|
numerics
|
OpenMP backend – parallelised matrix operations. More...
Go to the source code of this file.
Namespaces | |
| namespace | num |
| namespace | num::backends |
| namespace | num::backends::omp |
Functions | |
| void | num::backends::omp::matmul (const Matrix &A, const Matrix &B, Matrix &C) |
| void | num::backends::omp::matvec (const Matrix &A, const Vector &x, Vector &y) |
| void | num::backends::omp::matadd (real alpha, const Matrix &A, real beta, const Matrix &B, Matrix &C) |
OpenMP backend – parallelised matrix operations.
The (ii, jj) cache-tile pairs are distributed across threads. Each thread owns a distinct C tile so there are no data races. Falls back to sequential when NUMERICS_HAS_OMP is not defined.
Definition in file matrix.cpp.