#ifndef __ARM_MVE_PRESERVE_USER_NAMESPACE
#define vst4q(__addr, __value) __arm_vst4q(__addr, __value)
#define vstrbq_scatter_offset(__base, __offset, __value) __arm_vstrbq_scatter_offset(__base, __offset, __value)
-#define vstrbq(__addr, __value) __arm_vstrbq(__addr, __value)
#define vstrwq_scatter_base(__addr, __offset, __value) __arm_vstrwq_scatter_base(__addr, __offset, __value)
#define vldrbq_gather_offset(__base, __offset) __arm_vldrbq_gather_offset(__base, __offset)
-#define vstrbq_p(__addr, __value, __p) __arm_vstrbq_p(__addr, __value, __p)
#define vstrbq_scatter_offset_p(__base, __offset, __value, __p) __arm_vstrbq_scatter_offset_p(__base, __offset, __value, __p)
#define vstrwq_scatter_base_p(__addr, __offset, __value, __p) __arm_vstrwq_scatter_base_p(__addr, __offset, __value, __p)
#define vldrbq_gather_offset_z(__base, __offset, __p) __arm_vldrbq_gather_offset_z(__base, __offset, __p)
#define vstrhq_scatter_offset_p(__base, __offset, __value, __p) __arm_vstrhq_scatter_offset_p(__base, __offset, __value, __p)
#define vstrhq_scatter_shifted_offset(__base, __offset, __value) __arm_vstrhq_scatter_shifted_offset(__base, __offset, __value)
#define vstrhq_scatter_shifted_offset_p(__base, __offset, __value, __p) __arm_vstrhq_scatter_shifted_offset_p(__base, __offset, __value, __p)
-#define vstrhq(__addr, __value) __arm_vstrhq(__addr, __value)
-#define vstrhq_p(__addr, __value, __p) __arm_vstrhq_p(__addr, __value, __p)
-#define vstrwq(__addr, __value) __arm_vstrwq(__addr, __value)
-#define vstrwq_p(__addr, __value, __p) __arm_vstrwq_p(__addr, __value, __p)
#define vstrdq_scatter_base_p(__addr, __offset, __value, __p) __arm_vstrdq_scatter_base_p(__addr, __offset, __value, __p)
#define vstrdq_scatter_base(__addr, __offset, __value) __arm_vstrdq_scatter_base(__addr, __offset, __value)
#define vstrdq_scatter_offset_p(__base, __offset, __value, __p) __arm_vstrdq_scatter_offset_p(__base, __offset, __value, __p)
#define vstrdq_scatter_base_wb_p(__addr, __offset, __value, __p) __arm_vstrdq_scatter_base_wb_p(__addr, __offset, __value, __p)
#define vstrwq_scatter_base_wb_p(__addr, __offset, __value, __p) __arm_vstrwq_scatter_base_wb_p(__addr, __offset, __value, __p)
#define vstrwq_scatter_base_wb(__addr, __offset, __value) __arm_vstrwq_scatter_base_wb(__addr, __offset, __value)
-#define vst1q_p(__addr, __value, __p) __arm_vst1q_p(__addr, __value, __p)
#define vst2q(__addr, __value) __arm_vst2q(__addr, __value)
-#define vld1q_z(__base, __p) __arm_vld1q_z(__base, __p)
#define vld2q(__addr) __arm_vld2q(__addr)
#define vld4q(__addr) __arm_vld4q(__addr)
#define vsetq_lane(__a, __b, __idx) __arm_vsetq_lane(__a, __b, __idx)
#define vst4q_f16( __addr, __value) __arm_vst4q_f16( __addr, __value)
#define vst4q_f32( __addr, __value) __arm_vst4q_f32( __addr, __value)
#define vpnot(__a) __arm_vpnot(__a)
-#define vstrbq_s8( __addr, __value) __arm_vstrbq_s8( __addr, __value)
-#define vstrbq_u8( __addr, __value) __arm_vstrbq_u8( __addr, __value)
-#define vstrbq_u16( __addr, __value) __arm_vstrbq_u16( __addr, __value)
#define vstrbq_scatter_offset_s8( __base, __offset, __value) __arm_vstrbq_scatter_offset_s8( __base, __offset, __value)
#define vstrbq_scatter_offset_u8( __base, __offset, __value) __arm_vstrbq_scatter_offset_u8( __base, __offset, __value)
#define vstrbq_scatter_offset_u16( __base, __offset, __value) __arm_vstrbq_scatter_offset_u16( __base, __offset, __value)
-#define vstrbq_s16( __addr, __value) __arm_vstrbq_s16( __addr, __value)
-#define vstrbq_u32( __addr, __value) __arm_vstrbq_u32( __addr, __value)
#define vstrbq_scatter_offset_s16( __base, __offset, __value) __arm_vstrbq_scatter_offset_s16( __base, __offset, __value)
#define vstrbq_scatter_offset_u32( __base, __offset, __value) __arm_vstrbq_scatter_offset_u32( __base, __offset, __value)
-#define vstrbq_s32( __addr, __value) __arm_vstrbq_s32( __addr, __value)
#define vstrbq_scatter_offset_s32( __base, __offset, __value) __arm_vstrbq_scatter_offset_s32( __base, __offset, __value)
#define vstrwq_scatter_base_s32(__addr, __offset, __value) __arm_vstrwq_scatter_base_s32(__addr, __offset, __value)
#define vstrwq_scatter_base_u32(__addr, __offset, __value) __arm_vstrwq_scatter_base_u32(__addr, __offset, __value)
#define vldrbq_gather_offset_u8(__base, __offset) __arm_vldrbq_gather_offset_u8(__base, __offset)
#define vldrbq_gather_offset_s8(__base, __offset) __arm_vldrbq_gather_offset_s8(__base, __offset)
-#define vldrbq_s8(__base) __arm_vldrbq_s8(__base)
-#define vldrbq_u8(__base) __arm_vldrbq_u8(__base)
#define vldrbq_gather_offset_u16(__base, __offset) __arm_vldrbq_gather_offset_u16(__base, __offset)
#define vldrbq_gather_offset_s16(__base, __offset) __arm_vldrbq_gather_offset_s16(__base, __offset)
-#define vldrbq_s16(__base) __arm_vldrbq_s16(__base)
-#define vldrbq_u16(__base) __arm_vldrbq_u16(__base)
#define vldrbq_gather_offset_u32(__base, __offset) __arm_vldrbq_gather_offset_u32(__base, __offset)
#define vldrbq_gather_offset_s32(__base, __offset) __arm_vldrbq_gather_offset_s32(__base, __offset)
-#define vldrbq_s32(__base) __arm_vldrbq_s32(__base)
-#define vldrbq_u32(__base) __arm_vldrbq_u32(__base)
#define vldrwq_gather_base_s32(__addr, __offset) __arm_vldrwq_gather_base_s32(__addr, __offset)
#define vldrwq_gather_base_u32(__addr, __offset) __arm_vldrwq_gather_base_u32(__addr, __offset)
-#define vstrbq_p_s8( __addr, __value, __p) __arm_vstrbq_p_s8( __addr, __value, __p)
-#define vstrbq_p_s32( __addr, __value, __p) __arm_vstrbq_p_s32( __addr, __value, __p)
-#define vstrbq_p_s16( __addr, __value, __p) __arm_vstrbq_p_s16( __addr, __value, __p)
-#define vstrbq_p_u8( __addr, __value, __p) __arm_vstrbq_p_u8( __addr, __value, __p)
-#define vstrbq_p_u32( __addr, __value, __p) __arm_vstrbq_p_u32( __addr, __value, __p)
-#define vstrbq_p_u16( __addr, __value, __p) __arm_vstrbq_p_u16( __addr, __value, __p)
#define vstrbq_scatter_offset_p_s8( __base, __offset, __value, __p) __arm_vstrbq_scatter_offset_p_s8( __base, __offset, __value, __p)
#define vstrbq_scatter_offset_p_s32( __base, __offset, __value, __p) __arm_vstrbq_scatter_offset_p_s32( __base, __offset, __value, __p)
#define vstrbq_scatter_offset_p_s16( __base, __offset, __value, __p) __arm_vstrbq_scatter_offset_p_s16( __base, __offset, __value, __p)
#define vldrbq_gather_offset_z_u16(__base, __offset, __p) __arm_vldrbq_gather_offset_z_u16(__base, __offset, __p)
#define vldrbq_gather_offset_z_u32(__base, __offset, __p) __arm_vldrbq_gather_offset_z_u32(__base, __offset, __p)
#define vldrbq_gather_offset_z_s8(__base, __offset, __p) __arm_vldrbq_gather_offset_z_s8(__base, __offset, __p)
-#define vldrbq_z_s16(__base, __p) __arm_vldrbq_z_s16(__base, __p)
-#define vldrbq_z_u8(__base, __p) __arm_vldrbq_z_u8(__base, __p)
-#define vldrbq_z_s8(__base, __p) __arm_vldrbq_z_s8(__base, __p)
-#define vldrbq_z_s32(__base, __p) __arm_vldrbq_z_s32(__base, __p)
-#define vldrbq_z_u16(__base, __p) __arm_vldrbq_z_u16(__base, __p)
-#define vldrbq_z_u32(__base, __p) __arm_vldrbq_z_u32(__base, __p)
#define vldrwq_gather_base_z_u32(__addr, __offset, __p) __arm_vldrwq_gather_base_z_u32(__addr, __offset, __p)
#define vldrwq_gather_base_z_s32(__addr, __offset, __p) __arm_vldrwq_gather_base_z_s32(__addr, __offset, __p)
#define vldrhq_gather_offset_s32(__base, __offset) __arm_vldrhq_gather_offset_s32(__base, __offset)
#define vldrhq_gather_shifted_offset_z_s16(__base, __offset, __p) __arm_vldrhq_gather_shifted_offset_z_s16(__base, __offset, __p)
#define vldrhq_gather_shifted_offset_z_u32(__base, __offset, __p) __arm_vldrhq_gather_shifted_offset_z_u32(__base, __offset, __p)
#define vldrhq_gather_shifted_offset_z_u16(__base, __offset, __p) __arm_vldrhq_gather_shifted_offset_z_u16(__base, __offset, __p)
-#define vldrhq_s32(__base) __arm_vldrhq_s32(__base)
-#define vldrhq_s16(__base) __arm_vldrhq_s16(__base)
-#define vldrhq_u32(__base) __arm_vldrhq_u32(__base)
-#define vldrhq_u16(__base) __arm_vldrhq_u16(__base)
-#define vldrhq_z_s32(__base, __p) __arm_vldrhq_z_s32(__base, __p)
-#define vldrhq_z_s16(__base, __p) __arm_vldrhq_z_s16(__base, __p)
-#define vldrhq_z_u32(__base, __p) __arm_vldrhq_z_u32(__base, __p)
-#define vldrhq_z_u16(__base, __p) __arm_vldrhq_z_u16(__base, __p)
-#define vldrwq_s32(__base) __arm_vldrwq_s32(__base)
-#define vldrwq_u32(__base) __arm_vldrwq_u32(__base)
-#define vldrwq_z_s32(__base, __p) __arm_vldrwq_z_s32(__base, __p)
-#define vldrwq_z_u32(__base, __p) __arm_vldrwq_z_u32(__base, __p)
-#define vldrhq_f16(__base) __arm_vldrhq_f16(__base)
-#define vldrhq_z_f16(__base, __p) __arm_vldrhq_z_f16(__base, __p)
