arch/arm/arm_neon.h

39 typedef __Int32x4_t int32x4_t;
48 typedef __simd128_int32_t int32x4_t;
81 typedef __attribute__((neon_vector_type(4))) int32_t int32x4_t;
118 vcltq_s32(int32x4_t __v0, int32x4_t __v1)
124 static __inline int32x4_t
127 	return (int32x4_t) { __x, __x, __x, __x };
237 	return (uint32_t)__builtin_neon_vget_laneuv4si((int32x4_t)__v, __i);
242 	(uint32_t)__builtin_neon_vgetq_lane_i32((int32x4_t)(__v),	      \
366 static __inline int32x4_t
369 	return (int32x4_t)__v;
416 vreinterpretq_u8_s32(int32x4_t __v)
469 	(uint32x4_t)__builtin_neon_vsetq_lane_i32((__x), (int32x4_t)(__v),    \
489 static __inline int32x4_t
490 vshlq_n_s32(int32x4_t __v, uint8_t __bits)
493 	return (int32x4_t)__builtin_aarch64_ashlv4si(__v, __bits);
495 	return (int32x4_t)__builtin_neon_vshl_nv4si(__v, __bits);
500 	(int32x4_t)__builtin_neon_vshlq_n_v((int32x4_t)(__v), (__bits), 34)
509 	return (uint32x4_t)__builtin_aarch64_ashlv4si((int32x4_t)__v, __bits);
511 	return (uint32x4_t)__builtin_neon_vshl_nv4si((int32x4_t)__v, __bits);
516 	(uint32x4_t)__builtin_neon_vshlq_n_v((int32x4_t)(__v), (__bits), 50)
528 	return (uint32x4_t)__builtin_aarch64_lshrv4si((int32x4_t)__v, __bits);
531 	return (uint32x4_t)__builtin_neon_vshru_nv4si((int32x4_t)__v, __bits);
536 	(uint32x4_t)__builtin_neon_vshrq_n_v((int32x4_t)(__v), (__bits), 50)
561 static __inline int32x4_t
562 vsliq_n_s32(int32x4_t __vins, int32x4_t __vsh, uint8_t __bits)
565 	return (int32x4_t)__builtin_aarch64_ssli_nv4si(__vins, __vsh, __bits);
567 	return (int32x4_t)__builtin_neon_vsli_nv4si(__vins, __vsh, __bits);
573 	(int32x4_t)__builtin_neon_vsliq_n_v((int32x4_t)(__vins),	      \
574 	    (int32x4_t)(__vsh), (__bits), 34)
578 	int32x4_t __tvins = (__vins);					      \
579 	int32x4_t __tvsh = (__vsh);					      \
581 	int32x4_t __vins_r = __builtin_shufflevector(__tvins, __tvins,	      \
583 	int32x4_t __vsh_r = __builtin_shufflevector(__tvsh, __tvsh,	      \
585 	int32x4_t __r = __builtin_neon_vsliq_n_v(__tvins, __tvsh, __tbits,    \
600 	return (uint32x4_t)__builtin_neon_vsri_nv4si((int32x4_t)__vins,
601 	    (int32x4_t)__vsh, __bits);
607 	(int32x4_t)__builtin_neon_vsriq_n_v((int32x4_t)(__vins),	      \
608 	    (int32x4_t)(__vsh), (__bits), 34)
612 	int32x4_t __tvins = (__vins);					      \
613 	int32x4_t __tvsh = (__vsh);					      \
615 	int32x4_t __vins_r = __builtin_shufflevector(__tvins, __tvins,	      \
617 	int32x4_t __vsh_r = __builtin_shufflevector(__tvsh, __tvsh,	      \
619 	int32x4_t __r = __builtin_neon_vsriq_n_v(__tvins, __tvsh, __tbits,    \
634 	__builtin_aarch64_st1v4si(__p, (int32x4_t)__v);
638 	__builtin_neon_vst1v4si(__p, (int32x4_t)__v);