diff mbox

[AArch64/ARM,2/3] Rewrite AArch64 ZIP Intrinsics using __builtin_shuffle

Message ID 53340301.2070209@arm.com
State New
Headers show

Commit Message

Alan Lawrence March 27, 2014, 10:52 a.m. UTC 
This patch replaces the temporary inline assembler for vzip_* in arm_neon.h with 
equivalent calls to __builtin_shuffle. These are matched by 
aarch64_expand_vec_perm_const{,_1} to output the same assembler instructions.

Tests from first patch still passing on aarch64-none-elf and aarch64_be-none-elf.

gcc/ChangeLog:

2012-03-27  Alan Lawrence  <alan.lawrence@arm.com>

	* config/aarch64/arm_neon.h (vzip1_f32, vzip1_p8, vzip1_p16, vzip1_s8,
	vzip1_s16, vzip1_s32, vzip1_u8, vzip1_u16, vzip1_u32, vzip1q_f32,
	vzip1q_f64, vzip1q_p8, vzip1q_p16, vzip1q_s8, vzip1q_s16, vzip1q_s32,
	vzip1q_s64, vzip1q_u8, vzip1q_u16, vzip1q_u32, vzip1q_u64, vzip2_f32,
	vzip2_p8, vzip2_p16, vzip2_s8, vzip2_s16, vzip2_s32, vzip2_u8,
	vzip2_u16, vzip2_u32, vzip2q_f32, vzip2q_f64, vzip2q_p8, vzip2q_p16,
	vzip2q_s8, vzip2q_s16, vzip2q_s32, vzip2q_s64, vzip2q_u8, vzip2q_u16,
	vzip2q_u32, vzip2q_u64): Replace inline __asm__ with __builtin_shuffle.

Comments

Marcus Shawcroft April 23, 2014, 5:41 p.m. UTC | #1 
On 27 March 2014 10:52, Alan Lawrence <alan.lawrence@arm.com> wrote:
> This patch replaces the temporary inline assembler for vzip_* in arm_neon.h
> with equivalent calls to __builtin_shuffle. These are matched by
> aarch64_expand_vec_perm_const{,_1} to output the same assembler
> instructions.
>
> Tests from first patch still passing on aarch64-none-elf and
> aarch64_be-none-elf.
>
> gcc/ChangeLog:
>
> 2012-03-27  Alan Lawrence  <alan.lawrence@arm.com>
>
>         * config/aarch64/arm_neon.h (vzip1_f32, vzip1_p8, vzip1_p16,
> vzip1_s8,
>         vzip1_s16, vzip1_s32, vzip1_u8, vzip1_u16, vzip1_u32, vzip1q_f32,
>         vzip1q_f64, vzip1q_p8, vzip1q_p16, vzip1q_s8, vzip1q_s16,
> vzip1q_s32,
>         vzip1q_s64, vzip1q_u8, vzip1q_u16, vzip1q_u32, vzip1q_u64,
> vzip2_f32,
>         vzip2_p8, vzip2_p16, vzip2_s8, vzip2_s16, vzip2_s32, vzip2_u8,
>         vzip2_u16, vzip2_u32, vzip2q_f32, vzip2q_f64, vzip2q_p8, vzip2q_p16,
>         vzip2q_s8, vzip2q_s16, vzip2q_s32, vzip2q_s64, vzip2q_u8,
> vzip2q_u16,
>         vzip2q_u32, vzip2q_u64): Replace inline __asm__ with
> __builtin_shuffle.

OK /Marcus
Alan Lawrence April 29, 2014, 4:58 p.m. UTC | #2 
Committed as revision 209906.

A.