-#define vldrwq_f32(__base) __arm_vldrwq_f32(__base)
-#define vldrwq_z_f32(__base, __p) __arm_vldrwq_z_f32(__base, __p)
#define vldrdq_gather_base_s64(__addr, __offset) __arm_vldrdq_gather_base_s64(__addr, __offset)
#define vldrdq_gather_base_u64(__addr, __offset) __arm_vldrdq_gather_base_u64(__addr, __offset)
#define vldrdq_gather_base_z_s64(__addr, __offset, __p) __arm_vldrdq_gather_base_z_s64(__addr, __offset, __p)
#define vldrwq_gather_shifted_offset_z_f32(__base, __offset, __p) __arm_vldrwq_gather_shifted_offset_z_f32(__base, __offset, __p)
#define vldrwq_gather_shifted_offset_z_s32(__base, __offset, __p) __arm_vldrwq_gather_shifted_offset_z_s32(__base, __offset, __p)
#define vldrwq_gather_shifted_offset_z_u32(__base, __offset, __p) __arm_vldrwq_gather_shifted_offset_z_u32(__base, __offset, __p)
-#define vstrhq_f16(__addr, __value) __arm_vstrhq_f16(__addr, __value)
#define vstrhq_scatter_offset_s32( __base, __offset, __value) __arm_vstrhq_scatter_offset_s32( __base, __offset, __value)
#define vstrhq_scatter_offset_s16( __base, __offset, __value) __arm_vstrhq_scatter_offset_s16( __base, __offset, __value)
#define vstrhq_scatter_offset_u32( __base, __offset, __value) __arm_vstrhq_scatter_offset_u32( __base, __offset, __value)
#define vstrhq_scatter_shifted_offset_p_s16( __base, __offset, __value, __p) __arm_vstrhq_scatter_shifted_offset_p_s16( __base, __offset, __value, __p)
#define vstrhq_scatter_shifted_offset_p_u32( __base, __offset, __value, __p) __arm_vstrhq_scatter_shifted_offset_p_u32( __base, __offset, __value, __p)
#define vstrhq_scatter_shifted_offset_p_u16( __base, __offset, __value, __p) __arm_vstrhq_scatter_shifted_offset_p_u16( __base, __offset, __value, __p)
-#define vstrhq_s32(__addr, __value) __arm_vstrhq_s32(__addr, __value)
-#define vstrhq_s16(__addr, __value) __arm_vstrhq_s16(__addr, __value)
-#define vstrhq_u32(__addr, __value) __arm_vstrhq_u32(__addr, __value)
-#define vstrhq_u16(__addr, __value) __arm_vstrhq_u16(__addr, __value)
-#define vstrhq_p_f16(__addr, __value, __p) __arm_vstrhq_p_f16(__addr, __value, __p)
-#define vstrhq_p_s32(__addr, __value, __p) __arm_vstrhq_p_s32(__addr, __value, __p)
-#define vstrhq_p_s16(__addr, __value, __p) __arm_vstrhq_p_s16(__addr, __value, __p)
-#define vstrhq_p_u32(__addr, __value, __p) __arm_vstrhq_p_u32(__addr, __value, __p)
-#define vstrhq_p_u16(__addr, __value, __p) __arm_vstrhq_p_u16(__addr, __value, __p)
-#define vstrwq_f32(__addr, __value) __arm_vstrwq_f32(__addr, __value)
-#define vstrwq_s32(__addr, __value) __arm_vstrwq_s32(__addr, __value)
-#define vstrwq_u32(__addr, __value) __arm_vstrwq_u32(__addr, __value)
-#define vstrwq_p_f32(__addr, __value, __p) __arm_vstrwq_p_f32(__addr, __value, __p)
-#define vstrwq_p_s32(__addr, __value, __p) __arm_vstrwq_p_s32(__addr, __value, __p)
-#define vstrwq_p_u32(__addr, __value, __p) __arm_vstrwq_p_u32(__addr, __value, __p)
#define vstrdq_scatter_base_p_s64(__addr, __offset, __value, __p) __arm_vstrdq_scatter_base_p_s64(__addr, __offset, __value, __p)
#define vstrdq_scatter_base_p_u64(__addr, __offset, __value, __p) __arm_vstrdq_scatter_base_p_u64(__addr, __offset, __value, __p)
#define vstrdq_scatter_base_s64(__addr, __offset, __value) __arm_vstrdq_scatter_base_s64(__addr, __offset, __value)
#define vstrwq_scatter_base_wb_s32(__addr, __offset, __value) __arm_vstrwq_scatter_base_wb_s32(__addr, __offset, __value)
#define vstrwq_scatter_base_wb_u32(__addr, __offset, __value) __arm_vstrwq_scatter_base_wb_u32(__addr, __offset, __value)
#define vstrwq_scatter_base_wb_f32(__addr, __offset, __value) __arm_vstrwq_scatter_base_wb_f32(__addr, __offset, __value)
-#define vst1q_p_u8(__addr, __value, __p) __arm_vst1q_p_u8(__addr, __value, __p)
-#define vst1q_p_s8(__addr, __value, __p) __arm_vst1q_p_s8(__addr, __value, __p)
#define vst2q_s8(__addr, __value) __arm_vst2q_s8(__addr, __value)
#define vst2q_u8(__addr, __value) __arm_vst2q_u8(__addr, __value)
-#define vld1q_z_u8(__base, __p) __arm_vld1q_z_u8(__base, __p)
-#define vld1q_z_s8(__base, __p) __arm_vld1q_z_s8(__base, __p)
#define vld2q_s8(__addr) __arm_vld2q_s8(__addr)
#define vld2q_u8(__addr) __arm_vld2q_u8(__addr)
#define vld4q_s8(__addr) __arm_vld4q_s8(__addr)
#define vld4q_u8(__addr) __arm_vld4q_u8(__addr)
-#define vst1q_p_u16(__addr, __value, __p) __arm_vst1q_p_u16(__addr, __value, __p)
-#define vst1q_p_s16(__addr, __value, __p) __arm_vst1q_p_s16(__addr, __value, __p)
#define vst2q_s16(__addr, __value) __arm_vst2q_s16(__addr, __value)
#define vst2q_u16(__addr, __value) __arm_vst2q_u16(__addr, __value)
-#define vld1q_z_u16(__base, __p) __arm_vld1q_z_u16(__base, __p)
-#define vld1q_z_s16(__base, __p) __arm_vld1q_z_s16(__base, __p)
#define vld2q_s16(__addr) __arm_vld2q_s16(__addr)
#define vld2q_u16(__addr) __arm_vld2q_u16(__addr)
#define vld4q_s16(__addr) __arm_vld4q_s16(__addr)
#define vld4q_u16(__addr) __arm_vld4q_u16(__addr)
-#define vst1q_p_u32(__addr, __value, __p) __arm_vst1q_p_u32(__addr, __value, __p)
-#define vst1q_p_s32(__addr, __value, __p) __arm_vst1q_p_s32(__addr, __value, __p)
#define vst2q_s32(__addr, __value) __arm_vst2q_s32(__addr, __value)
#define vst2q_u32(__addr, __value) __arm_vst2q_u32(__addr, __value)
-#define vld1q_z_u32(__base, __p) __arm_vld1q_z_u32(__base, __p)
-#define vld1q_z_s32(__base, __p) __arm_vld1q_z_s32(__base, __p)
#define vld2q_s32(__addr) __arm_vld2q_s32(__addr)
#define vld2q_u32(__addr) __arm_vld2q_u32(__addr)
#define vld4q_s32(__addr) __arm_vld4q_s32(__addr)
#define vld4q_u32(__addr) __arm_vld4q_u32(__addr)
#define vld4q_f16(__addr) __arm_vld4q_f16(__addr)
#define vld2q_f16(__addr) __arm_vld2q_f16(__addr)
-#define vld1q_z_f16(__base, __p) __arm_vld1q_z_f16(__base, __p)
#define vst2q_f16(__addr, __value) __arm_vst2q_f16(__addr, __value)
-#define vst1q_p_f16(__addr, __value, __p) __arm_vst1q_p_f16(__addr, __value, __p)
#define vld4q_f32(__addr) __arm_vld4q_f32(__addr)
#define vld2q_f32(__addr) __arm_vld2q_f32(__addr)
-#define vld1q_z_f32(__base, __p) __arm_vld1q_z_f32(__base, __p)
#define vst2q_f32(__addr, __value) __arm_vst2q_f32(__addr, __value)
-#define vst1q_p_f32(__addr, __value, __p) __arm_vst1q_p_f32(__addr, __value, __p)
#define vsetq_lane_f16(__a, __b, __idx) __arm_vsetq_lane_f16(__a, __b, __idx)
#define vsetq_lane_f32(__a, __b, __idx) __arm_vsetq_lane_f32(__a, __b, __idx)
#define vsetq_lane_s16(__a, __b, __idx) __arm_vsetq_lane_s16(__a, __b, __idx)
__builtin_mve_vstrbq_scatter_offset_uv8hi ((__builtin_neon_qi *) __base, __offset, __value);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_s8 (int8_t * __addr, int8x16_t __value)
-{
- __builtin_mve_vstrbq_sv16qi ((__builtin_neon_qi *) __addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_s32 (int8_t * __addr, int32x4_t __value)
-{
- __builtin_mve_vstrbq_sv4si ((__builtin_neon_qi *) __addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_s16 (int8_t * __addr, int16x8_t __value)
-{
- __builtin_mve_vstrbq_sv8hi ((__builtin_neon_qi *) __addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_u8 (uint8_t * __addr, uint8x16_t __value)
-{
- __builtin_mve_vstrbq_uv16qi ((__builtin_neon_qi *) __addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_u32 (uint8_t * __addr, uint32x4_t __value)
-{
- __builtin_mve_vstrbq_uv4si ((__builtin_neon_qi *) __addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_u16 (uint8_t * __addr, uint16x8_t __value)
-{
- __builtin_mve_vstrbq_uv8hi ((__builtin_neon_qi *) __addr, __value);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vstrwq_scatter_base_s32 (uint32x4_t __addr, const int __offset, int32x4_t __value)
return __builtin_mve_vldrbq_gather_offset_sv16qi ((__builtin_neon_qi *) __base, __offset);
}
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_s8 (int8_t const * __base)
-{
- return __builtin_mve_vldrbq_sv16qi ((__builtin_neon_qi *) __base);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_u8 (uint8_t const * __base)
-{
- return __builtin_mve_vldrbq_uv16qi ((__builtin_neon_qi *) __base);
-}
-
__extension__ extern __inline uint16x8_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrbq_gather_offset_u16 (uint8_t const * __base, uint16x8_t __offset)
return __builtin_mve_vldrbq_gather_offset_sv8hi ((__builtin_neon_qi *) __base, __offset);
}
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_s16 (int8_t const * __base)
-{
- return __builtin_mve_vldrbq_sv8hi ((__builtin_neon_qi *) __base);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_u16 (uint8_t const * __base)
-{
- return __builtin_mve_vldrbq_uv8hi ((__builtin_neon_qi *) __base);
-}
-
