src/runtime/GLES_COMPUTE/functions/GCGEMM.cpp

/*
 * Copyright (c) 2017-2019 ARM Limited.
 *
 * SPDX-License-Identifier: MIT
 *
 * 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.
 */
#include "arm_compute/runtime/GLES_COMPUTE/functions/GCGEMM.h"

#include "arm_compute/core/Error.h"
#include "arm_compute/core/GLES_COMPUTE/GCHelpers.h"
#include "arm_compute/core/GLES_COMPUTE/IGCTensor.h"
#include "arm_compute/core/GLES_COMPUTE/kernels/GCGEMMInterleave4x4Kernel.h"
#include "arm_compute/core/GLES_COMPUTE/kernels/GCGEMMMatrixAdditionKernel.h"
#include "arm_compute/core/GLES_COMPUTE/kernels/GCGEMMMatrixMultiplyKernel.h"
#include "arm_compute/core/GLES_COMPUTE/kernels/GCGEMMTranspose1xWKernel.h"
#include "arm_compute/core/Helpers.h"
#include "arm_compute/core/TensorInfo.h"
#include "arm_compute/core/Types.h"
#include "arm_compute/core/Validate.h"
#include "arm_compute/runtime/GLES_COMPUTE/GCScheduler.h"
#include "arm_compute/runtime/ITensorAllocator.h"

using namespace arm_compute;

namespace
{
Status validate_arguments(const ITensorInfo *a, const ITensorInfo *b, const IGCTensor *c, const ITensorInfo *output, const float alpha, const float beta, const GEMMInfo &gemm_info = GEMMInfo())
{
    ARM_COMPUTE_ERROR_ON_NULLPTR(a, b, output);

    ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(a, 1, DataType::F16, DataType::F32);
    ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(a, b, output);
    ARM_COMPUTE_ERROR_ON_MSG(gemm_info.is_a_reshaped(), "Matrix A already reshaped is not supported");
    ARM_COMPUTE_ERROR_ON_MSG(gemm_info.is_b_reshaped(), "Matrix B already reshaped is not supported");

    if(c != nullptr)
    {
        ARM_COMPUTE_ERROR_ON_MISMATCHING_DATA_TYPES(a, c->info());
        ARM_COMPUTE_ERROR_ON_MSG(a->dimension(1) != c->info()->dimension(1), "The C matrix must have the same number of rows as the matrix A");
        ARM_COMPUTE_ERROR_ON_MSG(b->dimension(0) != c->info()->dimension(0), "The C matrix must have the same number of columns as the matrix B");
    }

    if(output->total_size() != 0)
    {
        ARM_COMPUTE_RETURN_ERROR_ON_MSG(b->dimension(0) != output->dimension(0), "The output matrix must have the same number of columns as the matrix B");
        ARM_COMPUTE_RETURN_ERROR_ON_MSG(a->dimension(1) != output->dimension(1), "The output matrix must have the same number of rows as the matrix A");
    }

    ARM_COMPUTE_RETURN_ERROR_ON_MSG(a->dimension(0) != b->dimension(1), "The product AB is defined only if the number of columns in A is equal to the number of rows in B");

    ARM_COMPUTE_UNUSED(alpha);
    ARM_COMPUTE_UNUSED(beta);
    ARM_COMPUTE_UNUSED(gemm_info);
    return Status{};
}
} // namespace

GCGEMM::GCGEMM(std::shared_ptr<IMemoryManager> memory_manager)
    : _memory_group(std::move(memory_manager)), _interleave_kernel(), _transpose_kernel(), _mm_kernel(), _ma_kernel(), _tmp_a(), _tmp_b(), _original_b(nullptr), _is_interleaved_transposed(false),
      _run_addition(false), _reshape_b_only_on_first_run(false), _is_prepared(false)
{
}

void GCGEMM::configure(const IGCTensor *a, const IGCTensor *b, const IGCTensor *c, IGCTensor *output, float alpha, float beta, const GEMMInfo &gemm_info)
{
    ARM_COMPUTE_ERROR_ON_NULLPTR(a, b, output);

    // Perform validation step
    ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(a->info(), b->info(), c, output->info(), alpha, beta, gemm_info));

