thread_multiply_add.h

Go to the documentation of this file.

/***************************************************************************************************
 * Copyright (c) 2017-2018, NVIDIA CORPORATION.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without modification, are permitted
 * provided that the following conditions are met:
 *     * Redistributions of source code must retain the above copyright notice, this list of
 *       conditions and the following disclaimer.
 *     * Redistributions in binary form must reproduce the above copyright notice, this list of
 *       conditions and the following disclaimer in the documentation and/or other materials
 *       provided with the distribution.
 *     * Neither the name of the NVIDIA CORPORATION nor the names of its contributors may be used
 *       to endorse or promote products derived from this software without specific prior written
 *       permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR
 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
 * FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
 * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
 * STRICT LIABILITY, OR TOR (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 **************************************************************************************************/
#pragma once

#include <cutlass/fragment.h>

namespace cutlass {
namespace gemm {


template <typename AccumulatorsPerThread_,
          typename ThreadsPerWarp_,
          typename ScalarA_,
          typename ScalarB_,
          typename ScalarC_>
struct ThreadMultiplyAdd {
  typedef Shape<1, 1, 1, 1> InstructionShape;
  typedef AccumulatorsPerThread_ AccumulatorsPerThread;
  typedef ThreadsPerWarp_ ThreadsPerWarp;
  typedef typename ShapeMul<AccumulatorsPerThread, ThreadsPerWarp>::Shape AccumulatorsPerWarp;
  typedef ScalarA_ ScalarA;
  typedef Fragment<ScalarA, AccumulatorsPerThread::kW> FragmentA;
  typedef ScalarB_ ScalarB;
  typedef Fragment<ScalarB, AccumulatorsPerThread::kH> FragmentB;
  typedef ScalarC_ ScalarC;
  typedef Fragment<ScalarC, AccumulatorsPerThread::kH * AccumulatorsPerThread::kW, 16> Accumulators;

  CUTLASS_DEVICE ThreadMultiplyAdd() {}

  CUTLASS_DEVICE void multiply_add(FragmentA const& a,
                                   FragmentB const& b,
                                   Accumulators const& c,
                                   Accumulators& d) {
    for (int j = 0; j < AccumulatorsPerThread::kH; ++j) {
      for (int i = 0; i < AccumulatorsPerThread::kW; ++i) {
        d[j * AccumulatorsPerThread::kW + i] = a[i] * b[j] + c[j * AccumulatorsPerThread::kW + i];
      }
    }
  }
};


}  // namespace gemm
}  // namespace cutlass