__extension__ extern __inline uint32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrbq_gather_offset_u32 (uint8_t const * __base, uint32x4_t __offset)
return __builtin_mve_vldrbq_gather_offset_sv4si ((__builtin_neon_qi *) __base, __offset);
}
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_s32 (int8_t const * __base)
-{
- return __builtin_mve_vldrbq_sv4si ((__builtin_neon_qi *) __base);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_u32 (uint8_t const * __base)
-{
- return __builtin_mve_vldrbq_uv4si ((__builtin_neon_qi *) __base);
-}
-
__extension__ extern __inline int32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrwq_gather_base_s32 (uint32x4_t __addr, const int __offset)
return __builtin_mve_vldrwq_gather_base_uv4si (__addr, __offset);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_p_s8 (int8_t * __addr, int8x16_t __value, mve_pred16_t __p)
-{
- __builtin_mve_vstrbq_p_sv16qi ((__builtin_neon_qi *) __addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_p_s32 (int8_t * __addr, int32x4_t __value, mve_pred16_t __p)
-{
- __builtin_mve_vstrbq_p_sv4si ((__builtin_neon_qi *) __addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_p_s16 (int8_t * __addr, int16x8_t __value, mve_pred16_t __p)
-{
- __builtin_mve_vstrbq_p_sv8hi ((__builtin_neon_qi *) __addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_p_u8 (uint8_t * __addr, uint8x16_t __value, mve_pred16_t __p)
-{
- __builtin_mve_vstrbq_p_uv16qi ((__builtin_neon_qi *) __addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_p_u32 (uint8_t * __addr, uint32x4_t __value, mve_pred16_t __p)
-{
- __builtin_mve_vstrbq_p_uv4si ((__builtin_neon_qi *) __addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_p_u16 (uint8_t * __addr, uint16x8_t __value, mve_pred16_t __p)
-{
- __builtin_mve_vstrbq_p_uv8hi ((__builtin_neon_qi *) __addr, __value, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vstrbq_scatter_offset_p_s8 (int8_t * __base, uint8x16_t __offset, int8x16_t __value, mve_pred16_t __p)
return __builtin_mve_vldrbq_gather_offset_z_uv8hi ((__builtin_neon_qi *) __base, __offset, __p);
}
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_z_s8 (int8_t const * __base, mve_pred16_t __p)
-{
- return __builtin_mve_vldrbq_z_sv16qi ((__builtin_neon_qi *) __base, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_z_s32 (int8_t const * __base, mve_pred16_t __p)
-{
- return __builtin_mve_vldrbq_z_sv4si ((__builtin_neon_qi *) __base, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_z_s16 (int8_t const * __base, mve_pred16_t __p)
-{
- return __builtin_mve_vldrbq_z_sv8hi ((__builtin_neon_qi *) __base, __p);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_z_u8 (uint8_t const * __base, mve_pred16_t __p)
-{
- return __builtin_mve_vldrbq_z_uv16qi ((__builtin_neon_qi *) __base, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_z_u32 (uint8_t const * __base, mve_pred16_t __p)
-{
- return __builtin_mve_vldrbq_z_uv4si ((__builtin_neon_qi *) __base, __p);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrbq_z_u16 (uint8_t const * __base, mve_pred16_t __p)
-{
- return __builtin_mve_vldrbq_z_uv8hi ((__builtin_neon_qi *) __base, __p);
-}
-
__extension__ extern __inline int32x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrwq_gather_base_z_s32 (uint32x4_t __addr, const int __offset, mve_pred16_t __p)
return __builtin_mve_vldrhq_gather_shifted_offset_z_uv8hi ((__builtin_neon_hi *) __base, __offset, __p);
}
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_s32 (int16_t const * __base)
-{
- return __builtin_mve_vldrhq_sv4si ((__builtin_neon_hi *) __base);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_s16 (int16_t const * __base)
-{
- return __builtin_mve_vldrhq_sv8hi ((__builtin_neon_hi *) __base);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_u32 (uint16_t const * __base)
-{
- return __builtin_mve_vldrhq_uv4si ((__builtin_neon_hi *) __base);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_u16 (uint16_t const * __base)
-{
- return __builtin_mve_vldrhq_uv8hi ((__builtin_neon_hi *) __base);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_z_s32 (int16_t const * __base, mve_pred16_t __p)
-{
- return __builtin_mve_vldrhq_z_sv4si ((__builtin_neon_hi *) __base, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_z_s16 (int16_t const * __base, mve_pred16_t __p)
-{
- return __builtin_mve_vldrhq_z_sv8hi ((__builtin_neon_hi *) __base, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_z_u32 (uint16_t const * __base, mve_pred16_t __p)
-{
- return __builtin_mve_vldrhq_z_uv4si ((__builtin_neon_hi *) __base, __p);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_z_u16 (uint16_t const * __base, mve_pred16_t __p)
-{
- return __builtin_mve_vldrhq_z_uv8hi ((__builtin_neon_hi *) __base, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_s32 (int32_t const * __base)
-{
- return __builtin_mve_vldrwq_sv4si ((__builtin_neon_si *) __base);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_u32 (uint32_t const * __base)
-{
- return __builtin_mve_vldrwq_uv4si ((__builtin_neon_si *) __base);
-}
-
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_z_s32 (int32_t const * __base, mve_pred16_t __p)
-{
- return __builtin_mve_vldrwq_z_sv4si ((__builtin_neon_si *) __base, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_z_u32 (uint32_t const * __base, mve_pred16_t __p)
-{
- return __builtin_mve_vldrwq_z_uv4si ((__builtin_neon_si *) __base, __p);
-}
-
__extension__ extern __inline int64x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrdq_gather_base_s64 (uint64x2_t __addr, const int __offset)
__builtin_mve_vstrhq_scatter_shifted_offset_p_uv8hi ((__builtin_neon_hi *) __base, __offset, __value, __p);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_s32 (int16_t * __addr, int32x4_t __value)
-{
- __builtin_mve_vstrhq_sv4si ((__builtin_neon_hi *) __addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_s16 (int16_t * __addr, int16x8_t __value)
-{
- __builtin_mve_vstrhq_sv8hi ((__builtin_neon_hi *) __addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_u32 (uint16_t * __addr, uint32x4_t __value)
-{
- __builtin_mve_vstrhq_uv4si ((__builtin_neon_hi *) __addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_u16 (uint16_t * __addr, uint16x8_t __value)
-{
- __builtin_mve_vstrhq_uv8hi ((__builtin_neon_hi *) __addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_p_s32 (int16_t * __addr, int32x4_t __value, mve_pred16_t __p)
-{
- __builtin_mve_vstrhq_p_sv4si ((__builtin_neon_hi *) __addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_p_s16 (int16_t * __addr, int16x8_t __value, mve_pred16_t __p)
-{
- __builtin_mve_vstrhq_p_sv8hi ((__builtin_neon_hi *) __addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_p_u32 (uint16_t * __addr, uint32x4_t __value, mve_pred16_t __p)
-{
- __builtin_mve_vstrhq_p_uv4si ((__builtin_neon_hi *) __addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_p_u16 (uint16_t * __addr, uint16x8_t __value, mve_pred16_t __p)
-{
- __builtin_mve_vstrhq_p_uv8hi ((__builtin_neon_hi *) __addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrwq_s32 (int32_t * __addr, int32x4_t __value)
-{
- __builtin_mve_vstrwq_sv4si ((__builtin_neon_si *) __addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrwq_u32 (uint32_t * __addr, uint32x4_t __value)
-{
- __builtin_mve_vstrwq_uv4si ((__builtin_neon_si *) __addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrwq_p_s32 (int32_t * __addr, int32x4_t __value, mve_pred16_t __p)
-{
- __builtin_mve_vstrwq_p_sv4si ((__builtin_neon_si *) __addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrwq_p_u32 (uint32_t * __addr, uint32x4_t __value, mve_pred16_t __p)
-{
- __builtin_mve_vstrwq_p_uv4si ((__builtin_neon_si *) __addr, __value, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vstrdq_scatter_base_p_s64 (uint64x2_t __addr, const int __offset, int64x2_t __value, mve_pred16_t __p)
*__addr = __builtin_mve_vstrwq_scatter_base_wb_uv4si (*__addr, __offset, __value);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p_u8 (uint8_t * __addr, uint8x16_t __value, mve_pred16_t __p)
-{
- return __arm_vstrbq_p_u8 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p_s8 (int8_t * __addr, int8x16_t __value, mve_pred16_t __p)
-{
- return __arm_vstrbq_p_s8 (__addr, __value, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vst2q_s8 (int8_t * __addr, int8x16x2_t __value)
__builtin_mve_vst2qv16qi ((__builtin_neon_qi *) __addr, __rv.__o);
}
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z_u8 (uint8_t const *__base, mve_pred16_t __p)
-{
