mirror/eigen
2
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-03-07 18:27:40 +08:00
Code Issues Packages Projects Releases Wiki Activity

bug #936, patch 1/3: some cleanup and renaming for consistency.

Browse Source
This commit is contained in:
Benoit Jacob 2015-01-30 17:43:56 -05:00
parent 759bd92a85
commit 9f99f61e69
3 changed files with 13 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
Eigen/src/Core/arch/AltiVec/PacketMath.h
View File

@ -22,8 +22,8 @@ namespace internal {
#define EIGEN_HAS_FUSED_MADD 1
#endif
#ifndef EIGEN_HAS_FUSE_CJMADD
#define EIGEN_HAS_FUSE_CJMADD 1
#ifndef EIGEN_HAS_FUSED_CJMADD
#define EIGEN_HAS_FUSED_CJMADD
#endif
// NOTE Altivec has 32 registers, but Eigen only accepts a value of 8 or 16

4
Eigen/src/Core/arch/NEON/PacketMath.h
View File

@ -24,8 +24,8 @@ namespace internal {
#define EIGEN_HAS_FUSED_MADD 1
#endif
#ifndef EIGEN_HAS_FUSE_CJMADD
#define EIGEN_HAS_FUSE_CJMADD 1
#ifndef EIGEN_HAS_FUSED_CJMADD
#define EIGEN_HAS_FUSED_CJMADD
#endif
// FIXME NEON has 16 quad registers, but since the current register allocator

18
Eigen/src/Core/products/GeneralBlockPanelKernel.h
View File

@ -120,8 +120,8 @@ inline void computeProductBlockingSizes(SizeType& k, SizeType& m, SizeType& n)
computeProductBlockingSizes<LhsScalar,RhsScalar,1>(k, m, n);
}
#ifdef EIGEN_HAS_FUSE_CJMADD
#define MADD(CJ,A,B,C,T) C = CJ.pmadd(A,B,C);
#ifdef EIGEN_HAS_FUSED_CJMADD
#define CJMADD(CJ,A,B,C,T) C = CJ.pmadd(A,B,C);
#else
// FIXME (a bit overkill maybe ?)
@ -146,8 +146,8 @@ inline void computeProductBlockingSizes(SizeType& k, SizeType& m, SizeType& n)
gebp_madd_selector<CJ,A,B,C,T>::run(cj,a,b,c,t);
}
#define MADD(CJ,A,B,C,T) gebp_madd(CJ,A,B,C,T);
// #define MADD(CJ,A,B,C,T) T = B; T = CJ.pmul(A,T); C = padd(C,T);
#define CJMADD(CJ,A,B,C,T) gebp_madd(CJ,A,B,C,T);
// #define CJMADD(CJ,A,B,C,T) T = B; T = CJ.pmul(A,T); C = padd(C,T);
#endif
/* Vectorization logic
@ -1402,13 +1402,13 @@ void gebp_kernel<LhsScalar,RhsScalar,Index,mr,nr,ConjugateLhs,ConjugateRhs>
B_0 = blB[0];
B_1 = blB[1];
MADD(cj,A0,B_0,C0, B_0);
MADD(cj,A0,B_1,C1, B_1);
CJMADD(cj,A0,B_0,C0, B_0);
CJMADD(cj,A0,B_1,C1, B_1);
B_0 = blB[2];
B_1 = blB[3];
MADD(cj,A0,B_0,C2, B_0);
MADD(cj,A0,B_1,C3, B_1);
CJMADD(cj,A0,B_0,C2, B_0);
CJMADD(cj,A0,B_1,C3, B_1);
blB += 4;
}
@ -1434,7 +1434,7 @@ void gebp_kernel<LhsScalar,RhsScalar,Index,mr,nr,ConjugateLhs,ConjugateRhs>
{
LhsScalar A0 = blA[k];
RhsScalar B_0 = blB[k];
MADD(cj, A0, B_0, C0, B_0);
CJMADD(cj, A0, B_0, C0, B_0);
}
res[(j2+0)*resStride + i] += alpha*C0;
}