Prefer __builtin_prefetch over inline asm

Reorder the ifdefs for the PREFETCH macros so that the compiler builtin is
favored over the inline assembly for aarch64.
This commit is contained in:
Nick Terrell
2020-08-10 22:17:18 -07:00
parent 63eebc86cd
commit e3bda594ae
+3 -3
View File
@@ -125,12 +125,12 @@
# include <mmintrin.h> /* https://msdn.microsoft.com/fr-fr/library/84szxsww(v=vs.90).aspx */ # include <mmintrin.h> /* https://msdn.microsoft.com/fr-fr/library/84szxsww(v=vs.90).aspx */
# define PREFETCH_L1(ptr) _mm_prefetch((const char*)(ptr), _MM_HINT_T0) # define PREFETCH_L1(ptr) _mm_prefetch((const char*)(ptr), _MM_HINT_T0)
# define PREFETCH_L2(ptr) _mm_prefetch((const char*)(ptr), _MM_HINT_T1) # define PREFETCH_L2(ptr) _mm_prefetch((const char*)(ptr), _MM_HINT_T1)
# elif defined(__aarch64__)
# define PREFETCH_L1(ptr) __asm__ __volatile__("prfm pldl1keep, %0" ::"Q"(*(ptr)))
# define PREFETCH_L2(ptr) __asm__ __volatile__("prfm pldl2keep, %0" ::"Q"(*(ptr)))
# elif defined(__GNUC__) && ( (__GNUC__ >= 4) || ( (__GNUC__ == 3) && (__GNUC_MINOR__ >= 1) ) ) # elif defined(__GNUC__) && ( (__GNUC__ >= 4) || ( (__GNUC__ == 3) && (__GNUC_MINOR__ >= 1) ) )
# define PREFETCH_L1(ptr) __builtin_prefetch((ptr), 0 /* rw==read */, 3 /* locality */) # define PREFETCH_L1(ptr) __builtin_prefetch((ptr), 0 /* rw==read */, 3 /* locality */)
# define PREFETCH_L2(ptr) __builtin_prefetch((ptr), 0 /* rw==read */, 2 /* locality */) # define PREFETCH_L2(ptr) __builtin_prefetch((ptr), 0 /* rw==read */, 2 /* locality */)
# elif defined(__aarch64__)
# define PREFETCH_L1(ptr) __asm__ __volatile__("prfm pldl1keep, %0" ::"Q"(*(ptr)))
# define PREFETCH_L2(ptr) __asm__ __volatile__("prfm pldl2keep, %0" ::"Q"(*(ptr)))
# else # else
# define PREFETCH_L1(ptr) (void)(ptr) /* disabled */ # define PREFETCH_L1(ptr) (void)(ptr) /* disabled */
# define PREFETCH_L2(ptr) (void)(ptr) /* disabled */ # define PREFETCH_L2(ptr) (void)(ptr) /* disabled */