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

Fix excessive GEBP register spilling for 32-bit NEON.

Browse Source 
Clang does a poor job of optimizing the GEBP microkernel on 32-bit ARM,
leading to excessive 16-byte register spills, slowing down basic f32
matrix multiplication by approx 50%.

By specializing `gebp_traits`, we can eliminate the register spills.
Volatile inline ASM both acts as a barrier to prevent reordering and
enforces strict register use. In a simple f32 matrix multiply example,
this modification reduces 16-byte spills from 109 instances to zero,
leading to a 1.5x speed increase (search for `16-byte Spill` in the
assembly in https://godbolt.org/z/chsPbE).

This is a replacement of !379.  See there for further discussion.

Also moved `gebp_traits` specializations for NEON to
`Eigen/src/Core/arch/NEON/GeneralBlockPanelKernel.h` to be alongside
other NEON-specific code.

Fixes #2138.
This commit is contained in:
Antonio Sanchez
2021-02-03 08:18:28 -08:00
parent 56c8b14d87
commit f85038b7f3
3 changed files with 186 additions and 143 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
Eigen/Core
View File

@@ -340,7 +340,9 @@ using std::ptrdiff_t;
#include "src/Core/ConditionEstimator.h"
#if defined(EIGEN_VECTORIZE_ALTIVEC) || defined(EIGEN_VECTORIZE_VSX)
#include "src/Core/arch/AltiVec/MatrixProduct.h"
#include "src/Core/arch/AltiVec/MatrixProduct.h"
#elif defined EIGEN_VECTORIZE_NEON
#include "src/Core/arch/NEON/GeneralBlockPanelKernel.h"
#endif
#include "src/Core/BooleanRedux.h"