // AlmostEqual tolerance constants per type.
        // Formula: |a - b| < tolerance * (|a| + |b| + 1)
        constexpr float  AlmostEqualTolerance_Half     = 0.01f;
        constexpr float  AlmostEqualTolerance_BFloat16 = 0.1f;
        // tolerance * epsilon = 2 * 0.0625; 2*eps needed for SR
        constexpr float  AlmostEqualTolerance_Float8   = 0.125f;
        // tolerance * epsilon = 2 * 0.125; 2*eps needed for SR
        constexpr float  AlmostEqualTolerance_BFloat8  = 0.25f;
        // 7 digits precision - 2
        constexpr float  AlmostEqualTolerance_Float    = 0.0001f;
        // 15 digits precision - 2
        constexpr double AlmostEqualTolerance_Double   = 1e-12;

        // threshold is largest allowed delta. -1 uses default for each type
        template <typename T>
        inline bool AlmostEqual(T a, T b, double threshold = -1.0);

        template <>
        inline bool AlmostEqual(Half a, Half b, double threshold)
        {
            float fa      = static_cast<float>(a);
            float fb      = static_cast<float>(b);
            float absDiff = std::fabs(fa - fb);
            return fa == fb
                   || absDiff < AlmostEqualTolerance_Half * (std::fabs(fa) + std::fabs(fb) + 1.0f);
        }

        template <>
        inline bool AlmostEqual(Float8 a, Float8 b, double threshold)
        {
            float fa      = static_cast<float>(a);
            float fb      = static_cast<float>(b);
            float absDiff = std::fabs(fa - fb);
            return fa == fb
                   || absDiff < AlmostEqualTolerance_Float8 * (std::fabs(fa) + std::fabs(fb) + 1.0f);
        }

        template <>
        inline bool AlmostEqual(BFloat8 a, BFloat8 b, double threshold)
        {
            float fa      = static_cast<float>(a);
            float fb      = static_cast<float>(b);
            float absDiff = std::fabs(fa - fb);
            
            return fa == fb
                   || absDiff < AlmostEqualTolerance_BFloat8 * (std::fabs(fa) + std::fabs(fb) + 1.0f);
        }

        template <>
        inline bool AlmostEqual(Float8_fnuz a, Float8_fnuz b, double threshold)
        {
            float fa      = static_cast<float>(a);
            float fb      = static_cast<float>(b);
            float absDiff = std::fabs(fa - fb);
            return fa == fb
                   || absDiff < AlmostEqualTolerance_Float8 * (std::fabs(fa) + std::fabs(fb) + 1.0f);
        }

        template <>
        inline bool AlmostEqual(BFloat8_fnuz a, BFloat8_fnuz b, double threshold)
        {
            float fa      = static_cast<float>(a);
            float fb      = static_cast<float>(b);
            float absDiff = std::fabs(fa - fb);
            return fa == fb
                   || absDiff < AlmostEqualTolerance_BFloat8 * (std::fabs(fa) + std::fabs(fb) + 1.0f);
        }

        template <>
        inline bool AlmostEqual(BFloat16 a, BFloat16 b, double threshold)
        {
            float fa      = static_cast<float>(a);
            float fb      = static_cast<float>(b);
            float absDiff = std::fabs(fa - fb);
            return fa == fb
                   || absDiff < AlmostEqualTolerance_BFloat16 * (std::fabs(fa) + std::fabs(fb) + 1.0f);
        }

        template <>
        inline bool AlmostEqual(float a, float b, double threshold)
        {
            float tol     = (threshold > 0.0) ? static_cast<float>(threshold) : AlmostEqualTolerance_Float;
            float absDiff = std::fabs(a - b);
            return a == b
                   || absDiff < tol * (std::fabs(a) + std::fabs(b) + 1);
        }

        template <>
        inline bool AlmostEqual(double a, double b, double threshold)
        {
            double absDiff = std::fabs(a - b);
            return a == b
                   || absDiff < AlmostEqualTolerance_Double * (std::fabs(a) + std::fabs(b) + 1);
        }
        template <>
        inline bool AlmostEqual(int8_t a, int8_t b, double threshold)
        {
            return a == b;
        }
        template <>
        inline bool AlmostEqual(int a, int b, double threshold)
        {
            return a == b;
        }
        template <>
        inline bool AlmostEqual(unsigned int a, unsigned int b, double threshold)
        {
            return a == b;
        }
        template <>
        inline bool AlmostEqual(std::complex<float> a, std::complex<float> b, double threshold)
        {
            return AlmostEqual(a.real(), b.real()) && AlmostEqual(a.imag(), b.imag());
        }

        template <>
        inline bool AlmostEqual(std::complex<double> a, std::complex<double> b, double threshold)
        {
            return AlmostEqual(a.real(), b.real()) && AlmostEqual(a.imag(), b.imag());
        }

        template <typename Inputs,
                  typename Accumulator = typename Inputs::DType,
                  typename MathOpAccum = Accumulator>
        struct ReferenceSolution
        {
            static void SolveCPU(ContractionProblemGemm const& contraction,
                                 ContractionInputs const&      inputs,
                                 size_t                        elementsToValidate);
            static void SolveCPU(ContractionProblemGroupedGemm const& contractions,
                                 ContractionGroupedInputs const&      inputs,
                                 size_t                               elementsToValidate);
        };

        void SolveCPU(ContractionProblem const* contraction,
                      ProblemInputs const*      inputs,
                      size_t                    elementsToValidate);

        // Specialized solver for ungrouped GEMM problems. There are currently 2 implementations
        // of reference CPU GEMM. One is substantially faster but only supports a limited set of
        // data types and problem configurations. The other is a more general but slower implementation.
        // When `tryFastPath` is true, the function will attempt to use the fast implementation first,
        // and fall back to the general implementation if the fast path is not applicable. If `tryFastPath`
        // is false, the function will directly use the general implementation.
        void SolveGemmCPU(ContractionProblemGemm const& problem,
                          ContractionInputs const&      inputs,
                          size_t                        elementsToValidate,
                          bool                          tryFastPath = true);

        // Check whether a given contraction problem is eligible for the fast CPU GEMM path.
        // This inspects problem geometry, data types, and feature flags but does not
        // look at runtime input buffers.
        bool isFastPathEligible(ContractionProblemGemm const& problem);

    } // namespace Client
} // namespace TensileLite