- return __arm_vldrbq_z_u8 ( __base, __p);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z_s8 (int8_t const *__base, mve_pred16_t __p)
-{
- return __arm_vldrbq_z_s8 ( __base, __p);
-}
-
__extension__ extern __inline int8x16x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vld2q_s8 (int8_t const * __addr)
return __rv.__i;
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p_u16 (uint16_t * __addr, uint16x8_t __value, mve_pred16_t __p)
-{
- return __arm_vstrhq_p_u16 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p_s16 (int16_t * __addr, int16x8_t __value, mve_pred16_t __p)
-{
- return __arm_vstrhq_p_s16 (__addr, __value, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vst2q_s16 (int16_t * __addr, int16x8x2_t __value)
__builtin_mve_vst2qv8hi ((__builtin_neon_hi *) __addr, __rv.__o);
}
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z_u16 (uint16_t const *__base, mve_pred16_t __p)
-{
- return __arm_vldrhq_z_u16 ( __base, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z_s16 (int16_t const *__base, mve_pred16_t __p)
-{
- return __arm_vldrhq_z_s16 ( __base, __p);
-}
-
__extension__ extern __inline int16x8x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vld2q_s16 (int16_t const * __addr)
return __rv.__i;
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p_u32 (uint32_t * __addr, uint32x4_t __value, mve_pred16_t __p)
-{
- return __arm_vstrwq_p_u32 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p_s32 (int32_t * __addr, int32x4_t __value, mve_pred16_t __p)
-{
- return __arm_vstrwq_p_s32 (__addr, __value, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vst2q_s32 (int32_t * __addr, int32x4x2_t __value)
__builtin_mve_vst2qv4si ((__builtin_neon_si *) __addr, __rv.__o);
}
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z_u32 (uint32_t const *__base, mve_pred16_t __p)
-{
- return __arm_vldrwq_z_u32 ( __base, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z_s32 (int32_t const *__base, mve_pred16_t __p)
-{
- return __arm_vldrwq_z_s32 ( __base, __p);
-}
-
__extension__ extern __inline int32x4x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vld2q_s32 (int32_t const * __addr)
__builtin_mve_vst4qv4sf (__addr, __rv.__o);
}
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_f32 (float32_t const * __base)
-{
- return __builtin_mve_vldrwq_fv4sf((__builtin_neon_si *) __base);
-}
-
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrwq_z_f32 (float32_t const * __base, mve_pred16_t __p)
-{
- return __builtin_mve_vldrwq_z_fv4sf((__builtin_neon_si *) __base, __p);
-}
-
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_z_f16 (float16_t const * __base, mve_pred16_t __p)
-{
- return __builtin_mve_vldrhq_z_fv8hf((__builtin_neon_hi *) __base, __p);
-}
-
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vldrhq_f16 (float16_t const * __base)
-{
- return __builtin_mve_vldrhq_fv8hf((__builtin_neon_hi *) __base);
-}
-
__extension__ extern __inline float16x8_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vldrhq_gather_offset_f16 (float16_t const * __base, uint16x8_t __offset)
return __builtin_mve_vldrwq_gather_shifted_offset_z_fv4sf ((__builtin_neon_si *) __base, __offset, __p);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrwq_p_f32 (float32_t * __addr, float32x4_t __value, mve_pred16_t __p)
-{
- __builtin_mve_vstrwq_p_fv4sf ((__builtin_neon_si *) __addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrwq_f32 (float32_t * __addr, float32x4_t __value)
-{
- __builtin_mve_vstrwq_fv4sf ((__builtin_neon_si *) __addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_f16 (float16_t * __addr, float16x8_t __value)
-{
- __builtin_mve_vstrhq_fv8hf ((__builtin_neon_hi *) __addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_p_f16 (float16_t * __addr, float16x8_t __value, mve_pred16_t __p)
-{
- __builtin_mve_vstrhq_p_fv8hf ((__builtin_neon_hi *) __addr, __value, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vstrhq_scatter_offset_f16 (float16_t * __base, uint16x8_t __offset, float16x8_t __value)
return __rv.__i;
}
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z_f16 (float16_t const *__base, mve_pred16_t __p)
-{
- return __arm_vldrhq_z_f16 (__base, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vst2q_f16 (float16_t * __addr, float16x8x2_t __value)
__builtin_mve_vst2qv8hf (__addr, __rv.__o);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p_f16 (float16_t * __addr, float16x8_t __value, mve_pred16_t __p)
-{
- return __arm_vstrhq_p_f16 (__addr, __value, __p);
-}
-
__extension__ extern __inline float32x4x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vld4q_f32 (float32_t const * __addr)
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vld2q_f32 (float32_t const * __addr)
{
- union { float32x4x2_t __i; __builtin_neon_oi __o; } __rv;
- __rv.__o = __builtin_mve_vld2qv4sf (__addr);
- return __rv.__i;
-}
-
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z_f32 (float32_t const *__base, mve_pred16_t __p)
-{
- return __arm_vldrwq_z_f32 (__base, __p);
+ union { float32x4x2_t __i; __builtin_neon_oi __o; } __rv;
+ __rv.__o = __builtin_mve_vld2qv4sf (__addr);
+ return __rv.__i;
}
__extension__ extern __inline void
__builtin_mve_vst2qv4sf (__addr, __rv.__o);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p_f32 (float32_t * __addr, float32x4_t __value, mve_pred16_t __p)
-{
- return __arm_vstrwq_p_f32 (__addr, __value, __p);
-}
-
__extension__ extern __inline float16x8_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vsetq_lane_f16 (float16_t __a, float16x8_t __b, const int __idx)
__arm_vstrbq_scatter_offset_u16 (__base, __offset, __value);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq (int8_t * __addr, int8x16_t __value)
-{
- __arm_vstrbq_s8 (__addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq (int8_t * __addr, int32x4_t __value)
-{
- __arm_vstrbq_s32 (__addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq (int8_t * __addr, int16x8_t __value)
-{
- __arm_vstrbq_s16 (__addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq (uint8_t * __addr, uint8x16_t __value)
-{
- __arm_vstrbq_u8 (__addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq (uint8_t * __addr, uint32x4_t __value)
-{
- __arm_vstrbq_u32 (__addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq (uint8_t * __addr, uint16x8_t __value)
-{
- __arm_vstrbq_u16 (__addr, __value);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vstrwq_scatter_base (uint32x4_t __addr, const int __offset, int32x4_t __value)
return __arm_vldrbq_gather_offset_s32 (__base, __offset);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_p (int8_t * __addr, int8x16_t __value, mve_pred16_t __p)
-{
- __arm_vstrbq_p_s8 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_p (int8_t * __addr, int32x4_t __value, mve_pred16_t __p)
-{
- __arm_vstrbq_p_s32 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_p (int8_t * __addr, int16x8_t __value, mve_pred16_t __p)
-{
- __arm_vstrbq_p_s16 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_p (uint8_t * __addr, uint8x16_t __value, mve_pred16_t __p)
-{
- __arm_vstrbq_p_u8 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_p (uint8_t * __addr, uint32x4_t __value, mve_pred16_t __p)
-{
- __arm_vstrbq_p_u32 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrbq_p (uint8_t * __addr, uint16x8_t __value, mve_pred16_t __p)
-{
- __arm_vstrbq_p_u16 (__addr, __value, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vstrbq_scatter_offset_p (int8_t * __base, uint8x16_t __offset, int8x16_t __value, mve_pred16_t __p)
__arm_vstrhq_scatter_shifted_offset_p_u16 (__base, __offset, __value, __p);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq (int16_t * __addr, int32x4_t __value)
-{
- __arm_vstrhq_s32 (__addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq (int16_t * __addr, int16x8_t __value)
-{
- __arm_vstrhq_s16 (__addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq (uint16_t * __addr, uint32x4_t __value)
-{
- __arm_vstrhq_u32 (__addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq (uint16_t * __addr, uint16x8_t __value)
-{
- __arm_vstrhq_u16 (__addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_p (int16_t * __addr, int32x4_t __value, mve_pred16_t __p)
-{
- __arm_vstrhq_p_s32 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_p (int16_t * __addr, int16x8_t __value, mve_pred16_t __p)
-{
- __arm_vstrhq_p_s16 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_p (uint16_t * __addr, uint32x4_t __value, mve_pred16_t __p)
-{
