lisa.pizzo/LisaPizzoExercises
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Delete Sheet5/bench_funcs.cpp

Browse source
This commit is contained in:
Lisa Pizzo 2025-12-04 12:06:27 +01:00
commit c363644922
1 changed files with 0 additions and 103 deletions
Download patch file Download diff file Expand all files Collapse all files

103
Sheet5/bench_funcs.cpp
View file

@ -1,103 +0,0 @@
// bench_funcs.cpp
#include "bench_funcs.h"
#include <omp.h>
#include <cmath>
#include <cstddef>
#include <vector>
#include <algorithm>
// A: parallel sum
double sum_basic(const std::vector<double>& x)
{
double sum = 0.0;
#pragma omp parallel for reduction(+:sum)
for (std::size_t i = 0; i < x.size(); ++i) sum += x[i];
return sum;
}
// A: parallel dot product
double dot_basic(const std::vector<double>& x, const std::vector<double>& y)
{
std::size_t N = x.size();
double sum = 0.0;
#pragma omp parallel for reduction(+:sum)
for (std::size_t i = 0; i < N; ++i) sum += x[i] * y[i];
return sum;
}
// Kahan remains same
double dot_kahan(const std::vector<double>& x, const std::vector<double>& y)
{
double sum = 0.0;
double c = 0.0;
for (std::size_t i = 0; i < x.size(); ++i)
{
double prod = x[i] * y[i];
double yk = prod - c;
double t = sum + yk;
c = (t - sum) - yk;
sum = t;
}
return sum;
}
// Norm (sum of squares)
double norm_basic(const std::vector<double>& x)
{
double sumsq = 0.0;
#pragma omp parallel for reduction(+:sumsq)
for (std::size_t i = 0; i < x.size(); ++i) sumsq += x[i]*x[i];
return sumsq;
}
// B: matvec (row-major)
void matvec_rowmajor(const std::vector<double>& A, std::size_t M, std::size_t N,
const std::vector<double>& x, std::vector<double>& b)
{
b.assign(M, 0.0);
#pragma omp parallel for
for (std::size_t i = 0; i < M; ++i) {
double tmp = 0.0;
const double* Ai = &A[i*N];
for (std::size_t j = 0; j < N; ++j) tmp += Ai[j] * x[j];
b[i] = tmp;
}
}
// C: matmul (row-major)
void matmul_rowmajor(const std::vector<double>& A, std::size_t M, std::size_t L,
const std::vector<double>& B, std::size_t N,
std::vector<double>& C)
{
C.assign(M * N, 0.0);
// Parallelize over output rows (i)
#pragma omp parallel for
for (std::size_t i = 0; i < M; ++i) {
for (std::size_t k = 0; k < L; ++k) {
double Aik = A[i*L + k];
const double* Bk = &B[k*N];
double* Ci = &C[i*N];
for (std::size_t j = 0; j < N; ++j) {
Ci[j] += Aik * Bk[j];
}
}
}
}
// D: polynomial evaluation (Horner), parallel over x points
void polyp_horner(const std::vector<double>& a, const std::vector<double>& x, std::vector<double>& y)
{
std::size_t p = a.size() - 1;
std::size_t N = x.size();
y.assign(N, 0.0);
#pragma omp parallel for
for (std::size_t i = 0; i < N; ++i) {
double xi = x[i];
double val = a[p];
for (std::size_t k = p; k-- > 0; ) {
val = val * xi + a[k];
}
y[i] = val;
}
}