devtools/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2026-04-10 11:34:33 +08:00
Code Issues Packages Projects Releases Wiki Activity

mixing types step 3:

Browse Source 
- improve support of colmajor by vector and matrix - matrix
- now all configurations are well handled, but the perf are not always very good
This commit is contained in:
Gael Guennebaud
2010-07-11 23:57:23 +02:00
parent 8e3c4283f5
commit f8678272a4
6 changed files with 122 additions and 38 deletions
Download Patch File Download Diff File Expand all files Collapse all files

36
Eigen/src/Core/Product.h
View File

@@ -323,9 +323,10 @@ template<> struct ei_gemv_selector<OnTheRight,ColMajor,true>
static void run(const ProductType& prod, Dest& dest, typename ProductType::Scalar alpha)
{
typedef typename ProductType::Index Index;
typedef typename ProductType::LhsScalar LhsScalar;
typedef typename ProductType::RhsScalar RhsScalar;
typedef typename ProductType::Scalar ResScalar;
typedef typename ProductType::LhsScalar LhsScalar;
typedef typename ProductType::RhsScalar RhsScalar;
typedef typename ProductType::Scalar ResScalar;
typedef typename ProductType::RealScalar RealScalar;
typedef typename ProductType::ActualLhsType ActualLhsType;
typedef typename ProductType::ActualRhsType ActualRhsType;
typedef typename ProductType::LhsBlasTraits LhsBlasTraits;
@@ -340,16 +341,30 @@ template<> struct ei_gemv_selector<OnTheRight,ColMajor,true>
enum {
// FIXME find a way to allow an inner stride on the result if ei_packet_traits<Scalar>::size==1
EvalToDest = Dest::InnerStrideAtCompileTime==1
EvalToDestAtCompileTime = Dest::InnerStrideAtCompileTime==1,
ComplexByReal = (NumTraits<LhsScalar>::IsComplex) && (!NumTraits<RhsScalar>::IsComplex)
};
bool alphaIsCompatible = (!ComplexByReal) || (ei_imag(actualAlpha)==RealScalar(0));
bool evalToDest = EvalToDestAtCompileTime && alphaIsCompatible;
RhsScalar compatibleAlpha = ei_get_factor<ResScalar,RhsScalar>::run(actualAlpha);
ResScalar* actualDest;
if (EvalToDest)
if (evalToDest)
{
actualDest = &dest.coeffRef(0);
}
else
{
actualDest = ei_aligned_stack_new(ResScalar,dest.size());
MappedDest(actualDest, dest.size()) = dest;
if(!alphaIsCompatible)
{
MappedDest(actualDest, dest.size()).setZero();
compatibleAlpha = RhsScalar(1);
}
else
MappedDest(actualDest, dest.size()) = dest;
}
ei_general_matrix_vector_product
@@ -358,11 +373,14 @@ template<> struct ei_gemv_selector<OnTheRight,ColMajor,true>
&actualLhs.const_cast_derived().coeffRef(0,0), actualLhs.outerStride(),
actualRhs.data(), actualRhs.innerStride(),
actualDest, 1,
actualAlpha);
compatibleAlpha);
if (!EvalToDest)
if (!evalToDest)
{
dest = MappedDest(actualDest, dest.size());
if(!alphaIsCompatible)
dest += actualAlpha * MappedDest(actualDest, dest.size());
else
dest = MappedDest(actualDest, dest.size());
ei_aligned_stack_delete(ResScalar, actualDest, dest.size());
}
}