#include <iostream>
static size_t dimension = 4 * 1024;
static const int maxIter = 10;
static const int sparsityFactor = 7;
void setupInputs() {
A = A * ((A % sparsityFactor) == 0) / 1000;
std::cout << "Sparsity of A = "
<< 100.f * (float)
sparseGetNNZ(spA) / (float)spA.elements() <<
"%"
<< std::endl;
std::cout << "Memory Usage of A = " << A.bytes() / (1024.f * 1024.f)
<< " MB" << std::endl;
std::cout << "Memory Usage of spA = "
(1024.f * 1024.f)
<< " MB" << std::endl;
}
void sparseConjugateGradient(void) {
for (int i = 0; i < maxIter; ++i) {
array alpha = alpha_num / alpha_den;
r -=
tile(alpha, Ap.dims()) * Ap;
x +=
tile(alpha, Ap.dims()) * p;
array beta = beta_num / alpha_num;
p = r +
tile(beta, p.dims()) * p;
}
}
void denseConjugateGradient(void) {
for (int i = 0; i < maxIter; ++i) {
array alpha = alpha_num / alpha_den;
r -=
tile(alpha, Ap.dims()) * Ap;
x +=
tile(alpha, Ap.dims()) * p;
array beta = beta_num / alpha_num;
p = r +
tile(beta, p.dims()) * p;
}
}
void checkConjugateGradient(
const af::array in) {
for (int i = 0; i < maxIter; ++i) {
array alpha = alpha_num / alpha_den;
r -=
tile(alpha, Ap.dims()) * Ap;
x +=
tile(alpha, Ap.dims()) * p;
array beta = beta_num / alpha_num;
p = r +
tile(beta, p.dims()) * p;
}
std::cout << "Final difference in solutions:\n";
}
int main(int, char **) {
setupInputs();
std::cout << "Verifying Dense Conjugate Gradient:" << std::endl;
checkConjugateGradient(A);
std::cout << "Verifying Sparse Conjugate Gradient:" << std::endl;
checkConjugateGradient(spA);
std::cout << "Dense Conjugate Gradient Time: "
<<
timeit(denseConjugateGradient) * 1000 <<
"ms" << std::endl;
std::cout << "Sparse Conjugate Gradient Time: "
<<
timeit(sparseConjugateGradient) * 1000 <<
"ms" << std::endl;
return 0;
}
A multi dimensional data container.
dim4 dims() const
Get dimensions of the array.
size_t bytes() const
Get the size of the array in bytes.
@ f32
32-bit floating point values
AFAPI array floor(const array &in)
C++ Interface for flooring an array of numbers.
T dot(const array &lhs, const array &rhs, const matProp optLhs=AF_MAT_NONE, const matProp optRhs=AF_MAT_NONE)
Dot Product.
AFAPI array matmul(const array &lhs, const array &rhs, const matProp optLhs=AF_MAT_NONE, const matProp optRhs=AF_MAT_NONE)
Matrix multiply of two arrays.
AFAPI array transpose(const array &in, const bool conjugate=false)
Transposes a matrix.
array constant(T val, const dim4 &dims, const dtype ty=(af_dtype) dtype_traits< T >::ctype)
AFAPI array identity(const dim4 &dims, const dtype ty=f32)
AFAPI void sync(const int device=-1)
Blocks until the device is finished processing.
AFAPI array tile(const array &in, const unsigned x, const unsigned y=1, const unsigned z=1, const unsigned w=1)
static af::array array(af::dim4 idims, cl_mem buf, af::dtype type, bool retain=false)
Create an af::array object from an OpenCL cl_mem buffer.
AFAPI array randu(const dim4 &dims, const dtype ty, randomEngine &r)
AFAPI array sparseGetColIdx(const array in)
AFAPI array sparse(const dim_t nRows, const dim_t nCols, const array values, const array rowIdx, const array colIdx, const af::storage stype=AF_STORAGE_CSR)
This function converts af::array of values, row indices and column indices into a sparse array.
AFAPI dim_t sparseGetNNZ(const array in)
AFAPI array sparseGetRowIdx(const array in)
AFAPI array sparseGetValues(const array in)
AFAPI double timeit(void(*fn)())