Marcus Shawcroft wrote:
> On 27 March 2014 10:52, Alan Lawrence <alan.lawrence@arm.com> wrote:
>> This patch replaces the temporary inline assembler for vzip_* in arm_neon.h
>> with equivalent calls to __builtin_shuffle. These are matched by
>> aarch64_expand_vec_perm_const{,_1} to output the same assembler
>> instructions.
>>
>> Tests from first patch still passing on aarch64-none-elf and
>> aarch64_be-none-elf.
>>
>> gcc/ChangeLog:
>>
>> 2012-03-27  Alan Lawrence  <alan.lawrence@arm.com>
>>
>>         * config/aarch64/arm_neon.h (vzip1_f32, vzip1_p8, vzip1_p16,
>> vzip1_s8,
>>         vzip1_s16, vzip1_s32, vzip1_u8, vzip1_u16, vzip1_u32, vzip1q_f32,
>>         vzip1q_f64, vzip1q_p8, vzip1q_p16, vzip1q_s8, vzip1q_s16,
>> vzip1q_s32,
>>         vzip1q_s64, vzip1q_u8, vzip1q_u16, vzip1q_u32, vzip1q_u64,
>> vzip2_f32,
>>         vzip2_p8, vzip2_p16, vzip2_s8, vzip2_s16, vzip2_s32, vzip2_u8,
>>         vzip2_u16, vzip2_u32, vzip2q_f32, vzip2q_f64, vzip2q_p8, vzip2q_p16,
>>         vzip2q_s8, vzip2q_s16, vzip2q_s32, vzip2q_s64, vzip2q_u8,
>> vzip2q_u16,
>>         vzip2q_u32, vzip2q_u64): Replace inline __asm__ with
>> __builtin_shuffle.
> 
> OK /Marcus
>
diff mbox

Patch

diff --git a/gcc/config/aarch64/arm_neon.h b/gcc/config/aarch64/arm_neon.h
index 6af99361..0ee0aae 100644
--- a/gcc/config/aarch64/arm_neon.h
+++ b/gcc/config/aarch64/arm_neon.h
@@ -13414,468 +13414,6 @@  vuzp2q_u64 (uint64x2_t a, uint64x2_t b)
   return result;
 }
 
