public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc r12-5169] aarch64: Use type-qualified builtins for [R]ADDHN[2] Neon intrinsics
@ 2021-11-11 15:35 Jonathan Wright
0 siblings, 0 replies; only message in thread
From: Jonathan Wright @ 2021-11-11 15:35 UTC (permalink / raw)
To: gcc-cvs
https://gcc.gnu.org/g:7bde2a6ecd7c7ce7930fe048ae43a5d177d91b39
commit r12-5169-g7bde2a6ecd7c7ce7930fe048ae43a5d177d91b39
Author: Jonathan Wright <jonathan.wright@arm.com>
Date: Tue Nov 9 12:16:01 2021 +0000
aarch64: Use type-qualified builtins for [R]ADDHN[2] Neon intrinsics
Declare unsigned type-qualified builtins and use them to implement
(rounding) halving-narrowing-add Neon intrinsics. This removes the
need for many casts in arm_neon.h.
gcc/ChangeLog:
2021-11-09 Jonathan Wright <jonathan.wright@arm.com>
* config/aarch64/aarch64-simd-builtins.def: Declare unsigned
builtins for [r]addhn[2].
* config/aarch64/arm_neon.h (vaddhn_s16): Remove unnecessary
cast.
(vaddhn_s32): Likewise.
(vaddhn_s64): Likewise.
(vaddhn_u16): Use type-qualified builtin and remove casts.
(vaddhn_u32): Likewise.
(vaddhn_u64): Likewise.
(vraddhn_s16): Remove unnecessary cast.
(vraddhn_s32): Likewise.
(vraddhn_s64): Likewise.
(vraddhn_u16): Use type-qualified builtin and remove casts.
(vraddhn_u32): Likewise.
(vraddhn_u64): Likewise.
(vaddhn_high_s16): Remove unnecessary cast.
(vaddhn_high_s32): Likewise.
(vaddhn_high_s64): Likewise.
(vaddhn_high_u16): Use type-qualified builtin and remove
casts.
(vaddhn_high_u32): Likewise.
(vaddhn_high_u64): Likewise.
(vraddhn_high_s16): Remove unnecessary cast.
(vraddhn_high_s32): Likewise.
(vraddhn_high_s64): Likewise.
(vraddhn_high_u16): Use type-qualified builtin and remove
casts.
(vraddhn_high_u32): Likewise.
(vraddhn_high_u64): Likewise.
Diff:
---
gcc/config/aarch64/aarch64-simd-builtins.def | 4 ++
gcc/config/aarch64/arm_neon.h | 66 ++++++++++------------------
2 files changed, 28 insertions(+), 42 deletions(-)
diff --git a/gcc/config/aarch64/aarch64-simd-builtins.def b/gcc/config/aarch64/aarch64-simd-builtins.def
index 0d5d22e11e2..e57dee9456f 100644
--- a/gcc/config/aarch64/aarch64-simd-builtins.def
+++ b/gcc/config/aarch64/aarch64-simd-builtins.def
@@ -220,13 +220,17 @@
/* Implemented by aarch64_<sur><addsub>hn<mode>. */
BUILTIN_VQN (BINOP, addhn, 0, NONE)
+ BUILTIN_VQN (BINOPU, addhn, 0, NONE)
BUILTIN_VQN (BINOP, subhn, 0, NONE)
BUILTIN_VQN (BINOP, raddhn, 0, NONE)
+ BUILTIN_VQN (BINOPU, raddhn, 0, NONE)
BUILTIN_VQN (BINOP, rsubhn, 0, NONE)
/* Implemented by aarch64_<sur><addsub>hn2<mode>. */
BUILTIN_VQN (TERNOP, addhn2, 0, NONE)
+ BUILTIN_VQN (TERNOPU, addhn2, 0, NONE)
BUILTIN_VQN (TERNOP, subhn2, 0, NONE)
BUILTIN_VQN (TERNOP, raddhn2, 0, NONE)
+ BUILTIN_VQN (TERNOPU, raddhn2, 0, NONE)
BUILTIN_VQN (TERNOP, rsubhn2, 0, NONE)
/* Implemented by aarch64_<us>xtl<mode>. */
diff --git a/gcc/config/aarch64/arm_neon.h b/gcc/config/aarch64/arm_neon.h
index 05af357f9d5..fa404e0c640 100644
--- a/gcc/config/aarch64/arm_neon.h
+++ b/gcc/config/aarch64/arm_neon.h
@@ -713,186 +713,168 @@ __extension__ extern __inline int8x8_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vaddhn_s16 (int16x8_t __a, int16x8_t __b)
{
- return (int8x8_t) __builtin_aarch64_addhnv8hi (__a, __b);
+ return __builtin_aarch64_addhnv8hi (__a, __b);
}
__extension__ extern __inline int16x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vaddhn_s32 (int32x4_t __a, int32x4_t __b)
{
