GitHub-Proxy/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-04-23 10:09:36 +08:00
Code Issues Packages Projects Releases Wiki Activity

update to support mixin types

Browse Source
This commit is contained in:
Gael Guennebaud 2010-07-07 19:49:48 +02:00
parent 31a36aa9c4
commit 02fd3acd81
1 changed files with 11 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
bench/bench_gemm.cpp
View File

@ -15,7 +15,10 @@ using namespace Eigen;
#endif #endif
typedef SCALAR Scalar; typedef SCALAR Scalar;
typedef Matrix<Scalar,Dynamic,Dynamic> M; typedef NumTraits<Scalar>::Real RealScalar;
typedef Matrix<RealScalar,Dynamic,Dynamic> A;
typedef Matrix<Scalar,Dynamic,Dynamic> B;
typedef Matrix<Scalar,Dynamic,Dynamic> C;
#ifdef HAVE_BLAS #ifdef HAVE_BLAS
@ -84,8 +87,8 @@ void blas_gemm(const MatrixXd& a, const MatrixXd& b, MatrixXd& c)
#endif #endif
template<typename M> template<typename A, typename B, typename C>
EIGEN_DONT_INLINE void gemm(const M& a, const M& b, M& c) EIGEN_DONT_INLINE void gemm(const A& a, const B& b, C& c)
{ {
c.noalias() += a * b; c.noalias() += a * b;
} }
@ -96,7 +99,7 @@ int main(int argc, char ** argv)
std::ptrdiff_t l2 = ei_queryTopLevelCacheSize(); std::ptrdiff_t l2 = ei_queryTopLevelCacheSize();
std::cout << "L1 cache size = " << (l1>0 ? l1/1024 : -1) << " KB\n"; std::cout << "L1 cache size = " << (l1>0 ? l1/1024 : -1) << " KB\n";
std::cout << "L2/L3 cache size = " << (l2>0 ? l2/1024 : -1) << " KB\n"; std::cout << "L2/L3 cache size = " << (l2>0 ? l2/1024 : -1) << " KB\n";
typedef ei_product_blocking_traits<Scalar> Blocking; typedef ei_product_blocking_traits<Scalar,Scalar> Blocking;
std::cout << "Register blocking = " << Blocking::mr << " x " << Blocking::nr << "\n"; std::cout << "Register blocking = " << Blocking::mr << " x " << Blocking::nr << "\n";
int rep = 1; // number of repetitions per try int rep = 1; // number of repetitions per try
@ -132,16 +135,16 @@ int main(int argc, char ** argv)
int m = s; int m = s;
int n = s; int n = s;
int p = s; int p = s;
M a(m,n); a.setRandom(); A a(m,n); a.setRandom();
M b(n,p); b.setRandom(); B b(n,p); b.setRandom();
M c(m,p); c.setOnes(); C c(m,p); c.setOnes();
std::cout << "Matrix sizes = " << m << "x" << p << " * " << p << "x" << n << "\n"; std::cout << "Matrix sizes = " << m << "x" << p << " * " << p << "x" << n << "\n";
std::ptrdiff_t cm(m), cn(n), ck(p); std::ptrdiff_t cm(m), cn(n), ck(p);
computeProductBlockingSizes<Scalar,Scalar>(ck, cm, cn); computeProductBlockingSizes<Scalar,Scalar>(ck, cm, cn);
std::cout << "blocking size = " << cm << " x " << ck << "\n"; std::cout << "blocking size = " << cm << " x " << ck << "\n";
M r = c; C r = c;
// check the parallel product is correct // check the parallel product is correct
#ifdef EIGEN_HAS_OPENMP #ifdef EIGEN_HAS_OPENMP