1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190
|
/* ************************************************************************
* Copyright (C) 2018-2023 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 cop-
* ies 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 IM-
* PLIED, 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 CONNE-
* CTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
*
* ************************************************************************ */
#pragma once
#include <cmath>
#include <type_traits>
#include <vector>
#include "device_vector.hpp"
#include "host_alloc.hpp"
//!
//! @brief Pseudo-matrix subclass which uses host memory.
//!
template <typename T>
struct host_matrix : std::vector<T, host_memory_allocator<T>>
{
// Inherit constructors
using std::vector<T, host_memory_allocator<T>>::vector;
//!
//! @brief Constructor.
//!
host_matrix(size_t m, size_t n, size_t lda)
: std::vector<T, host_memory_allocator<T>>(n * lda)
, m_m(m)
, m_n(n)
, m_lda(lda)
{
}
//!
//! @brief Copy constructor from host_matrix of other types convertible to T
//!
template <typename U, std::enable_if_t<std::is_convertible<U, T>{}, int> = 0>
host_matrix(const host_matrix<U>& x)
: std::vector<T, host_memory_allocator<T>>(x.size())
, m_m(x.size())
, m_n(1)
, m_lda(1)
{
for(size_t i = 0; i < m_m; ++i)
(*this)[i] = x[i];
}
//!
//! @brief Decay into pointer wherever pointer is expected
//!
operator T*()
{
return this->data();
}
//!
//! @brief Decay into constant pointer wherever constant pointer is expected
//!
operator const T*() const
{
return this->data();
}
//!
//! @brief Transfer from a device matrix.
//! @param that That device matrix.
//! @return the hip error.
//!
hipError_t transfer_from(const device_matrix<T>& that)
{
hipError_t hip_err;
if(that.use_HMM && hipSuccess != (hip_err = hipDeviceSynchronize()))
return hip_err;
return hipMemcpy(*this,
that,
sizeof(T) * this->size(),
that.use_HMM ? hipMemcpyHostToHost : hipMemcpyDeviceToHost);
}
//!
//! @brief Transfer only the first matrix from a device_strided_batch matrix.
//! @param that That device_strided_batch matrix.
//! @return the hip error.
//!
hipError_t transfer_one_matrix_from(const device_strided_batch_matrix<T>& that)
{
hipError_t hip_err;
if(that.use_HMM && hipSuccess != (hip_err = hipDeviceSynchronize()))
return hip_err;
return hipMemcpy(*this,
that,
sizeof(T) * this->size(),
that.use_HMM ? hipMemcpyHostToHost : hipMemcpyDeviceToHost);
}
//!
//! @brief Returns the rows of the Matrix.
//!
size_t m() const
{
return this->m_m;
}
//!
//! @brief Returns the cols of the Matrix.
//!
size_t n() const
{
return this->m_n;
}
//!
//! @brief Returns the leading dimension of the Matrix.
//!
size_t lda() const
{
return this->m_lda;
}
//!
//! @brief Returns the batch count (always 1).
//!
static constexpr rocblas_int batch_count()
{
return 1;
}
//!
//! @brief Returns the stride (out of context, always 0)
//!
static constexpr rocblas_stride stride()
{
return 0;
}
//!
//! @brief Random access to the Matrices.
//! @param batch_index the batch index.
//! @return The mutable pointer.
//!
T* operator[](int64_t batch_index)
{
return this->data();
}
//!
//! @brief Constant random access to the Matrices.
//! @param batch_index the batch index.
//! @return The non-mutable pointer.
//!
const T* operator[](int64_t batch_index) const
{
return this->data();
}
//!
//! @brief Check if memory exists (out of context, always hipSuccess)
//!
static constexpr hipError_t memcheck()
{
return hipSuccess;
}
private:
size_t m_m = 0;
size_t m_n = 0;
size_t m_lda = 0;
};
|