- return (int16x4_t) __builtin_aarch64_addhnv4si (__a, __b);
+ return __builtin_aarch64_addhnv4si (__a, __b);
}
__extension__ extern __inline int32x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vaddhn_s64 (int64x2_t __a, int64x2_t __b)
{
- return (int32x2_t) __builtin_aarch64_addhnv2di (__a, __b);
+ return __builtin_aarch64_addhnv2di (__a, __b);
}
__extension__ extern __inline uint8x8_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vaddhn_u16 (uint16x8_t __a, uint16x8_t __b)
{
- return (uint8x8_t) __builtin_aarch64_addhnv8hi ((int16x8_t) __a,
- (int16x8_t) __b);
+ return __builtin_aarch64_addhnv8hi_uuu (__a, __b);
}
__extension__ extern __inline uint16x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vaddhn_u32 (uint32x4_t __a, uint32x4_t __b)
{
- return (uint16x4_t) __builtin_aarch64_addhnv4si ((int32x4_t) __a,
- (int32x4_t) __b);
+ return __builtin_aarch64_addhnv4si_uuu (__a, __b);
}
__extension__ extern __inline uint32x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vaddhn_u64 (uint64x2_t __a, uint64x2_t __b)
{
- return (uint32x2_t) __builtin_aarch64_addhnv2di ((int64x2_t) __a,
- (int64x2_t) __b);
+ return __builtin_aarch64_addhnv2di_uuu (__a, __b);
}
__extension__ extern __inline int8x8_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vraddhn_s16 (int16x8_t __a, int16x8_t __b)
{
- return (int8x8_t) __builtin_aarch64_raddhnv8hi (__a, __b);
+ return __builtin_aarch64_raddhnv8hi (__a, __b);
}
__extension__ extern __inline int16x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vraddhn_s32 (int32x4_t __a, int32x4_t __b)
{
- return (int16x4_t) __builtin_aarch64_raddhnv4si (__a, __b);
+ return __builtin_aarch64_raddhnv4si (__a, __b);
}
__extension__ extern __inline int32x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vraddhn_s64 (int64x2_t __a, int64x2_t __b)
{
- return (int32x2_t) __builtin_aarch64_raddhnv2di (__a, __b);
+ return __builtin_aarch64_raddhnv2di (__a, __b);
}
__extension__ extern __inline uint8x8_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vraddhn_u16 (uint16x8_t __a, uint16x8_t __b)
{
- return (uint8x8_t) __builtin_aarch64_raddhnv8hi ((int16x8_t) __a,
- (int16x8_t) __b);
+ return __builtin_aarch64_raddhnv8hi_uuu (__a, __b);
}
__extension__ extern __inline uint16x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vraddhn_u32 (uint32x4_t __a, uint32x4_t __b)
{
- return (uint16x4_t) __builtin_aarch64_raddhnv4si ((int32x4_t) __a,
- (int32x4_t) __b);
+ return __builtin_aarch64_raddhnv4si_uuu (__a, __b);
}
__extension__ extern __inline uint32x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vraddhn_u64 (uint64x2_t __a, uint64x2_t __b)
{
- return (uint32x2_t) __builtin_aarch64_raddhnv2di ((int64x2_t) __a,
- (int64x2_t) __b);
+ return __builtin_aarch64_raddhnv2di_uuu (__a, __b);
}
__extension__ extern __inline int8x16_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vaddhn_high_s16 (int8x8_t __a, int16x8_t __b, int16x8_t __c)
{
- return (int8x16_t) __builtin_aarch64_addhn2v8hi (__a, __b, __c);
+ return __builtin_aarch64_addhn2v8hi (__a, __b, __c);
}
__extension__ extern __inline int16x8_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vaddhn_high_s32 (int16x4_t __a, int32x4_t __b, int32x4_t __c)
{
- return (int16x8_t) __builtin_aarch64_addhn2v4si (__a, __b, __c);
+ return __builtin_aarch64_addhn2v4si (__a, __b, __c);
}