-__extension__ static __inline float32x2_t __attribute__ ((__always_inline__))
-vzip1_f32 (float32x2_t a, float32x2_t b)
-{
-  float32x2_t result;
-  __asm__ ("zip1 %0.2s,%1.2s,%2.2s"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline poly8x8_t __attribute__ ((__always_inline__))
-vzip1_p8 (poly8x8_t a, poly8x8_t b)
-{
-  poly8x8_t result;
-  __asm__ ("zip1 %0.8b,%1.8b,%2.8b"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline poly16x4_t __attribute__ ((__always_inline__))
-vzip1_p16 (poly16x4_t a, poly16x4_t b)
-{
-  poly16x4_t result;
-  __asm__ ("zip1 %0.4h,%1.4h,%2.4h"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline int8x8_t __attribute__ ((__always_inline__))
-vzip1_s8 (int8x8_t a, int8x8_t b)
-{
-  int8x8_t result;
-  __asm__ ("zip1 %0.8b,%1.8b,%2.8b"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline int16x4_t __attribute__ ((__always_inline__))
-vzip1_s16 (int16x4_t a, int16x4_t b)
-{
-  int16x4_t result;
-  __asm__ ("zip1 %0.4h,%1.4h,%2.4h"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline int32x2_t __attribute__ ((__always_inline__))
-vzip1_s32 (int32x2_t a, int32x2_t b)
-{
-  int32x2_t result;
-  __asm__ ("zip1 %0.2s,%1.2s,%2.2s"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline uint8x8_t __attribute__ ((__always_inline__))
-vzip1_u8 (uint8x8_t a, uint8x8_t b)
-{
-  uint8x8_t result;
-  __asm__ ("zip1 %0.8b,%1.8b,%2.8b"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline uint16x4_t __attribute__ ((__always_inline__))
-vzip1_u16 (uint16x4_t a, uint16x4_t b)
-{
-  uint16x4_t result;
-  __asm__ ("zip1 %0.4h,%1.4h,%2.4h"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline uint32x2_t __attribute__ ((__always_inline__))
-vzip1_u32 (uint32x2_t a, uint32x2_t b)
-{
-  uint32x2_t result;
-  __asm__ ("zip1 %0.2s,%1.2s,%2.2s"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline float32x4_t __attribute__ ((__always_inline__))
-vzip1q_f32 (float32x4_t a, float32x4_t b)
-{
-  float32x4_t result;
-  __asm__ ("zip1 %0.4s,%1.4s,%2.4s"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline float64x2_t __attribute__ ((__always_inline__))
-vzip1q_f64 (float64x2_t a, float64x2_t b)
-{
-  float64x2_t result;
-  __asm__ ("zip1 %0.2d,%1.2d,%2.2d"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline poly8x16_t __attribute__ ((__always_inline__))
-vzip1q_p8 (poly8x16_t a, poly8x16_t b)
-{
-  poly8x16_t result;
-  __asm__ ("zip1 %0.16b,%1.16b,%2.16b"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline poly16x8_t __attribute__ ((__always_inline__))
-vzip1q_p16 (poly16x8_t a, poly16x8_t b)
-{
-  poly16x8_t result;
-  __asm__ ("zip1 %0.8h,%1.8h,%2.8h"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline int8x16_t __attribute__ ((__always_inline__))
-vzip1q_s8 (int8x16_t a, int8x16_t b)
-{
-  int8x16_t result;
-  __asm__ ("zip1 %0.16b,%1.16b,%2.16b"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline int16x8_t __attribute__ ((__always_inline__))
-vzip1q_s16 (int16x8_t a, int16x8_t b)
-{
-  int16x8_t result;
-  __asm__ ("zip1 %0.8h,%1.8h,%2.8h"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline int32x4_t __attribute__ ((__always_inline__))
-vzip1q_s32 (int32x4_t a, int32x4_t b)
-{
-  int32x4_t result;
-  __asm__ ("zip1 %0.4s,%1.4s,%2.4s"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline int64x2_t __attribute__ ((__always_inline__))
-vzip1q_s64 (int64x2_t a, int64x2_t b)
-{
-  int64x2_t result;
-  __asm__ ("zip1 %0.2d,%1.2d,%2.2d"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline uint8x16_t __attribute__ ((__always_inline__))
-vzip1q_u8 (uint8x16_t a, uint8x16_t b)
-{
-  uint8x16_t result;
-  __asm__ ("zip1 %0.16b,%1.16b,%2.16b"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline uint16x8_t __attribute__ ((__always_inline__))
-vzip1q_u16 (uint16x8_t a, uint16x8_t b)
-{
-  uint16x8_t result;
-  __asm__ ("zip1 %0.8h,%1.8h,%2.8h"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline uint32x4_t __attribute__ ((__always_inline__))
-vzip1q_u32 (uint32x4_t a, uint32x4_t b)
-{
-  uint32x4_t result;
-  __asm__ ("zip1 %0.4s,%1.4s,%2.4s"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline uint64x2_t __attribute__ ((__always_inline__))
-vzip1q_u64 (uint64x2_t a, uint64x2_t b)
-{
-  uint64x2_t result;
-  __asm__ ("zip1 %0.2d,%1.2d,%2.2d"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline float32x2_t __attribute__ ((__always_inline__))