- __arm_vstrhq_p_u32 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_p (uint16_t * __addr, uint16x8_t __value, mve_pred16_t __p)
-{
- __arm_vstrhq_p_u16 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrwq (int32_t * __addr, int32x4_t __value)
-{
- __arm_vstrwq_s32 (__addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrwq (uint32_t * __addr, uint32x4_t __value)
-{
- __arm_vstrwq_u32 (__addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrwq_p (int32_t * __addr, int32x4_t __value, mve_pred16_t __p)
-{
- __arm_vstrwq_p_s32 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrwq_p (uint32_t * __addr, uint32x4_t __value, mve_pred16_t __p)
-{
- __arm_vstrwq_p_u32 (__addr, __value, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vstrdq_scatter_base_p (uint64x2_t __addr, const int __offset, int64x2_t __value, mve_pred16_t __p)
__arm_vstrwq_scatter_base_wb_u32 (__addr, __offset, __value);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p (uint8_t * __addr, uint8x16_t __value, mve_pred16_t __p)
-{
- __arm_vst1q_p_u8 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p (int8_t * __addr, int8x16_t __value, mve_pred16_t __p)
-{
- __arm_vst1q_p_s8 (__addr, __value, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vst2q (int8_t * __addr, int8x16x2_t __value)
__arm_vst2q_u8 (__addr, __value);
}
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z (uint8_t const *__base, mve_pred16_t __p)
-{
- return __arm_vld1q_z_u8 (__base, __p);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z (int8_t const *__base, mve_pred16_t __p)
-{
- return __arm_vld1q_z_s8 (__base, __p);
-}
-
__extension__ extern __inline int8x16x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vld2q (int8_t const * __addr)
return __arm_vld4q_u8 (__addr);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p (uint16_t * __addr, uint16x8_t __value, mve_pred16_t __p)
-{
- __arm_vst1q_p_u16 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p (int16_t * __addr, int16x8_t __value, mve_pred16_t __p)
-{
- __arm_vst1q_p_s16 (__addr, __value, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vst2q (int16_t * __addr, int16x8x2_t __value)
__arm_vst2q_u16 (__addr, __value);
}
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z (uint16_t const *__base, mve_pred16_t __p)
-{
- return __arm_vld1q_z_u16 (__base, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z (int16_t const *__base, mve_pred16_t __p)
-{
- return __arm_vld1q_z_s16 (__base, __p);
-}
-
__extension__ extern __inline int16x8x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vld2q (int16_t const * __addr)
return __arm_vld4q_u16 (__addr);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p (uint32_t * __addr, uint32x4_t __value, mve_pred16_t __p)
-{
- __arm_vst1q_p_u32 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p (int32_t * __addr, int32x4_t __value, mve_pred16_t __p)
-{
- __arm_vst1q_p_s32 (__addr, __value, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vst2q (int32_t * __addr, int32x4x2_t __value)
__arm_vst2q_u32 (__addr, __value);
}
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z (uint32_t const *__base, mve_pred16_t __p)
-{
- return __arm_vld1q_z_u32 (__base, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z (int32_t const *__base, mve_pred16_t __p)
-{
- return __arm_vld1q_z_s32 (__base, __p);
-}
-
__extension__ extern __inline int32x4x2_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vld2q (int32_t const * __addr)
return __arm_vldrwq_gather_shifted_offset_z_f32 (__base, __offset, __p);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrwq_p (float32_t * __addr, float32x4_t __value, mve_pred16_t __p)
-{
- __arm_vstrwq_p_f32 (__addr, __value, __p);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrwq (float32_t * __addr, float32x4_t __value)
-{
- __arm_vstrwq_f32 (__addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq (float16_t * __addr, float16x8_t __value)
-{
- __arm_vstrhq_f16 (__addr, __value);
-}
-
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vstrhq_p (float16_t * __addr, float16x8_t __value, mve_pred16_t __p)
-{
- __arm_vstrhq_p_f16 (__addr, __value, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vstrhq_scatter_offset (float16_t * __base, uint16x8_t __offset, float16x8_t __value)
return __arm_vld2q_f16 (__addr);
}
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z (float16_t const *__base, mve_pred16_t __p)
-{
- return __arm_vld1q_z_f16 (__base, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vst2q (float16_t * __addr, float16x8x2_t __value)
__arm_vst2q_f16 (__addr, __value);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p (float16_t * __addr, float16x8_t __value, mve_pred16_t __p)
-{
- __arm_vst1q_p_f16 (__addr, __value, __p);
-}
-
__extension__ extern __inline float32x4x4_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vld4q (float32_t const * __addr)
return __arm_vld2q_f32 (__addr);
}
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vld1q_z (float32_t const *__base, mve_pred16_t __p)
-{
- return __arm_vld1q_z_f32 (__base, __p);
-}
-
__extension__ extern __inline void
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vst2q (float32_t * __addr, float32x4x2_t __value)
__arm_vst2q_f32 (__addr, __value);
}
-__extension__ extern __inline void
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vst1q_p (float32_t * __addr, float32x4_t __value, mve_pred16_t __p)
-{
- __arm_vst1q_p_f32 (__addr, __value, __p);
-}
-
__extension__ extern __inline float16x8_t
__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
__arm_vsetq_lane (float16_t __a, float16x8_t __b, const int __idx)
int (*)[__ARM_mve_type_float16_t_ptr][__ARM_mve_type_float16x8x4_t]: __arm_vst4q_f16 (__ARM_mve_coerce_f16_ptr(__p0, float16_t *), __ARM_mve_coerce(__p1, float16x8x4_t)), \
int (*)[__ARM_mve_type_float32_t_ptr][__ARM_mve_type_float32x4x4_t]: __arm_vst4q_f32 (__ARM_mve_coerce_f32_ptr(__p0, float32_t *), __ARM_mve_coerce(__p1, float32x4x4_t)));})
-#define __arm_vld1q_z(p0,p1) ( \
- _Generic( (int (*)[__ARM_mve_typeid(p0)])0, \
- int (*)[__ARM_mve_type_int8_t_ptr]: __arm_vld1q_z_s8 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), p1), \
- int (*)[__ARM_mve_type_int16_t_ptr]: __arm_vld1q_z_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), p1), \
- int (*)[__ARM_mve_type_int32_t_ptr]: __arm_vld1q_z_s32 (__ARM_mve_coerce_s32_ptr(p0, int32_t *), p1), \
- int (*)[__ARM_mve_type_uint8_t_ptr]: __arm_vld1q_z_u8 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), p1), \
- int (*)[__ARM_mve_type_uint16_t_ptr]: __arm_vld1q_z_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), p1), \
- int (*)[__ARM_mve_type_uint32_t_ptr]: __arm_vld1q_z_u32 (__ARM_mve_coerce_u32_ptr(p0, uint32_t *), p1), \
- int (*)[__ARM_mve_type_float16_t_ptr]: __arm_vld1q_z_f16 (__ARM_mve_coerce_f16_ptr(p0, float16_t *), p1), \
- int (*)[__ARM_mve_type_float32_t_ptr]: __arm_vld1q_z_f32 (__ARM_mve_coerce_f32_ptr(p0, float32_t *), p1)))
-
#define __arm_vld2q(p0) ( \
_Generic( (int (*)[__ARM_mve_typeid(p0)])0, \
int (*)[__ARM_mve_type_int8_t_ptr]: __arm_vld2q_s8 (__ARM_mve_coerce_s8_ptr(p0, int8_t *)), \
int (*)[__ARM_mve_type_uint32_t_ptr]: __arm_vldrwq_gather_shifted_offset_z_u32 (__ARM_mve_coerce_u32_ptr(p0, uint32_t *), p1, p2), \
int (*)[__ARM_mve_type_float32_t_ptr]: __arm_vldrwq_gather_shifted_offset_z_f32 (__ARM_mve_coerce_f32_ptr(p0, float32_t *), p1, p2)))
-#define __arm_vst1q_p(p0,p1,p2) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_int8x16_t]: __arm_vst1q_p_s8 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, int8x16_t), p2), \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_int16x8_t]: __arm_vst1q_p_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, int16x8_t), p2), \
- int (*)[__ARM_mve_type_int32_t_ptr][__ARM_mve_type_int32x4_t]: __arm_vst1q_p_s32 (__ARM_mve_coerce_s32_ptr(p0, int32_t *), __ARM_mve_coerce(__p1, int32x4_t), p2), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint8x16_t]: __arm_vst1q_p_u8 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint8x16_t), p2), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vst1q_p_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2), \
- int (*)[__ARM_mve_type_uint32_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vst1q_p_u32 (__ARM_mve_coerce_u32_ptr(p0, uint32_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2), \
- int (*)[__ARM_mve_type_float16_t_ptr][__ARM_mve_type_float16x8_t]: __arm_vst1q_p_f16 (__ARM_mve_coerce_f16_ptr(p0, float16_t *), __ARM_mve_coerce(__p1, float16x8_t), p2), \
