path: root/src/routines/levelx/xgemmstridedbatched.cpp

// =================================================================================================
// This file is part of the CLBlast project. The project is licensed under Apache Version 2.0. This
// project loosely follows the Google C++ styleguide and uses a tab-size of two spaces and a max-
// width of 100 characters per line.
//
// Author(s):
//   Cedric Nugteren <www.cedricnugteren.nl>
//
// This file implements the XgemmStridedBatched class (see the header for information about the class).
//
// =================================================================================================

#include "routines/levelx/xgemmstridedbatched.hpp"
#include "routines/level3/xgemm.hpp"

#include <string>
#include <vector>

namespace clblast {
// =================================================================================================

// Constructor: forwards to base class constructor
template <typename T>
XgemmStridedBatched<T>::XgemmStridedBatched(Queue &queue, EventPointer event, const std::string &name):
    Routine(queue, event, name, {"Copy","Pad","Transpose","Padtranspose","Xgemm","XgemmDirect","GemmRoutine"},
        PrecisionValue<T>(), {}, {
            #include "../../kernels/level3/level3.opencl"
            #include "../../kernels/level3/copy_fast.opencl"
            #include "../../kernels/level3/copy_pad.opencl"
            #include "../../kernels/level3/transpose_fast.opencl"
            #include "../../kernels/level3/transpose_pad.opencl"
            , // separated in multiple parts to prevent C1091 in MSVC 2013
            #include "../../kernels/level3/xgemm_direct_part1.opencl"
            #include "../../kernels/level3/xgemm_direct_part2.opencl"
            #include "../../kernels/level3/xgemm_direct_part3.opencl"
            , // separated in multiple parts to prevent C1091 in MSVC 2013
            #include "../../kernels/level3/xgemm_part1.opencl"
            #include "../../kernels/level3/xgemm_part2.opencl"
            , // separated in multiple parts to prevent C1091 in MSVC 2013
            #include "../../kernels/level3/xgemm_part3.opencl"
            #include "../../kernels/level3/xgemm_part4.opencl"
            , // separated in multiple parts to prevent C1091 in MSVC 2013
            #include "../../kernels/level3/xgemm_batched.opencl"
            #include "../../kernels/level3/xgemm_direct_batched.opencl"
        }) {
}

// =================================================================================================

// The main routine
template <typename T>
void XgemmStridedBatched<T>::DoGemmStridedBatched(const Layout layout, const Transpose a_transpose, const Transpose b_transpose,
                                                  const size_t m, const size_t n, const size_t k, const T alpha,
                                                  const Buffer<T> &a_buffer, const size_t a_offset, const size_t a_ld, const size_t a_stride,
                                                  const Buffer<T> &b_buffer, const size_t b_offset, const size_t b_ld, const size_t b_stride, const T beta,
                                                  const Buffer<T> &c_buffer, const size_t c_offset, const size_t c_ld, const size_t c_stride,
                                                  const size_t batch_count) {

  // Tests for a valid batch count
  if (batch_count < 1) {
    throw BLASError(StatusCode::kInvalidBatchCount);
  }

  // Makes sure the strides are valid
  if (c_stride == 0) { throw BLASError(StatusCode::kInvalidDimension); }

  // Two methods to choose from, select which one to run
  const auto do_gemm_direct = Xgemm<T>::UseDirectKernel(m, n, k, db_["XGEMM_MIN_INDIRECT_SIZE"]);
  const auto gemm_kernel_id = (do_gemm_direct) ? 0 : db_["GEMMK"];

  // Computes the transpose/conjugate options and sets the a/b/c sizes based on that
  bool a_do_transpose, b_do_transpose, c_do_transpose, a_conjugate, b_conjugate;
  size_t a_one, a_two, b_one, b_two, c_one, c_two;
  Xgemm<T>::ProcessArguments(layout, a_transpose, b_transpose, m, n, k,
                             a_one, a_two, b_one, b_two, c_one, c_two,
                             a_do_transpose, b_do_transpose, c_do_transpose, a_conjugate, b_conjugate,
                             gemm_kernel_id);