-vzip2_f32 (float32x2_t a, float32x2_t b)
-{
-  float32x2_t result;
-  __asm__ ("zip2 %0.2s,%1.2s,%2.2s"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline poly8x8_t __attribute__ ((__always_inline__))
-vzip2_p8 (poly8x8_t a, poly8x8_t b)
-{
-  poly8x8_t result;
-  __asm__ ("zip2 %0.8b,%1.8b,%2.8b"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline poly16x4_t __attribute__ ((__always_inline__))
-vzip2_p16 (poly16x4_t a, poly16x4_t b)
-{
-  poly16x4_t result;
-  __asm__ ("zip2 %0.4h,%1.4h,%2.4h"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline int8x8_t __attribute__ ((__always_inline__))
-vzip2_s8 (int8x8_t a, int8x8_t b)
-{
-  int8x8_t result;
-  __asm__ ("zip2 %0.8b,%1.8b,%2.8b"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline int16x4_t __attribute__ ((__always_inline__))
-vzip2_s16 (int16x4_t a, int16x4_t b)
-{
-  int16x4_t result;
-  __asm__ ("zip2 %0.4h,%1.4h,%2.4h"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline int32x2_t __attribute__ ((__always_inline__))
-vzip2_s32 (int32x2_t a, int32x2_t b)
-{
-  int32x2_t result;
-  __asm__ ("zip2 %0.2s,%1.2s,%2.2s"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline uint8x8_t __attribute__ ((__always_inline__))
-vzip2_u8 (uint8x8_t a, uint8x8_t b)
-{
-  uint8x8_t result;
-  __asm__ ("zip2 %0.8b,%1.8b,%2.8b"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline uint16x4_t __attribute__ ((__always_inline__))
-vzip2_u16 (uint16x4_t a, uint16x4_t b)
-{
-  uint16x4_t result;
-  __asm__ ("zip2 %0.4h,%1.4h,%2.4h"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline uint32x2_t __attribute__ ((__always_inline__))
-vzip2_u32 (uint32x2_t a, uint32x2_t b)
-{
-  uint32x2_t result;
-  __asm__ ("zip2 %0.2s,%1.2s,%2.2s"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline float32x4_t __attribute__ ((__always_inline__))
-vzip2q_f32 (float32x4_t a, float32x4_t b)
-{
-  float32x4_t result;
-  __asm__ ("zip2 %0.4s,%1.4s,%2.4s"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline float64x2_t __attribute__ ((__always_inline__))
-vzip2q_f64 (float64x2_t a, float64x2_t b)
-{
-  float64x2_t result;
-  __asm__ ("zip2 %0.2d,%1.2d,%2.2d"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline poly8x16_t __attribute__ ((__always_inline__))
-vzip2q_p8 (poly8x16_t a, poly8x16_t b)
-{
-  poly8x16_t result;
-  __asm__ ("zip2 %0.16b,%1.16b,%2.16b"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline poly16x8_t __attribute__ ((__always_inline__))
-vzip2q_p16 (poly16x8_t a, poly16x8_t b)
-{
-  poly16x8_t result;
-  __asm__ ("zip2 %0.8h,%1.8h,%2.8h"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline int8x16_t __attribute__ ((__always_inline__))
-vzip2q_s8 (int8x16_t a, int8x16_t b)
-{
-  int8x16_t result;
-  __asm__ ("zip2 %0.16b,%1.16b,%2.16b"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline int16x8_t __attribute__ ((__always_inline__))
-vzip2q_s16 (int16x8_t a, int16x8_t b)
-{
-  int16x8_t result;
-  __asm__ ("zip2 %0.8h,%1.8h,%2.8h"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline int32x4_t __attribute__ ((__always_inline__))
-vzip2q_s32 (int32x4_t a, int32x4_t b)
-{
-  int32x4_t result;
-  __asm__ ("zip2 %0.4s,%1.4s,%2.4s"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline int64x2_t __attribute__ ((__always_inline__))
-vzip2q_s64 (int64x2_t a, int64x2_t b)
-{
-  int64x2_t result;
-  __asm__ ("zip2 %0.2d,%1.2d,%2.2d"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline uint8x16_t __attribute__ ((__always_inline__))
-vzip2q_u8 (uint8x16_t a, uint8x16_t b)
-{
-  uint8x16_t result;
-  __asm__ ("zip2 %0.16b,%1.16b,%2.16b"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline uint16x8_t __attribute__ ((__always_inline__))
-vzip2q_u16 (uint16x8_t a, uint16x8_t b)
-{
-  uint16x8_t result;
-  __asm__ ("zip2 %0.8h,%1.8h,%2.8h"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline uint32x4_t __attribute__ ((__always_inline__))
-vzip2q_u32 (uint32x4_t a, uint32x4_t b)
-{
-  uint32x4_t result;
-  __asm__ ("zip2 %0.4s,%1.4s,%2.4s"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
-__extension__ static __inline uint64x2_t __attribute__ ((__always_inline__))
-vzip2q_u64 (uint64x2_t a, uint64x2_t b)
-{
-  uint64x2_t result;
-  __asm__ ("zip2 %0.2d,%1.2d,%2.2d"
-           : "=w"(result)
-           : "w"(a), "w"(b)
-           : /* No clobbers */);
-  return result;
-}
-
 /* End of temporary inline asm implementations.  */
 