- int (*)[__ARM_mve_type_float32_t_ptr][__ARM_mve_type_float32x4_t]: __arm_vst1q_p_f32 (__ARM_mve_coerce_f32_ptr(p0, float32_t *), __ARM_mve_coerce(__p1, float32x4_t), p2));})
-
#define __arm_vst2q(p0,p1) ({ __typeof(p1) __p1 = (p1); \
_Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_int8x16x2_t]: __arm_vst2q_s8 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, int8x16x2_t)), \
int (*)[__ARM_mve_type_float16_t_ptr][__ARM_mve_type_float16x8x2_t]: __arm_vst2q_f16 (__ARM_mve_coerce_f16_ptr(p0, float16_t *), __ARM_mve_coerce(__p1, float16x8x2_t)), \
int (*)[__ARM_mve_type_float32_t_ptr][__ARM_mve_type_float32x4x2_t]: __arm_vst2q_f32 (__ARM_mve_coerce_f32_ptr(p0, float32_t *), __ARM_mve_coerce(__p1, float32x4x2_t)));})
-#define __arm_vstrhq(p0,p1) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_int16x8_t]: __arm_vstrhq_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, int16x8_t)), \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_int32x4_t]: __arm_vstrhq_s32 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, int32x4_t)), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vstrhq_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vstrhq_u32 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint32x4_t)), \
- int (*)[__ARM_mve_type_float16_t_ptr][__ARM_mve_type_float16x8_t]: __arm_vstrhq_f16 (__ARM_mve_coerce_f16_ptr(p0, float16_t *), __ARM_mve_coerce(__p1, float16x8_t)));})
-
-#define __arm_vstrhq_p(p0,p1,p2) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_int16x8_t]: __arm_vstrhq_p_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, int16x8_t), p2), \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_int32x4_t]: __arm_vstrhq_p_s32 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, int32x4_t), p2), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vstrhq_p_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vstrhq_p_u32 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2), \
- int (*)[__ARM_mve_type_float16_t_ptr][__ARM_mve_type_float16x8_t]: __arm_vstrhq_p_f16 (__ARM_mve_coerce_f16_ptr(p0, float16_t *), __ARM_mve_coerce(__p1, float16x8_t), p2));})
-
#define __arm_vstrhq_scatter_offset_p(p0,p1,p2,p3) ({ __typeof(p1) __p1 = (p1); \
__typeof(p2) __p2 = (p2); \
_Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vstrhq_scatter_shifted_offset_u32 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t)), \
int (*)[__ARM_mve_type_float16_t_ptr][__ARM_mve_type_uint16x8_t][__ARM_mve_type_float16x8_t]: __arm_vstrhq_scatter_shifted_offset_f16 (__ARM_mve_coerce_f16_ptr(p0, float16_t *), __ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, float16x8_t)));})
-#define __arm_vstrwq_p(p0,p1,p2) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int32_t_ptr][__ARM_mve_type_int32x4_t]: __arm_vstrwq_p_s32 (__ARM_mve_coerce_s32_ptr(p0, int32_t *), __ARM_mve_coerce(__p1, int32x4_t), p2), \
- int (*)[__ARM_mve_type_uint32_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vstrwq_p_u32 (__ARM_mve_coerce_u32_ptr(p0, uint32_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2), \
- int (*)[__ARM_mve_type_float32_t_ptr][__ARM_mve_type_float32x4_t]: __arm_vstrwq_p_f32 (__ARM_mve_coerce_f32_ptr(p0, float32_t *), __ARM_mve_coerce(__p1, float32x4_t), p2));})
-
-#define __arm_vstrwq(p0,p1) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int32_t_ptr][__ARM_mve_type_int32x4_t]: __arm_vstrwq_s32 (__ARM_mve_coerce_s32_ptr(p0, int32_t *), __ARM_mve_coerce(__p1, int32x4_t)), \
- int (*)[__ARM_mve_type_uint32_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vstrwq_u32 (__ARM_mve_coerce_u32_ptr(p0, uint32_t *), __ARM_mve_coerce(__p1, uint32x4_t)), \
- int (*)[__ARM_mve_type_float32_t_ptr][__ARM_mve_type_float32x4_t]: __arm_vstrwq_f32 (__ARM_mve_coerce_f32_ptr(p0, float32_t *), __ARM_mve_coerce(__p1, float32x4_t)));})
-
#define __arm_vstrhq_scatter_offset(p0,p1,p2) ({ __typeof(p1) __p1 = (p1); \
__typeof(p2) __p2 = (p2); \
_Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
int (*)[__ARM_mve_type_int32_t_ptr]: __arm_vldrwq_gather_shifted_offset_z_s32 (__ARM_mve_coerce_s32_ptr(__p0, int32_t *), p1, p2), \
int (*)[__ARM_mve_type_uint32_t_ptr]: __arm_vldrwq_gather_shifted_offset_z_u32 (__ARM_mve_coerce_u32_ptr(__p0, uint32_t *), p1, p2));})
-#define __arm_vst1q_p(p0,p1,p2) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_int8x16_t]: __arm_vst1q_p_s8 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, int8x16_t), p2), \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_int16x8_t]: __arm_vst1q_p_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, int16x8_t), p2), \
- int (*)[__ARM_mve_type_int32_t_ptr][__ARM_mve_type_int32x4_t]: __arm_vst1q_p_s32 (__ARM_mve_coerce_s32_ptr(p0, int32_t *), __ARM_mve_coerce(__p1, int32x4_t), p2), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint8x16_t]: __arm_vst1q_p_u8 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint8x16_t), p2), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vst1q_p_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2), \
- int (*)[__ARM_mve_type_uint32_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vst1q_p_u32 (__ARM_mve_coerce_u32_ptr(p0, uint32_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2));})
-
#define __arm_vst2q(p0,p1) ({ __typeof(p1) __p1 = (p1); \
_Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_int8x16x2_t]: __arm_vst2q_s8 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, int8x16x2_t)), \
int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint16x8x2_t]: __arm_vst2q_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint16x8x2_t)), \
int (*)[__ARM_mve_type_uint32_t_ptr][__ARM_mve_type_uint32x4x2_t]: __arm_vst2q_u32 (__ARM_mve_coerce_u32_ptr(p0, uint32_t *), __ARM_mve_coerce(__p1, uint32x4x2_t)));})
-#define __arm_vstrhq(p0,p1) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_int16x8_t]: __arm_vstrhq_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, int16x8_t)), \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_int32x4_t]: __arm_vstrhq_s32 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, int32x4_t)), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vstrhq_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vstrhq_u32 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint32x4_t)));})
-
-#define __arm_vstrhq_p(p0,p1,p2) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_int16x8_t]: __arm_vstrhq_p_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, int16x8_t), p2), \
- int (*)[__ARM_mve_type_int16_t_ptr][__ARM_mve_type_int32x4_t]: __arm_vstrhq_p_s32 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), __ARM_mve_coerce(__p1, int32x4_t), p2), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vstrhq_p_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2), \
- int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vstrhq_p_u32 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2));})
-
#define __arm_vstrhq_scatter_offset_p(p0,p1,p2,p3) ({ __typeof(p1) __p1 = (p1); \
__typeof(p2) __p2 = (p2); \
_Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vstrhq_scatter_shifted_offset_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, uint16x8_t)), \
int (*)[__ARM_mve_type_uint16_t_ptr][__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vstrhq_scatter_shifted_offset_u32 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), __ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t)));})
-
-#define __arm_vstrwq(p0,p1) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int32_t_ptr][__ARM_mve_type_int32x4_t]: __arm_vstrwq_s32 (__ARM_mve_coerce_s32_ptr(p0, int32_t *), __ARM_mve_coerce(__p1, int32x4_t)), \
- int (*)[__ARM_mve_type_uint32_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vstrwq_u32 (__ARM_mve_coerce_u32_ptr(p0, uint32_t *), __ARM_mve_coerce(__p1, uint32x4_t)));})
-
-#define __arm_vstrwq_p(p0,p1,p2) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int32_t_ptr][__ARM_mve_type_int32x4_t]: __arm_vstrwq_p_s32 (__ARM_mve_coerce_s32_ptr(p0, int32_t *), __ARM_mve_coerce(__p1, int32x4_t), p2), \
- int (*)[__ARM_mve_type_uint32_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vstrwq_p_u32 (__ARM_mve_coerce_u32_ptr(p0, uint32_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2));})
-
#define __arm_vstrdq_scatter_base_p(p0,p1,p2,p3) ({ __typeof(p2) __p2 = (p2); \
_Generic( (int (*)[__ARM_mve_typeid(__p2)])0, \
int (*)[__ARM_mve_type_int64x2_t]: __arm_vstrdq_scatter_base_p_s64 (p0, p1, __ARM_mve_coerce(__p2, int64x2_t), p3), \
int (*)[__ARM_mve_type_uint32x4_t]: __arm_vuninitializedq_u32 (), \
int (*)[__ARM_mve_type_uint64x2_t]: __arm_vuninitializedq_u64 ());})