__extension__ extern __inline int32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vaddhn_high_s64 (int32x2_t __a, int64x2_t __b, int64x2_t __c)
{
- return (int32x4_t) __builtin_aarch64_addhn2v2di (__a, __b, __c);
+ return __builtin_aarch64_addhn2v2di (__a, __b, __c);
}
__extension__ extern __inline uint8x16_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vaddhn_high_u16 (uint8x8_t __a, uint16x8_t __b, uint16x8_t __c)
{
- return (uint8x16_t) __builtin_aarch64_addhn2v8hi ((int8x8_t) __a,
- (int16x8_t) __b,
- (int16x8_t) __c);
+ return __builtin_aarch64_addhn2v8hi_uuuu (__a, __b, __c);
}
__extension__ extern __inline uint16x8_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vaddhn_high_u32 (uint16x4_t __a, uint32x4_t __b, uint32x4_t __c)
{
- return (uint16x8_t) __builtin_aarch64_addhn2v4si ((int16x4_t) __a,
- (int32x4_t) __b,
- (int32x4_t) __c);
+ return __builtin_aarch64_addhn2v4si_uuuu (__a, __b, __c);
}
__extension__ extern __inline uint32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vaddhn_high_u64 (uint32x2_t __a, uint64x2_t __b, uint64x2_t __c)
{
- return (uint32x4_t) __builtin_aarch64_addhn2v2di ((int32x2_t) __a,
- (int64x2_t) __b,
- (int64x2_t) __c);
+ return __builtin_aarch64_addhn2v2di_uuuu (__a, __b, __c);
}
__extension__ extern __inline int8x16_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vraddhn_high_s16 (int8x8_t __a, int16x8_t __b, int16x8_t __c)
{
- return (int8x16_t) __builtin_aarch64_raddhn2v8hi (__a, __b, __c);
+ return __builtin_aarch64_raddhn2v8hi (__a, __b, __c);
}
__extension__ extern __inline int16x8_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vraddhn_high_s32 (int16x4_t __a, int32x4_t __b, int32x4_t __c)
{
- return (int16x8_t) __builtin_aarch64_raddhn2v4si (__a, __b, __c);
+ return __builtin_aarch64_raddhn2v4si (__a, __b, __c);
}
__extension__ extern __inline int32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vraddhn_high_s64 (int32x2_t __a, int64x2_t __b, int64x2_t __c)
{
- return (int32x4_t) __builtin_aarch64_raddhn2v2di (__a, __b, __c);
+ return __builtin_aarch64_raddhn2v2di (__a, __b, __c);
}
__extension__ extern __inline uint8x16_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vraddhn_high_u16 (uint8x8_t __a, uint16x8_t __b, uint16x8_t __c)
{
- return (uint8x16_t) __builtin_aarch64_raddhn2v8hi ((int8x8_t) __a,
- (int16x8_t) __b,
- (int16x8_t) __c);
+ return __builtin_aarch64_raddhn2v8hi_uuuu (__a, __b, __c);
}
__extension__ extern __inline uint16x8_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vraddhn_high_u32 (uint16x4_t __a, uint32x4_t __b, uint32x4_t __c)
{
- return (uint16x8_t) __builtin_aarch64_raddhn2v4si ((int16x4_t) __a,
- (int32x4_t) __b,
- (int32x4_t) __c);
+ return __builtin_aarch64_raddhn2v4si_uuuu (__a, __b, __c);
}
__extension__ extern __inline uint32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
vraddhn_high_u64 (uint32x2_t __a, uint64x2_t __b, uint64x2_t __c)
{
- return (uint32x4_t) __builtin_aarch64_raddhn2v2di ((int32x2_t) __a,
- (int64x2_t) __b,
- (int64x2_t) __c);
+ return __builtin_aarch64_raddhn2v2di_uuuu (__a, __b, __c);
}
__extension__ extern __inline float32x2_t
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2021-11-11 15:35 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-11-11 15:35 [gcc r12-5169] aarch64: Use type-qualified builtins for [R]ADDHN[2] Neon intrinsics Jonathan Wright
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).