 /* Start of temporary inline asm for vldn, vstn and friends.  */
@@ -25316,6 +24854,444 @@  __INTERLEAVE_LIST (uzp)
 
 /* vzip */
 
+__extension__ static __inline float32x2_t __attribute__ ((__always_inline__))
+vzip1_f32 (float32x2_t __a, float32x2_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint32x2_t) {3, 1});
+#else
+  return __builtin_shuffle (__a, __b, (uint32x2_t) {0, 2});
+#endif
+}
+
+__extension__ static __inline poly8x8_t __attribute__ ((__always_inline__))
+vzip1_p8 (poly8x8_t __a, poly8x8_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint8x8_t) {12, 4, 13, 5, 14, 6, 15, 7});
+#else
+  return __builtin_shuffle (__a, __b, (uint8x8_t) {0, 8, 1, 9, 2, 10, 3, 11});
+#endif
+}
+
+__extension__ static __inline poly16x4_t __attribute__ ((__always_inline__))
+vzip1_p16 (poly16x4_t __a, poly16x4_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint16x4_t) {6, 2, 7, 3});
+#else
+  return __builtin_shuffle (__a, __b, (uint16x4_t) {0, 4, 1, 5});
+#endif
+}
+
+__extension__ static __inline int8x8_t __attribute__ ((__always_inline__))
+vzip1_s8 (int8x8_t __a, int8x8_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint8x8_t) {12, 4, 13, 5, 14, 6, 15, 7});
+#else
+  return __builtin_shuffle (__a, __b, (uint8x8_t) {0, 8, 1, 9, 2, 10, 3, 11});
+#endif
+}
+
+__extension__ static __inline int16x4_t __attribute__ ((__always_inline__))
+vzip1_s16 (int16x4_t __a, int16x4_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint16x4_t) {6, 2, 7, 3});
+#else
+  return __builtin_shuffle (__a, __b, (uint16x4_t) {0, 4, 1, 5});
+#endif
+}
+
+__extension__ static __inline int32x2_t __attribute__ ((__always_inline__))
+vzip1_s32 (int32x2_t __a, int32x2_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint32x2_t) {3, 1});
+#else
+  return __builtin_shuffle (__a, __b, (uint32x2_t) {0, 2});
+#endif
+}
+
+__extension__ static __inline uint8x8_t __attribute__ ((__always_inline__))
+vzip1_u8 (uint8x8_t __a, uint8x8_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint8x8_t) {12, 4, 13, 5, 14, 6, 15, 7});
+#else
+  return __builtin_shuffle (__a, __b, (uint8x8_t) {0, 8, 1, 9, 2, 10, 3, 11});
+#endif
+}
+
+__extension__ static __inline uint16x4_t __attribute__ ((__always_inline__))
+vzip1_u16 (uint16x4_t __a, uint16x4_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint16x4_t) {6, 2, 7, 3});
+#else
+  return __builtin_shuffle (__a, __b, (uint16x4_t) {0, 4, 1, 5});
+#endif
+}
+
+__extension__ static __inline uint32x2_t __attribute__ ((__always_inline__))
+vzip1_u32 (uint32x2_t __a, uint32x2_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint32x2_t) {3, 1});
+#else
+  return __builtin_shuffle (__a, __b, (uint32x2_t) {0, 2});
+#endif
+}
+
+__extension__ static __inline float32x4_t __attribute__ ((__always_inline__))
+vzip1q_f32 (float32x4_t __a, float32x4_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint32x4_t) {6, 2, 7, 3});
+#else
+  return __builtin_shuffle (__a, __b, (uint32x4_t) {0, 4, 1, 5});
+#endif
+}
+
+__extension__ static __inline float64x2_t __attribute__ ((__always_inline__))
+vzip1q_f64 (float64x2_t __a, float64x2_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint64x2_t) {3, 1});
+#else
+  return __builtin_shuffle (__a, __b, (uint64x2_t) {0, 2});