-#define __arm_vld1q_z(p0,p1) ( _Generic( (int (*)[__ARM_mve_typeid(p0)])0, \
- int (*)[__ARM_mve_type_int8_t_ptr]: __arm_vld1q_z_s8 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), p1), \
- int (*)[__ARM_mve_type_int16_t_ptr]: __arm_vld1q_z_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *), p1), \
- int (*)[__ARM_mve_type_int32_t_ptr]: __arm_vld1q_z_s32 (__ARM_mve_coerce_s32_ptr(p0, int32_t *), p1), \
- int (*)[__ARM_mve_type_uint8_t_ptr]: __arm_vld1q_z_u8 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), p1), \
- int (*)[__ARM_mve_type_uint16_t_ptr]: __arm_vld1q_z_u16 (__ARM_mve_coerce_u16_ptr(p0, uint16_t *), p1), \
- int (*)[__ARM_mve_type_uint32_t_ptr]: __arm_vld1q_z_u32 (__ARM_mve_coerce_u32_ptr(p0, uint32_t *), p1)))
-
#define __arm_vld2q(p0) ( _Generic( (int (*)[__ARM_mve_typeid(p0)])0, \
int (*)[__ARM_mve_type_int8_t_ptr]: __arm_vld2q_s8 (__ARM_mve_coerce_s8_ptr(p0, int8_t *)), \
int (*)[__ARM_mve_type_int16_t_ptr]: __arm_vld2q_s16 (__ARM_mve_coerce_s16_ptr(p0, int16_t *)), \
int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrbq_gather_offset_u16(__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \
int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrbq_gather_offset_u32(__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint32x4_t)));})
-#define __arm_vstrbq(p0,p1) ({ __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_int8x16_t]: __arm_vstrbq_s8 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, int8x16_t)), \
- int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_int16x8_t]: __arm_vstrbq_s16 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, int16x8_t)), \
- int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_int32x4_t]: __arm_vstrbq_s32 (__ARM_mve_coerce_s8_ptr(p0, int8_t *), __ARM_mve_coerce(__p1, int32x4_t)), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint8x16_t]: __arm_vstrbq_u8 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint8x16_t)), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vstrbq_u16 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint16x8_t)), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vstrbq_u32 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint32x4_t)));})
-
-#define __arm_vstrbq_p(p0,p1,p2) ({ __typeof(p0) __p0 = (p0); \
- __typeof(p1) __p1 = (p1); \
- _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)])0, \
- int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_int8x16_t]: __arm_vstrbq_p_s8 (__ARM_mve_coerce_s8_ptr(__p0, int8_t *), __ARM_mve_coerce(__p1, int8x16_t), p2), \
- int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_int16x8_t]: __arm_vstrbq_p_s16 (__ARM_mve_coerce_s8_ptr(__p0, int8_t *), __ARM_mve_coerce(__p1, int16x8_t), p2), \
- int (*)[__ARM_mve_type_int8_t_ptr][__ARM_mve_type_int32x4_t]: __arm_vstrbq_p_s32 (__ARM_mve_coerce_s8_ptr(__p0, int8_t *), __ARM_mve_coerce(__p1, int32x4_t), p2), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint8x16_t]: __arm_vstrbq_p_u8 (__ARM_mve_coerce_u8_ptr(__p0, uint8_t *), __ARM_mve_coerce(__p1, uint8x16_t), p2), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vstrbq_p_u16 (__ARM_mve_coerce_u8_ptr(__p0, uint8_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2), \
- int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vstrbq_p_u32 (__ARM_mve_coerce_u8_ptr(__p0, uint8_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2));})
-
#define __arm_vstrdq_scatter_base(p0,p1,p2) ({ __typeof(p2) __p2 = (p2); \
_Generic( (int (*)[__ARM_mve_typeid(__p2)])0, \
int (*)[__ARM_mve_type_int64x2_t]: __arm_vstrdq_scatter_base_s64 (p0, p1, __ARM_mve_coerce(__p2, int64x2_t)), \
(set_attr "type" "mve_move")
(set_attr "length""8")])
-;;
-;; [vstrbq_s vstrbq_u]
-;;
-(define_insn "mve_vstrbq_<supf><mode>"
- [(set (match_operand:<MVE_B_ELEM> 0 "mve_memory_operand" "=Ux")
- (unspec:<MVE_B_ELEM> [(match_operand:MVE_2 1 "s_register_operand" "w")]
- VSTRBQ))
+;; Vector stores
+;; [vstrbq_s8, vstrhq_s16, vstrwq_s32,
+;; vstrbq_u8, vstrhq_u16, vstrwq_u32,
+;; vst1q ]
+(define_insn "@mve_vstrq_<mode>"
+ [(set (match_operand:MVE_VLD_ST 0 "mve_memory_operand" "=Ux")
+ (unspec:MVE_VLD_ST
+ [(match_operand:MVE_VLD_ST 1 "s_register_operand" "w")]
+ VSTRQ))
+ ]
+ "(TARGET_HAVE_MVE && VALID_MVE_SI_MODE (<MODE>mode))
+ || (TARGET_HAVE_MVE_FLOAT && VALID_MVE_SF_MODE (<MODE>mode))"
+{
+ rtx ops[2];
+ int regno = REGNO (operands[1]);
+ ops[1] = gen_rtx_REG (TImode, regno);
+ ops[0] = operands[0];
+ output_asm_insn ("vstr<MVE_elem_ch>.<V_sz_elem>\t%q1, %E0",ops);
+ return "";
+}
+ [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrq_<mode>"))
+ (set_attr "length" "4")])
+
+;; Predicated vector stores
+;; [vstrbq_p_s8, vstrhq_p_s16, vstrwq_p_s32,
+;; vstrbq_p_u8, vstrhq_p_u16, vstrwq_p_u32,
+;; vst1q_p ]
+(define_insn "@mve_vstrq_p_<mode>"
+ [(set (match_operand:MVE_VLD_ST 0 "mve_memory_operand" "=Ux")
+ (unspec:MVE_VLD_ST [
+ (match_operand:MVE_VLD_ST 1 "s_register_operand" "w")
+ (match_operand:<MVE_VPRED> 2 "vpr_register_operand" "Up")
+ (match_dup 0)
+ ] VSTRQ_P))
]
+ "(TARGET_HAVE_MVE && VALID_MVE_SI_MODE (<MODE>mode))
+ || (TARGET_HAVE_MVE_FLOAT && VALID_MVE_SF_MODE (<MODE>mode))"
+{
+ rtx ops[2];
+ int regno = REGNO (operands[1]);
+ ops[1] = gen_rtx_REG (TImode, regno);
+ ops[0] = operands[0];
+ output_asm_insn ("vpst\;vstr<MVE_elem_ch>t.<V_sz_elem>\t%q1, %E0",ops);
+ return "";
+}
+ [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrq_<mode>"))
+ (set_attr "type" "mve_move")
+ (set_attr "length" "8")])
+
+;; Truncating vector stores
+;; [vstrbq_s16, vstrbq_s32, vstrhq_s32,
+;; vstrbq_u16, vstrbq_u32, vstrhq_u32]
+(define_insn "@mve_vstrq_truncate_<mode>"
+ [(set (match_operand:MVE_w_narrow_TYPE 0 "mve_memory_operand" "=Ux")
+ (unspec:MVE_w_narrow_TYPE
+ [(truncate:MVE_w_narrow_TYPE
+ (match_operand:<MVE_wide_n_TYPE> 1 "s_register_operand" "w"))]
+ VSTRQ_TRUNC
+ ))]
"TARGET_HAVE_MVE"
{
- rtx ops[2];
- int regno = REGNO (operands[1]);
- ops[1] = gen_rtx_REG (TImode, regno);
- ops[0] = operands[0];
- output_asm_insn("vstrb.<V_sz_elem>\t%q1, %E0",ops);
- return "";
+ rtx ops[2];
+ int regno = REGNO (operands[1]);
+ ops[1] = gen_rtx_REG (TImode, regno);
+ ops[0] = operands[0];
+ output_asm_insn ("vstr<MVE_elem_ch>.<MVE_wide_n_sz_elem>\t%q1, %E0",ops);
+ return "";
+}
+ [(set (attr "mve_unpredicated_insn")
+ (symbol_ref "CODE_FOR_mve_vstrq_truncate_<mode>"))
+ (set_attr "length" "4")])
+
+;; Predicated truncating vector stores
+;; [vstrbq_p_s16, vstrbq_p_s32, vstrhq_p_s32,
+;; vstrbq_p_u16, vstrbq_p_u32, vstrhq_p_u32]
+(define_insn "@mve_vstrq_p_truncate_<mode>"
+ [(set (match_operand:MVE_w_narrow_TYPE 0 "mve_memory_operand" "=Ux")
+ (unspec:MVE_w_narrow_TYPE [
+ (truncate:MVE_w_narrow_TYPE
+ (match_operand:<MVE_wide_n_TYPE> 1 "s_register_operand" "w"))
+ (match_operand:<MVE_wide_n_VPRED> 2 "vpr_register_operand" "Up")
+ (match_dup 0)
+ ] VSTRQ_TRUNC_P))]
+ "TARGET_HAVE_MVE"
+{
+ rtx ops[2];
+ int regno = REGNO (operands[1]);
+ ops[1] = gen_rtx_REG (TImode, regno);
+ ops[0] = operands[0];
+ output_asm_insn (
+ "vpst\;vstr<MVE_elem_ch>t.<MVE_wide_n_sz_elem>\t%q1, %E0",
+ ops
+ );
+ return "";
}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrbq_<supf><mode>"))
+ [(set (attr "mve_unpredicated_insn")
+ (symbol_ref "CODE_FOR_mve_vstrq_truncate_<mode>"))
+ (set_attr "type" "mve_move")
+ (set_attr "length" "8")])
+
+;; Vector Loads
+;; [vldrbq_s8, vldrhq_s16, vldrwq_s32,
+;; vldrbq_u8, vldrhq_u16, vldrwq_u32,
+;; vld1q ]
+(define_insn "@mve_vldrq_<mode>"
+ [(set (match_operand:MVE_VLD_ST 0 "s_register_operand" "=w")
+ (unspec:MVE_VLD_ST
+ [(match_operand:MVE_VLD_ST 1 "mve_memory_operand" "Ux")]
+ VLDRQ))]
+ "(TARGET_HAVE_MVE && VALID_MVE_SI_MODE (<MODE>mode))
+ || (TARGET_HAVE_MVE_FLOAT && VALID_MVE_SF_MODE (<MODE>mode))"
+{
+ rtx ops[2];
+ int regno = REGNO (operands[0]);
+ ops[0] = gen_rtx_REG (TImode, regno);
+ ops[1] = operands[1];
+ output_asm_insn ("vldr<MVE_elem_ch>.<V_sz_elem>\t%q0, %E1",ops);
+ return "";
+ }
+ [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrq_<mode>"))
(set_attr "length" "4")])
+;; Predicated vector loads
+;; [vldrbq_z_s8, vldrhq_z_s16, vldrwq_z_s32,
+;; vldrbq_z_u8, vldrhq_z_u16, vldrwq_z_u32,
+;; vld1q_z ]
+(define_insn "@mve_vldrq_z_<mode>"
+ [(set (match_operand:MVE_VLD_ST 0 "s_register_operand" "=w")
+ (unspec:MVE_VLD_ST [
+ (match_operand:MVE_VLD_ST 1 "mve_memory_operand" "Ux")
+ (match_operand:<MVE_VPRED> 2 "vpr_register_operand" "Up")
+ ] VLDRQ_Z))]
+ "(TARGET_HAVE_MVE && VALID_MVE_SI_MODE (<MODE>mode))
+ || (TARGET_HAVE_MVE_FLOAT && VALID_MVE_SF_MODE (<MODE>mode))"
