CudaVector/CudaVector.cu at master · shreyanshsaha/CudaVector · GitHub

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
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
#pragma once
#include <cuda.h>
#include <cuda_runtime.h>
#include <cuda_runtime_api.h>

#include "CudaVector.h"

/**
 * Kernel Functions
 */

template <class T>
__global__ void add(T* a, T* b, T* c, unsigned long n) {
  unsigned long index = threadIdx.x + blockIdx.x * blockDim.x;
  if (index < n) c[index] = a[index] + b[index];
}

template <class T>
__global__ void subtract(T* a, T* b, T* c, unsigned long n) {
  unsigned long index = threadIdx.x + blockIdx.x * blockDim.x;
  if (index < n) c[index] = a[index] - b[index];
}

template <class T>
__global__ void dotProduct(T* a, T* b, T* c, unsigned long n) {
  unsigned long index = threadIdx.x + blockIdx.x * blockDim.x;
  if (index < n) c[index] = a[index] * b[index];
}

template <class T>
__global__ void constantMultiply(T* a, double b, T* c, unsigned long n) {
  unsigned long index = threadIdx.x + blockIdx.x * blockDim.x;
  if (index < n) c[index] = a[index] * b;
}

template <class T>
__global__ void contantDivide(T* a, double b, T* c, unsigned long n) {
  unsigned long index = threadIdx.x + blockIdx.x * blockDim.x;
  if (index < n) c[index] = a[index] / b;
}

/** =================
 *  Class definitions
 *  =================
 */

/**
 * Creates CudaVector object and sets the size
 */
template <class T>
CudaVector<T>::CudaVector(unsigned long size) {
  this->_array = new T[size];
  this->_size = size;
}

/**
 * Free up memory
 */
template <class T>
CudaVector<T>::~CudaVector() {
  delete[] _array;
}

/**
 * Adds vector in GPU
 *
 * Returns CudaVector object
 */
template <class T>
CudaVector<T> CudaVector<T>::operator+(const CudaVector<T>& a) {
  if (this->_size != a._size)
    throw CudaVectorException("Vector size not same!");

  // create result array
  CudaVector<T> result(this->_size);

  T *d_a, *d_b, *d_c;
  const size_t size = this->_size * sizeof(T);

  // allocate memory
  cudaMalloc((void**)&d_a, size);
  cudaMalloc((void**)&d_b, size);
  cudaMalloc((void**)&d_c, size);

  // copy memory to device
  cudaMemcpy(d_a, this->_array, size, cudaMemcpyHostToDevice);
  cudaMemcpy(d_b, a._array, size, cudaMemcpyHostToDevice);

  add<<<NUM_BLOCKS, NUM_THREADS_PER_BLOCK>>>(d_a, d_b, d_c, this->_size);

  // copy to host
  cudaMemcpy(result._array, d_c, size, cudaMemcpyDeviceToHost);

  return result;
}

/**
 * Subtract vector in GPU
 *
 * Returns CudaVector object
 */
template <class T>
CudaVector<T> CudaVector<T>::operator-(const CudaVector<T>& a) {
  if (this->_size != a._size)
    throw CudaVectorException("Vector size not same!");

  // create result array
  CudaVector<T> result(this->_size);

  T *d_a, *d_b, *d_c;
  const size_t size = this->_size * sizeof(T);

  // allocate memory
  cudaMalloc((void**)&d_a, size);
  cudaMalloc((void**)&d_b, size);
  cudaMalloc((void**)&d_c, size);

  // copy memory to device
  cudaMemcpy(d_a, this->_array, size, cudaMemcpyHostToDevice);
  cudaMemcpy(d_b, a._array, size, cudaMemcpyHostToDevice);

  subtract<<<NUM_BLOCKS, NUM_THREADS_PER_BLOCK>>>(d_a, d_b, d_c, this->_size);

  // copy to host
  cudaMemcpy(result._array, d_c, size, cudaMemcpyDeviceToHost);

  return result;
}

/**
 * Dot Product vector in GPU
 *
 * Returns CudaVector object
 */
template <class T>
CudaVector<T> CudaVector<T>::operator*(const CudaVector<T>& a) {
  if (this->_size != a._size)
    throw CudaVectorException("Vector size not same!");

  // create result array
  CudaVector<T> result(this->_size);

  T *d_a, *d_b, *d_c;
  const size_t size = this->_size * sizeof(T);

  // allocate memory
  cudaMalloc((void**)&d_a, size);
  cudaMalloc((void**)&d_b, size);
  cudaMalloc((void**)&d_c, size);

  // copy memory to device
  cudaMemcpy(d_a, this->_array, size, cudaMemcpyHostToDevice);
  cudaMemcpy(d_b, a._array, size, cudaMemcpyHostToDevice);

  dotProduct<<<NUM_BLOCKS, NUM_THREADS_PER_BLOCK>>>(d_a, d_b, d_c, this->_size);

  // copy to host
  cudaMemcpy(result._array, d_c, size, cudaMemcpyDeviceToHost);

  return result;
}

/**
 * Constant Multiplication of vector in GPU
 *
 * Returns CudaVector object
 */
template <class T>
CudaVector<T> CudaVector<T>::operator*(const double& a) {
  // create result array
  CudaVector<T> result(this->_size);

  T *d_a, *d_c;
  const size_t size = this->_size * sizeof(T);

  // allocate memory
  cudaMalloc((void**)&d_a, size);
  cudaMalloc((void**)&d_c, size);

  // copy memory to device
  cudaMemcpy(d_a, this->_array, size, cudaMemcpyHostToDevice);

  constantMultiply<<<NUM_BLOCKS, NUM_THREADS_PER_BLOCK>>>(d_a, a, d_c,
                                                          this->_size);

  // copy to host
  cudaMemcpy(result._array, d_c, size, cudaMemcpyDeviceToHost);

  return result;
}

/**
 * Constant Division of vector in GPU
 *
 * Returns CudaVector object
 */
template <class T>
CudaVector<T> CudaVector<T>::operator/(const double& a) {
  // create result array
  CudaVector<T> result(this->_size);

  T *d_a, *d_c;
  const size_t size = this->_size * sizeof(T);

  // allocate memory
  cudaMalloc((void**)&d_a, size);
  cudaMalloc((void**)&d_c, size);

  // copy memory to device
  cudaMemcpy(d_a, this->_array, size, cudaMemcpyHostToDevice);

  contantDivide<<<NUM_BLOCKS, NUM_THREADS_PER_BLOCK>>>(d_a, a, d_c,
                                                       this->_size);

  // copy to host
  cudaMemcpy(result._array, d_c, size, cudaMemcpyDeviceToHost);

  return result;
}

/**
 * Subscript operator overloading
 *
 * Returns CudaVector object
 */
template <class T>
T& CudaVector<T>::operator[](unsigned long i) {
  if (i > this->_size || i < 0)
    throw CudaVectorException("Index out of bounds!");

  return this->_array[i];
}

template <class T>
unsigned long CudaVector<T>::size() {
  return this->_size;
}

/**
 * Exception Class definition
 */

template <class T>
CudaVector<T>::CudaVectorException::CudaVectorException(
    const std::string& message)
    : message_(message) {}