+#endif
+}
+
+__extension__ static __inline poly8x16_t __attribute__ ((__always_inline__))
+vzip1q_p8 (poly8x16_t __a, poly8x16_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint8x16_t)
+      {24, 8, 25, 9, 26, 10, 27, 11, 28, 12, 29, 13, 30, 14, 31, 15});
+#else
+  return __builtin_shuffle (__a, __b, (uint8x16_t)
+      {0, 16, 1, 17, 2, 18, 3, 19, 4, 20, 5, 21, 6, 22, 7, 23});
+#endif
+}
+
+__extension__ static __inline poly16x8_t __attribute__ ((__always_inline__))
+vzip1q_p16 (poly16x8_t __a, poly16x8_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint16x8_t)
+      {12, 4, 13, 5, 14, 6, 15, 7});
+#else
+  return __builtin_shuffle (__a, __b, (uint16x8_t) {0, 8, 1, 9, 2, 10, 3, 11});
+#endif
+}
+
+__extension__ static __inline int8x16_t __attribute__ ((__always_inline__))
+vzip1q_s8 (int8x16_t __a, int8x16_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint8x16_t)
+      {24, 8, 25, 9, 26, 10, 27, 11, 28, 12, 29, 13, 30, 14, 31, 15});
+#else
+  return __builtin_shuffle (__a, __b, (uint8x16_t)
+      {0, 16, 1, 17, 2, 18, 3, 19, 4, 20, 5, 21, 6, 22, 7, 23});
+#endif
+}
+
+__extension__ static __inline int16x8_t __attribute__ ((__always_inline__))
+vzip1q_s16 (int16x8_t __a, int16x8_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint16x8_t)
+      {12, 4, 13, 5, 14, 6, 15, 7});
+#else
+  return __builtin_shuffle (__a, __b, (uint16x8_t) {0, 8, 1, 9, 2, 10, 3, 11});
+#endif
+}
+
+__extension__ static __inline int32x4_t __attribute__ ((__always_inline__))
+vzip1q_s32 (int32x4_t __a, int32x4_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint32x4_t) {6, 2, 7, 3});
+#else
+  return __builtin_shuffle (__a, __b, (uint32x4_t) {0, 4, 1, 5});
+#endif
+}
+
+__extension__ static __inline int64x2_t __attribute__ ((__always_inline__))
+vzip1q_s64 (int64x2_t __a, int64x2_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint64x2_t) {3, 1});
+#else
+  return __builtin_shuffle (__a, __b, (uint64x2_t) {0, 2});
+#endif
+}
+
+__extension__ static __inline uint8x16_t __attribute__ ((__always_inline__))
+vzip1q_u8 (uint8x16_t __a, uint8x16_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint8x16_t)
+      {24, 8, 25, 9, 26, 10, 27, 11, 28, 12, 29, 13, 30, 14, 31, 15});
+#else
+  return __builtin_shuffle (__a, __b, (uint8x16_t)
+      {0, 16, 1, 17, 2, 18, 3, 19, 4, 20, 5, 21, 6, 22, 7, 23});
+#endif
+}
+
+__extension__ static __inline uint16x8_t __attribute__ ((__always_inline__))
+vzip1q_u16 (uint16x8_t __a, uint16x8_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint16x8_t)
+      {12, 4, 13, 5, 14, 6, 15, 7});
+#else
+  return __builtin_shuffle (__a, __b, (uint16x8_t) {0, 8, 1, 9, 2, 10, 3, 11});
+#endif
+}
+
+__extension__ static __inline uint32x4_t __attribute__ ((__always_inline__))
+vzip1q_u32 (uint32x4_t __a, uint32x4_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint32x4_t) {6, 2, 7, 3});
+#else
+  return __builtin_shuffle (__a, __b, (uint32x4_t) {0, 4, 1, 5});
+#endif
+}
+
+__extension__ static __inline uint64x2_t __attribute__ ((__always_inline__))
+vzip1q_u64 (uint64x2_t __a, uint64x2_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint64x2_t) {3, 1});
+#else
+  return __builtin_shuffle (__a, __b, (uint64x2_t) {0, 2});