  // Tests the matrices for validity
  for (auto batch = size_t{0}; batch < batch_count; ++batch) {
    TestMatrixA(a_one, a_two, a_buffer, a_offset + a_stride * batch, a_ld);
    TestMatrixB(b_one, b_two, b_buffer, b_offset + b_stride * batch, b_ld);
    TestMatrixC(c_one, c_two, c_buffer, c_offset + c_stride * batch, c_ld);
  }

  // Selects which version of the batched GEMM to run
  if (do_gemm_direct) { // single generic kernel
    BatchedGemmDirect(m, n, k, alpha,
                      a_buffer, a_offset, a_ld, a_stride,
                      b_buffer, b_offset, b_ld, b_stride, beta,
                      c_buffer, c_offset, c_ld, c_stride,
                      a_do_transpose, b_do_transpose, c_do_transpose, a_conjugate, b_conjugate,
                      batch_count);
  }
  else { // pre/post-processing plus a very fast kernel
    BatchedGemmIndirect(m, n, k, alpha,
                        a_buffer, a_offset, a_ld, a_stride,
                        b_buffer, b_offset, b_ld, b_stride, beta,
                        c_buffer, c_offset, c_ld, c_stride,
                        a_do_transpose, b_do_transpose, c_do_transpose, a_conjugate, b_conjugate,
                        a_one, a_two, b_one, b_two, c_one, c_two, batch_count);
  }
}


// =================================================================================================

// The indirect version of batched GEMM. This uses the faster but non-general kernel. It has specific
// requirements, but several pre and post-processing kernels take care of those. However, the
// overhead of these extra kernels might not be ideal for certain devices/arguments.
template <typename T>
void XgemmStridedBatched<T>::BatchedGemmIndirect(const size_t m, const size_t n, const size_t k, const T alpha,
                                                 const Buffer<T> &a_buffer, const size_t a_offset, const size_t a_ld, const size_t a_stride,
                                                 const Buffer<T> &b_buffer, const size_t b_offset, const size_t b_ld, const size_t b_stride, const T beta,
                                                 const Buffer<T> &c_buffer, const size_t c_offset, const size_t c_ld, const size_t c_stride,
                                                 const bool a_do_transpose, const bool b_do_transpose, const bool c_do_transpose,
                                                 const bool a_conjugate, const bool b_conjugate,
                                                 const size_t a_one, const size_t a_two,
                                                 const size_t b_one, const size_t b_two,
                                                 const size_t c_one, const size_t c_two,
                                                 const size_t batch_count) {

  // Calculates the ceiled versions of m, n, and k
  const auto m_ceiled = Ceil(Ceil(m, db_["MWG"]), db_["VWM"]);
  const auto n_ceiled = Ceil(Ceil(n, db_["NWG"]), db_["VWN"]);
  const auto k_ceiled = Ceil(Ceil(k, db_["KWG"]), db_["VWM"]);

  // Computes the first and second "internal" (ceiled) dimensions of the 3 matrices taking into account
  // whether the matrices need to be rotated or not for the kernel.
  size_t a_one_i, a_two_i, b_one_i, b_two_i, c_one_i, c_two_i;
  Xgemm<T>::CalculateInternalDimensions(m, n, k, db_["MWG"], db_["NWG"], db_["KWG"],
                                        a_one_i, a_two_i, b_one_i, b_two_i, c_one_i, c_two_i,
                                        db_["GEMMK"]);

  // Determines whether or not temporary matrices are needed
  auto a_no_temp = a_one == a_one_i && a_two == a_two_i && a_ld == a_one && !a_do_transpose && !a_conjugate;
  auto b_no_temp = b_one == b_one_i && b_two == b_two_i && b_ld == b_one && !b_do_transpose && !b_conjugate;
  auto c_no_temp = c_one == c_one_i && c_two == c_two_i && c_ld == c_one && !c_do_transpose;

