GitHub-Proxy/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-08-12 03:39:01 +08:00
Code Issues Packages Projects Releases Wiki Activity

GCC 4.8 arm EIGEN_OPTIMIZATION_BARRIER fix (#2315).

Browse Source 
GCC 4.8 doesn't seem to like the `g` register constraint, failing to
compile with "error: 'asm' operand requires impossible reload".

Tested `r` instead, and that seems to work, even with latest compilers.

Also fixed some minor macro issues to eliminate warnings on armv7.

Fixes #2315.
This commit is contained in:
Antonio Sanchez 2021-08-31 12:01:53 -07:00 committed by Rasmus Munk Larsen
parent cc3573ab44
commit ff07a8a639
2 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
Eigen/src/Core/arch/NEON/PacketMath.h
View File

@ -155,7 +155,7 @@ EIGEN_STRONG_INLINE Packet4f vec4f_unpackhi(const Packet4f& a, const Packet4f& b
#define EIGEN_ARM_PREFETCH(ADDR) __builtin_prefetch(ADDR);
#elif defined __pld
#define EIGEN_ARM_PREFETCH(ADDR) __pld(ADDR)
#elif EIGEN_ARCH_ARM32
#elif EIGEN_ARCH_ARM
#define EIGEN_ARM_PREFETCH(ADDR) __asm__ __volatile__ ("pld [%[addr]]\n" :: [addr] "r" (ADDR) : );
#else
// by default no explicit prefetching
@ -3985,8 +3985,6 @@ template<> EIGEN_STRONG_INLINE Packet2d prsqrt(const Packet2d& a) {
template<> EIGEN_STRONG_INLINE Packet2d psqrt(const Packet2d& _x){ return vsqrtq_f64(_x); }
#endif // EIGEN_ARCH_ARM64
// Do we have an fp16 types and supporting Neon intrinsics?
#if EIGEN_HAS_ARM64_FP16_VECTOR_ARITHMETIC
typedef float16x4_t Packet4hf;
@ -4625,6 +4623,8 @@ EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE void ptranspose(PacketBlock<Packet8hf, 8>&
}
#endif // end EIGEN_HAS_ARM64_FP16_VECTOR_ARITHMETIC
#endif // EIGEN_ARCH_ARM64
} // end namespace internal
} // end namespace Eigen

2
Eigen/src/Core/util/Macros.h
View File

@ -1131,7 +1131,7 @@ namespace Eigen {
#define EIGEN_OPTIMIZATION_BARRIER(X) __asm__ ("" : "+r,v,wa" (X));
#elif EIGEN_ARCH_ARM_OR_ARM64
// General, NEON.
#define EIGEN_OPTIMIZATION_BARRIER(X) __asm__ ("" : "+g,w" (X));
#define EIGEN_OPTIMIZATION_BARRIER(X) __asm__ ("" : "+r,w" (X));
#elif EIGEN_ARCH_i386_OR_x86_64
// General, SSE.
#define EIGEN_OPTIMIZATION_BARRIER(X) __asm__ ("" : "+g,x" (X));