+#endif
+}
+
+__extension__ static __inline float32x2_t __attribute__ ((__always_inline__))
+vzip2_f32 (float32x2_t __a, float32x2_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint32x2_t) {2, 0});
+#else
+  return __builtin_shuffle (__a, __b, (uint32x2_t) {1, 3});
+#endif
+}
+
+__extension__ static __inline poly8x8_t __attribute__ ((__always_inline__))
+vzip2_p8 (poly8x8_t __a, poly8x8_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint8x8_t) {8, 0, 9, 1, 10, 2, 11, 3});
+#else
+  return __builtin_shuffle (__a, __b, (uint8x8_t) {4, 12, 5, 13, 6, 14, 7, 15});
+#endif
+}
+
+__extension__ static __inline poly16x4_t __attribute__ ((__always_inline__))
+vzip2_p16 (poly16x4_t __a, poly16x4_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint16x4_t) {4, 0, 5, 1});
+#else
+  return __builtin_shuffle (__a, __b, (uint16x4_t) {2, 6, 3, 7});
+#endif
+}
+
+__extension__ static __inline int8x8_t __attribute__ ((__always_inline__))
+vzip2_s8 (int8x8_t __a, int8x8_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint8x8_t) {8, 0, 9, 1, 10, 2, 11, 3});
+#else
+  return __builtin_shuffle (__a, __b, (uint8x8_t) {4, 12, 5, 13, 6, 14, 7, 15});
+#endif
+}
+
+__extension__ static __inline int16x4_t __attribute__ ((__always_inline__))
+vzip2_s16 (int16x4_t __a, int16x4_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint16x4_t) {4, 0, 5, 1});
+#else
+  return __builtin_shuffle (__a, __b, (uint16x4_t) {2, 6, 3, 7});
+#endif
+}
+
+__extension__ static __inline int32x2_t __attribute__ ((__always_inline__))
+vzip2_s32 (int32x2_t __a, int32x2_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint32x2_t) {2, 0});
+#else
+  return __builtin_shuffle (__a, __b, (uint32x2_t) {1, 3});
+#endif
+}
+
+__extension__ static __inline uint8x8_t __attribute__ ((__always_inline__))
+vzip2_u8 (uint8x8_t __a, uint8x8_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint8x8_t) {8, 0, 9, 1, 10, 2, 11, 3});
+#else
+  return __builtin_shuffle (__a, __b, (uint8x8_t) {4, 12, 5, 13, 6, 14, 7, 15});
+#endif
+}
+
+__extension__ static __inline uint16x4_t __attribute__ ((__always_inline__))
+vzip2_u16 (uint16x4_t __a, uint16x4_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint16x4_t) {4, 0, 5, 1});
+#else
+  return __builtin_shuffle (__a, __b, (uint16x4_t) {2, 6, 3, 7});
+#endif
+}
+
+__extension__ static __inline uint32x2_t __attribute__ ((__always_inline__))
+vzip2_u32 (uint32x2_t __a, uint32x2_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint32x2_t) {2, 0});
+#else
+  return __builtin_shuffle (__a, __b, (uint32x2_t) {1, 3});
+#endif
+}
+
+__extension__ static __inline float32x4_t __attribute__ ((__always_inline__))
+vzip2q_f32 (float32x4_t __a, float32x4_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint32x4_t) {4, 0, 5, 1});
+#else
+  return __builtin_shuffle (__a, __b, (uint32x4_t) {2, 6, 3, 7});
+#endif
+}
+
+__extension__ static __inline float64x2_t __attribute__ ((__always_inline__))
+vzip2q_f64 (float64x2_t __a, float64x2_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint64x2_t) {2, 0});
+#else
+  return __builtin_shuffle (__a, __b, (uint64x2_t) {1, 3});
+#endif
+}
+
+__extension__ static __inline poly8x16_t __attribute__ ((__always_inline__))