    // Check if we need to reshape the matrix B only on the first run
    _reshape_b_only_on_first_run = gemm_info.reshape_b_only_on_first_run();
    _is_prepared                 = false;
    _original_b                  = b;

    const IGCTensor *matrix_a = a;
    const IGCTensor *matrix_b = b;

    // Get the GPU target
    const GPUTarget gpu_target = GCScheduler::get().get_target();

    // Set the target for the kernels
    _interleave_kernel.set_target(gpu_target);
    _mm_kernel.set_target(gpu_target);

    // Arguments used by GEMMReshapeInfo
    // If we pass the matrix A and matrix B reshaped to GCGEMMMatrixMultiplyKernel, we need to pass m, n, k, mult_transpose1xW_width and mult_interleave4x4_height to GCGEMMReshapeInfo
    // in order to know how the matrices have been reshaped
    const int m                         = a->info()->dimension(1);
    const int n                         = b->info()->dimension(0);
    const int k                         = a->info()->dimension(0);
    int       mult_transpose1xW_width   = 1;
    int       mult_interleave4x4_height = 1;

    // If the input tensor has less than 16 rows, we run a special version of GEMM without reshaping the input tensors
    _is_interleaved_transposed = a->info()->dimension(1) > 16;

    if(_is_interleaved_transposed)
    {
        matrix_a = &_tmp_a;
        matrix_b = &_tmp_b;

        // Manage intermediate buffers
        _memory_group.manage(&_tmp_a);
        if(!_reshape_b_only_on_first_run)
        {
            _memory_group.manage(&_tmp_b);
        }
        // _tmp_a and _tmp_b will be auto configured in _interleave_kernel and in _transpose_kernel

        // Configure interleave kernel
        _interleave_kernel.configure(a, &_tmp_a);

        // Configure transpose kernel
        _transpose_kernel.configure(b, &_tmp_b);
    }

    _mm_kernel.configure(matrix_a, matrix_b, output, alpha, _is_interleaved_transposed, GEMMReshapeInfo(m, n, k, mult_transpose1xW_width, mult_interleave4x4_height));

    if(_is_interleaved_transposed)
    {
        // Allocate intermediate tensors
        _tmp_a.allocator()->allocate();
        if(!_reshape_b_only_on_first_run)
        {
            _tmp_b.allocator()->allocate();
        }
    }

    // Configure matrix addition kernel
    if(beta != 0 && c != nullptr)
    {
        _ma_kernel.configure(c, output, beta);
        _run_addition = true;
    }
}

Status GCGEMM::validate(const ITensorInfo *a, const ITensorInfo *b, const IGCTensor *c, const ITensorInfo *output, const float alpha, const float beta, const GEMMInfo &gemm_info)
{
    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(a, b, c, output, alpha, beta, gemm_info));
    return Status{};
}

void GCGEMM::run()
{
    prepare();

    MemoryGroupResourceScope scope_mg(_memory_group);

    if(_is_interleaved_transposed)
    {
        // Run interleave kernel
        GCScheduler::get().dispatch(_interleave_kernel, false);

        if(!_reshape_b_only_on_first_run)
        {
            // Run transpose kernel
            GCScheduler::get().dispatch(_transpose_kernel, false);
        }

        GCScheduler::get().memory_barrier();
    }

    // Run matrix multiply kernel
    GCScheduler::get().dispatch(_mm_kernel, !_run_addition);

    // Run matrix addition kernel
    if(_run_addition)
    {
        GCScheduler::get().memory_barrier();
        GCScheduler::get().dispatch(_ma_kernel);
    }
}

void GCGEMM::prepare()
{
    if(!_is_prepared)
    {
        if(_is_interleaved_transposed && _reshape_b_only_on_first_run)
        {
            ARM_COMPUTE_ERROR_ON(!_original_b->is_used());

            // Run transpose kernel
            _tmp_b.allocator()->allocate();
            GCScheduler::get().dispatch(_transpose_kernel, false);
            GCScheduler::get().memory_barrier();

            // Mark original weights tensor as unused
            _original_b->mark_as_unused();
        }

        _is_prepared = true;
    }
}