  // Creates the temporary matrices
  const auto a_temp = (a_no_temp) ? a_buffer : Buffer<T>(context_, batch_count * a_one_i * a_two_i);
  const auto b_temp = (b_no_temp) ? b_buffer : Buffer<T>(context_, batch_count * b_one_i * b_two_i);
  const auto c_temp = (c_no_temp) ? c_buffer : Buffer<T>(context_, batch_count * c_one_i * c_two_i);

  // Events of all kernels (including pre/post processing kernels)
  auto eventWaitList = std::vector<Event>();
  auto emptyEventList = std::vector<Event>();

  // Runs the pre-processing kernel for matrix A. This transposes the matrix, but also pads zeros
  // to fill it up until it reaches a certain multiple of size (kernel parameter dependent). In
  // case nothing has to be done, these kernels can be skipped.
  if (!a_no_temp) {
    auto eventProcessA = Event();
    PadCopyTransposeMatrixStridedBatched(queue_, device_, db_, eventProcessA.pointer(), emptyEventList,
                                         a_one, a_two, a_ld, a_offset, a_stride, a_buffer,
                                         a_one_i, a_two_i, a_one_i, 0, a_one_i * a_two_i, a_temp,
                                         program_, true, a_do_transpose, a_conjugate, batch_count);
    eventWaitList.push_back(eventProcessA);
  }

  // As above, but now for matrix B
  if (!b_no_temp) {
    auto eventProcessB = Event();
    PadCopyTransposeMatrixStridedBatched(queue_, device_, db_, eventProcessB.pointer(), emptyEventList,
                                         b_one, b_two, b_ld, b_offset, b_stride, b_buffer,
                                         b_one_i, b_two_i, b_one_i, 0, b_one_i * b_two_i, b_temp,
                                         program_, true, b_do_transpose, b_conjugate, batch_count);
    eventWaitList.push_back(eventProcessB);
  }

  // As above, but now for matrix C
  if (!c_no_temp) {
    auto eventProcessC = Event();
    PadCopyTransposeMatrixStridedBatched(queue_, device_, db_, eventProcessC.pointer(), emptyEventList,
                                         c_one, c_two, c_ld, c_offset, c_stride, c_buffer,
                                         c_one_i, c_two_i, c_one_i, 0, c_one_i * c_two_i, c_temp,
                                         program_, true, c_do_transpose, false, batch_count);
    eventWaitList.push_back(eventProcessC);
  }

  // Retrieves the Xgemm kernel from the compiled binary
  auto kernel = Kernel(program_, "XgemmStridedBatched");

  // Sets the kernel arguments
  kernel.SetArgument(0, static_cast<int>(m_ceiled));
  kernel.SetArgument(1, static_cast<int>(n_ceiled));
  kernel.SetArgument(2, static_cast<int>(k_ceiled));
  kernel.SetArgument(3, GetRealArg(alpha));
  kernel.SetArgument(4, GetRealArg(beta));
  kernel.SetArgument(5, a_temp());
  kernel.SetArgument(6, static_cast<int>(a_one_i));
  kernel.SetArgument(7, static_cast<int>(a_two_i));
  kernel.SetArgument(8, b_temp());
  kernel.SetArgument(9, static_cast<int>(b_one_i));
  kernel.SetArgument(10, static_cast<int>(b_two_i));
  kernel.SetArgument(11, c_temp());
  kernel.SetArgument(12, static_cast<int>(c_one_i));
  kernel.SetArgument(13, static_cast<int>(c_two_i));

  // Computes the global and local thread sizes
  const auto global = std::vector<size_t>{
      (c_one_i * db_["MDIMC"]) / db_["MWG"],
      (c_two_i * db_["NDIMC"]) / db_["NWG"],
      batch_count
  };
  const auto local = std::vector<size_t>{db_["MDIMC"], db_["NDIMC"], 1};

  // Launches the kernel
  auto eventKernel = Event();
  auto eventPointer = (!c_no_temp) ? eventKernel.pointer() : event_;
  RunKernel(kernel, queue_, device_, global, local, eventPointer, eventWaitList);