+vzip2q_p8 (poly8x16_t __a, poly8x16_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint8x16_t)
+      {16, 0, 17, 1, 18, 2, 19, 3, 20, 4, 21, 5, 22, 6, 23, 7});
+#else
+  return __builtin_shuffle (__a, __b, (uint8x16_t)
+      {8, 24, 9, 25, 10, 26, 11, 27, 12, 28, 13, 29, 14, 30, 15, 31});
+#endif
+}
+
+__extension__ static __inline poly16x8_t __attribute__ ((__always_inline__))
+vzip2q_p16 (poly16x8_t __a, poly16x8_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint16x8_t) {8, 0, 9, 1, 10, 2, 11, 3});
+#else
+  return __builtin_shuffle (__a, __b, (uint16x8_t)
+      {4, 12, 5, 13, 6, 14, 7, 15});
+#endif
+}
+
+__extension__ static __inline int8x16_t __attribute__ ((__always_inline__))
+vzip2q_s8 (int8x16_t __a, int8x16_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint8x16_t)
+      {16, 0, 17, 1, 18, 2, 19, 3, 20, 4, 21, 5, 22, 6, 23, 7});
+#else
+  return __builtin_shuffle (__a, __b, (uint8x16_t)
+      {8, 24, 9, 25, 10, 26, 11, 27, 12, 28, 13, 29, 14, 30, 15, 31});
+#endif
+}
+
+__extension__ static __inline int16x8_t __attribute__ ((__always_inline__))
+vzip2q_s16 (int16x8_t __a, int16x8_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint16x8_t) {8, 0, 9, 1, 10, 2, 11, 3});
+#else
+  return __builtin_shuffle (__a, __b, (uint16x8_t)
+      {4, 12, 5, 13, 6, 14, 7, 15});
+#endif
+}
+
+__extension__ static __inline int32x4_t __attribute__ ((__always_inline__))
+vzip2q_s32 (int32x4_t __a, int32x4_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint32x4_t) {4, 0, 5, 1});
+#else
+  return __builtin_shuffle (__a, __b, (uint32x4_t) {2, 6, 3, 7});
+#endif
+}
+
+__extension__ static __inline int64x2_t __attribute__ ((__always_inline__))
+vzip2q_s64 (int64x2_t __a, int64x2_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint64x2_t) {2, 0});
+#else
+  return __builtin_shuffle (__a, __b, (uint64x2_t) {1, 3});
+#endif
+}
+
+__extension__ static __inline uint8x16_t __attribute__ ((__always_inline__))
+vzip2q_u8 (uint8x16_t __a, uint8x16_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint8x16_t)
+      {16, 0, 17, 1, 18, 2, 19, 3, 20, 4, 21, 5, 22, 6, 23, 7});
+#else
+  return __builtin_shuffle (__a, __b, (uint8x16_t)
+      {8, 24, 9, 25, 10, 26, 11, 27, 12, 28, 13, 29, 14, 30, 15, 31});
+#endif
+}
+
+__extension__ static __inline uint16x8_t __attribute__ ((__always_inline__))
+vzip2q_u16 (uint16x8_t __a, uint16x8_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint16x8_t) {8, 0, 9, 1, 10, 2, 11, 3});
+#else
+  return __builtin_shuffle (__a, __b, (uint16x8_t)
+      {4, 12, 5, 13, 6, 14, 7, 15});
+#endif
+}
+
+__extension__ static __inline uint32x4_t __attribute__ ((__always_inline__))
+vzip2q_u32 (uint32x4_t __a, uint32x4_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint32x4_t) {4, 0, 5, 1});
+#else
+  return __builtin_shuffle (__a, __b, (uint32x4_t) {2, 6, 3, 7});
+#endif
+}
+
+__extension__ static __inline uint64x2_t __attribute__ ((__always_inline__))
+vzip2q_u64 (uint64x2_t __a, uint64x2_t __b)
+{
+#ifdef __AARCH64EB__
+  return __builtin_shuffle (__a, __b, (uint64x2_t) {2, 0});
+#else
+  return __builtin_shuffle (__a, __b, (uint64x2_t) {1, 3});
+#endif
+}
+
 __INTERLEAVE_LIST (zip)
 
 #undef __INTERLEAVE_LIST