GitHub-Proxy/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-04-19 16:19:37 +08:00
Code Issues Packages Projects Releases Wiki Activity

Optimize division operations in TensorVolumePatch.h

Browse Source
This commit is contained in:
Gopinath Vasalamarri 2025-02-28 22:34:13 +00:00 committed by Rasmus Munk Larsen
parent be5147b090
commit 9a86214039
1 changed files with 5 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
unsupported/Eigen/CXX11/src/Tensor/TensorVolumePatch.h
View File

@ -448,8 +448,11 @@ struct TensorEvaluator<const TensorVolumePatchOp<Planes, Rows, Cols, ArgType>, D
eigen_assert(otherIndex == indices[1] / m_fastOtherStride); eigen_assert(otherIndex == indices[1] / m_fastOtherStride);
// Find the offset of the element wrt the location of the first element. // Find the offset of the element wrt the location of the first element.
const Index patchOffsets[2] = {(indices[0] - patchIndex * m_patchStride) / m_fastOutputDepth, Index first_entry = (indices[0] - patchIndex * m_patchStride) / m_fastOutputDepth;
(indices[1] - patchIndex * m_patchStride) / m_fastOutputDepth}; Index second_entry = PacketSize == 1 ? first_entry :
(indices[1] - patchIndex * m_patchStride) / m_fastOutputDepth;
const Index patchOffsets[2] = {first_entry, second_entry};
const Index patch3DIndex = const Index patch3DIndex =
(NumDims == 5) ? patchIndex : (indices[0] - otherIndex * m_otherStride) / m_fastPatchStride; (NumDims == 5) ? patchIndex : (indices[0] - otherIndex * m_otherStride) / m_fastPatchStride;