  // Runs the post-processing kernel if needed
  if (!c_no_temp) {
    eventWaitList.push_back(eventKernel);
    PadCopyTransposeMatrixStridedBatched(queue_, device_, db_, event_, eventWaitList,
                                         c_one_i, c_two_i, c_one_i, 0, c_one_i * c_two_i, c_temp,
                                         c_one, c_two, c_ld, c_offset, c_stride, c_buffer,
                                         program_, false, c_do_transpose, false, batch_count);
  }
}

// =================================================================================================

// The direct version of batched GEMM, requiring just one kernel, no pre or post-processing kernels.
template <typename T>
void XgemmStridedBatched<T>::BatchedGemmDirect(const size_t m, const size_t n, const size_t k, const T alpha,
                                               const Buffer<T> &a_buffer, const size_t a_offset, const size_t a_ld, const size_t a_stride,
                                               const Buffer<T> &b_buffer, const size_t b_offset, const size_t b_ld, const size_t b_stride, const T beta,
                                               const Buffer<T> &c_buffer, const size_t c_offset, const size_t c_ld, const size_t c_stride,
                                               const bool a_do_transpose, const bool b_do_transpose, const bool c_do_transpose,
                                               const bool a_conjugate, const bool b_conjugate,
                                               const size_t batch_count) {

  // Retrieves the proper XgemmDirect kernel from the compiled binary
  const auto name = (a_do_transpose) ? (b_do_transpose ? "XgemmDirectStridedBatchedTT" : "XgemmDirectStridedBatchedTN") :
                    (b_do_transpose ? "XgemmDirectStridedBatchedNT" : "XgemmDirectStridedBatchedNN");
  auto kernel = Kernel(program_, name);

  // Sets the kernel arguments
  kernel.SetArgument(0, static_cast<int>(m));
  kernel.SetArgument(1, static_cast<int>(n));
  kernel.SetArgument(2, static_cast<int>(k));
  kernel.SetArgument(3, GetRealArg(alpha));
  kernel.SetArgument(4, GetRealArg(beta));
  kernel.SetArgument(5, a_buffer());
  kernel.SetArgument(6, static_cast<int>(a_offset));
  kernel.SetArgument(7, static_cast<int>(a_ld));
  kernel.SetArgument(8, static_cast<int>(a_stride));
  kernel.SetArgument(9, b_buffer());
  kernel.SetArgument(10, static_cast<int>(b_offset));
  kernel.SetArgument(11, static_cast<int>(b_ld));
  kernel.SetArgument(12, static_cast<int>(b_stride));
  kernel.SetArgument(13, c_buffer());
  kernel.SetArgument(14, static_cast<int>(c_offset));
  kernel.SetArgument(15, static_cast<int>(c_ld));
  kernel.SetArgument(16, static_cast<int>(c_stride));
  kernel.SetArgument(17, static_cast<int>(c_do_transpose));
  kernel.SetArgument(18, static_cast<int>(a_conjugate));
  kernel.SetArgument(19, static_cast<int>(b_conjugate));

  // Computes the global and local thread sizes
  const auto m_ceiled = Ceil(m, db_["WGD"]);
  const auto n_ceiled = Ceil(n, db_["WGD"]);
  const auto global = std::vector<size_t>{
      (m_ceiled * db_["MDIMCD"]) / db_["WGD"],
      (n_ceiled * db_["NDIMCD"]) / db_["WGD"],
      batch_count
  };
  const auto local = std::vector<size_t>{db_["MDIMCD"], db_["NDIMCD"], 1};

  // Launches the kernel
  RunKernel(kernel, queue_, device_, global, local, event_);
}

// =================================================================================================

// Compiles the templated class
template class XgemmStridedBatched<half>;
template class XgemmStridedBatched<float>;
template class XgemmStridedBatched<double>;
template class XgemmStridedBatched<float2>;
template class XgemmStridedBatched<double2>;

// =================================================================================================
} // namespace clblast