void mpkernel(int offset, int stride, int n, int sz,
double *dev_x, double *dev_y,
bool reset, bool print)
{
typedef typename vector_type::storage_type storage_type;
// multi-point expansions
vector_type x(sz, 0.0), y(sz, 0.0);
// Loop over elements, propagating sz samples simultaneously
for (int e=0; e<n; e++) {
// Initialize x
if (reset && storage_type::supports_reset) {
storage_type& x_s = x.storage();
storage_type& y_s = y.storage();
x_s.shallowReset(dev_x+offset, sz, stride, false);
y_s.shallowReset(dev_y+offset, sz, stride, false);
}
else {
for (int i=0; i<sz; i++)
x.fastAccessCoeff(i) = dev_x[offset+i*stride];
}
simple_function(x,y);
// Print x and y
if (print) {
std::cout << "x(0) = [ ";
for (int i=0; i<sz; i++)
std::cout << x.coeff(i) << " ";
std::cout << "]" << std::endl << std::endl;
std::cout << "y(0) = [ ";
for (int i=0; i<sz; i++)
std::cout << y.coeff(i) << " ";
std::cout << "]" << std::endl << std::endl;
}
// Return result
if (!(reset && vector_type::storage_type::supports_reset)) {
for (int i=0; i<sz; i++)
dev_y[offset+i*stride] = y.fastAccessCoeff(i);
}
offset += stride*sz;
}
}
void kernel(int offset, int stride, int n, int sz, double *dev_x, double *dev_y)
{
// Loop over elements
double x, y;
for (int e=0; e<n; e++) {
for (int i=0; i<sz; i++) {
// Initialize x
x = dev_x[offset+i*stride];
// Compute function
simple_function(x,y);
// Return result
dev_y[offset+i*stride] = y;
}
offset += stride*sz;
}
}
int main(int argc, char **argv)
{
try {
// Polynomial expansions
const int sz = 10;
vec_type u(sz, 0.0);
for (int i=0; i<sz; i++) {
u.fastAccessCoeff(i) = 0.1 * i;
}
// Compute PCE expansion of function
vec_type v = simple_function(u);
// Print u and v
std::cout << "v = 1.0 / (log(u)^2 + 1):" << std::endl;
std::cout << "\tu = " << u << std::endl;
std::cout << "\tv = " << v << std::endl;
}
catch (std::exception& e) {
std::cout << e.what() << std::endl;
}
}
int main() {
simple_function();
}