+{
+ rtx ops[2];
+ int regno = REGNO (operands[0]);
+ ops[0] = gen_rtx_REG (TImode, regno);
+ ops[1] = operands[1];
+ output_asm_insn ("vpst\;vldr<MVE_elem_ch>t.<V_sz_elem>\t%q0, %E1",ops);
+ return "";
+}
+ [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrq_<mode>"))
+ (set_attr "type" "mve_move")
+ (set_attr "length" "8")])
+
+;; Extending vector loads
+;; [vldrbq_s16, vldrbq_s32, vldrhq_s32,
+;; vldrbq_u16, vldrbq_u32, vldrhq_u32]
+(define_insn "@mve_vldrq_extend_<mode><US>"
+ [(set (match_operand:<MVE_wide_n_TYPE> 0 "s_register_operand" "=w")
+ (unspec:<MVE_wide_n_TYPE>
+ [(SE:<MVE_wide_n_TYPE>
+ (match_operand:MVE_w_narrow_TYPE 1 "mve_memory_operand" "Ux"))]
+ VLDRQ_EXT))]
+ "TARGET_HAVE_MVE"
+{
+ rtx ops[2];
+ int regno = REGNO (operands[0]);
+ ops[0] = gen_rtx_REG (TImode, regno);
+ ops[1] = operands[1];
+ output_asm_insn ("vldr<MVE_elem_ch>.<US><MVE_wide_n_sz_elem>\t%q0, %E1",ops);
+ return "";
+}
+ [(set (attr "mve_unpredicated_insn")
+ (symbol_ref "CODE_FOR_mve_vldrq_extend_<mode><US>"))
+ (set_attr "length" "4")])
+
+;; Predicated extending vector loads
+;; [vldrbq_z_s16, vldrbq_z_s32, vldrhq_z_s32,
+;; vldrbq_z_u16, vldrbq_z_u32, vldrhq_z_u32]
+(define_insn "@mve_vldrq_z_extend_<mode><US>"
+ [(set (match_operand:<MVE_wide_n_TYPE> 0 "s_register_operand" "=w")
+ (unspec:<MVE_wide_n_TYPE> [
+ (SE:<MVE_wide_n_TYPE>
+ (match_operand:MVE_w_narrow_TYPE 1 "mve_memory_operand" "Ux"))
+ (match_operand:<MVE_wide_n_VPRED> 2 "vpr_register_operand" "Up")
+ ] VLDRQ_EXT_Z))]
+ "TARGET_HAVE_MVE"
+{
+ rtx ops[2];
+ int regno = REGNO (operands[0]);
+ ops[0] = gen_rtx_REG (TImode, regno);
+ ops[1] = operands[1];
+ output_asm_insn (
+ "vpst\;vldr<MVE_elem_ch>t.<US><MVE_wide_n_sz_elem>\t%q0, %E1",
+ ops
+ );
+ return "";
+}
+ [(set (attr "mve_unpredicated_insn")
+ (symbol_ref "CODE_FOR_mve_vldrq_extend_<mode><US>"))
+ (set_attr "type" "mve_move")
+ (set_attr "length" "8")])
+
;;
;; [vstrbq_scatter_offset_s vstrbq_scatter_offset_u]
;;
[(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrbq_gather_offset_<supf><mode>"))
(set_attr "length" "4")])
-;;
-;; [vldrbq_s vldrbq_u]
-;;
-(define_insn "mve_vldrbq_<supf><mode>"
- [(set (match_operand:MVE_2 0 "s_register_operand" "=w")
- (unspec:MVE_2 [(match_operand:<MVE_B_ELEM> 1 "mve_memory_operand" "Ux")]
- VLDRBQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[2];
- int regno = REGNO (operands[0]);
- ops[0] = gen_rtx_REG (TImode, regno);
- ops[1] = operands[1];
- if (<V_sz_elem> == 8)
- output_asm_insn ("vldrb.<V_sz_elem>\t%q0, %E1",ops);
- else
- output_asm_insn ("vldrb.<supf><V_sz_elem>\t%q0, %E1",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrbq_<supf><mode>"))
- (set_attr "length" "4")])
-
;;
;; [vldrwq_gather_base_s vldrwq_gather_base_u]
;;
[(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrwq_scatter_base_<supf>v4si"))
(set_attr "length" "8")])
-(define_insn "mve_vstrbq_p_<supf><mode>"
- [(set (match_operand:<MVE_B_ELEM> 0 "mve_memory_operand" "=Ux")
- (unspec:<MVE_B_ELEM>
- [(match_operand:MVE_2 1 "s_register_operand" "w")
- (match_operand:<MVE_VPRED> 2 "vpr_register_operand" "Up")
- (match_dup 0)]
- VSTRBQ))]
- "TARGET_HAVE_MVE"
-{
- rtx ops[2];
- int regno = REGNO (operands[1]);
- ops[1] = gen_rtx_REG (TImode, regno);
- ops[0] = operands[0];
- output_asm_insn ("vpst\;vstrbt.<V_sz_elem>\t%q1, %E0",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrbq_<supf><mode>"))
- (set_attr "length" "8")])
-
;;
;; [vldrbq_gather_offset_z_s vldrbq_gather_offset_z_u]
;;
[(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrbq_gather_offset_<supf><mode>"))
(set_attr "length" "8")])
-;;
-;; [vldrbq_z_s vldrbq_z_u]
-;;
-(define_insn "mve_vldrbq_z_<supf><mode>"
- [(set (match_operand:MVE_2 0 "s_register_operand" "=w")
- (unspec:MVE_2 [(match_operand:<MVE_B_ELEM> 1 "mve_memory_operand" "Ux")
- (match_operand:<MVE_VPRED> 2 "vpr_register_operand" "Up")]
- VLDRBQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[2];
- int regno = REGNO (operands[0]);
- ops[0] = gen_rtx_REG (TImode, regno);
- ops[1] = operands[1];
- if (<V_sz_elem> == 8)
- output_asm_insn ("vpst\;vldrbt.<V_sz_elem>\t%q0, %E1",ops);
- else
- output_asm_insn ("vpst\;vldrbt.<supf><V_sz_elem>\t%q0, %E1",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrbq_<supf><mode>"))
- (set_attr "length" "8")])
-
;;
;; [vldrwq_gather_base_z_s vldrwq_gather_base_z_u]
;;
[(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrwq_gather_base_<supf>v4si"))
(set_attr "length" "8")])
-;;
-;; [vldrhq_f]
-;;
-(define_insn "mve_vldrhq_fv8hf"
- [(set (match_operand:V8HF 0 "s_register_operand" "=w")
- (unspec:V8HF [(match_operand:V8HI 1 "mve_memory_operand" "Ux")]
- VLDRHQ_F))
- ]
- "TARGET_HAVE_MVE && TARGET_HAVE_MVE_FLOAT"
-{
- rtx ops[2];
- int regno = REGNO (operands[0]);
- ops[0] = gen_rtx_REG (TImode, regno);
- ops[1] = operands[1];
- output_asm_insn ("vldrh.16\t%q0, %E1",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrhq_fv8hf"))
- (set_attr "length" "4")])
-
;;
;; [vldrhq_gather_offset_s vldrhq_gather_offset_u]
;;
[(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrhq_gather_shifted_offset_<supf><mode>"))
(set_attr "length" "8")])
-;;
-;; [vldrhq_s, vldrhq_u]
-;;
-(define_insn "mve_vldrhq_<supf><mode>"
- [(set (match_operand:MVE_5 0 "s_register_operand" "=w")
- (unspec:MVE_5 [(match_operand:<MVE_H_ELEM> 1 "mve_memory_operand" "Ux")]
- VLDRHQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[2];
- int regno = REGNO (operands[0]);
- ops[0] = gen_rtx_REG (TImode, regno);
- ops[1] = operands[1];
- if (<V_sz_elem> == 16)
- output_asm_insn ("vldrh.16\t%q0, %E1",ops);
- else
- output_asm_insn ("vldrh.<supf><V_sz_elem>\t%q0, %E1",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrhq_<supf><mode>"))
- (set_attr "length" "4")])
-
-;;
-;; [vldrhq_z_f]
-;;
-(define_insn "mve_vldrhq_z_fv8hf"
- [(set (match_operand:V8HF 0 "s_register_operand" "=w")
- (unspec:V8HF [(match_operand:V8HI 1 "mve_memory_operand" "Ux")
- (match_operand:<MVE_VPRED> 2 "vpr_register_operand" "Up")]
- VLDRHQ_F))
- ]
- "TARGET_HAVE_MVE && TARGET_HAVE_MVE_FLOAT"
-{
- rtx ops[2];
- int regno = REGNO (operands[0]);
- ops[0] = gen_rtx_REG (TImode, regno);
- ops[1] = operands[1];
- output_asm_insn ("vpst\;vldrht.16\t%q0, %E1",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrhq_fv8hf"))
- (set_attr "length" "8")])
-
-;;
-;; [vldrhq_z_s vldrhq_z_u]
-;;
-(define_insn "mve_vldrhq_z_<supf><mode>"
- [(set (match_operand:MVE_5 0 "s_register_operand" "=w")
- (unspec:MVE_5 [(match_operand:<MVE_H_ELEM> 1 "mve_memory_operand" "Ux")
- (match_operand:<MVE_VPRED> 2 "vpr_register_operand" "Up")]
- VLDRHQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[2];
- int regno = REGNO (operands[0]);
- ops[0] = gen_rtx_REG (TImode, regno);
- ops[1] = operands[1];
- if (<V_sz_elem> == 16)
- output_asm_insn ("vpst\;vldrht.16\t%q0, %E1",ops);
- else
- output_asm_insn ("vpst\;vldrht.<supf><V_sz_elem>\t%q0, %E1",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrhq_<supf><mode>"))
- (set_attr "length" "8")])
-
-;;
-;; [vldrwq_f]
-;;
-(define_insn "mve_vldrwq_fv4sf"
- [(set (match_operand:V4SF 0 "s_register_operand" "=w")
- (unspec:V4SF [(match_operand:V4SI 1 "mve_memory_operand" "Ux")]
- VLDRWQ_F))
- ]
- "TARGET_HAVE_MVE && TARGET_HAVE_MVE_FLOAT"
-{
- rtx ops[2];
- int regno = REGNO (operands[0]);
- ops[0] = gen_rtx_REG (TImode, regno);
- ops[1] = operands[1];
- output_asm_insn ("vldrw.32\t%q0, %E1",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrwq_fv4sf"))
- (set_attr "length" "4")])
-
-;;
-;; [vldrwq_s vldrwq_u]
-;;
-(define_insn "mve_vldrwq_<supf>v4si"
- [(set (match_operand:V4SI 0 "s_register_operand" "=w")
- (unspec:V4SI [(match_operand:V4SI 1 "mve_memory_operand" "Ux")]
- VLDRWQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[2];
- int regno = REGNO (operands[0]);
- ops[0] = gen_rtx_REG (TImode, regno);
- ops[1] = operands[1];
- output_asm_insn ("vldrw.32\t%q0, %E1",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrwq_<supf>v4si"))
- (set_attr "length" "4")])
-
-;;
-;; [vldrwq_z_f]
-;;
-(define_insn "mve_vldrwq_z_fv4sf"
- [(set (match_operand:V4SF 0 "s_register_operand" "=w")
- (unspec:V4SF [(match_operand:V4SI 1 "mve_memory_operand" "Ux")
- (match_operand:V4BI 2 "vpr_register_operand" "Up")]
- VLDRWQ_F))
- ]
- "TARGET_HAVE_MVE && TARGET_HAVE_MVE_FLOAT"
-{
- rtx ops[2];
- int regno = REGNO (operands[0]);
- ops[0] = gen_rtx_REG (TImode, regno);
- ops[1] = operands[1];
- output_asm_insn ("vpst\;vldrwt.32\t%q0, %E1",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrwq_fv4sf"))
- (set_attr "length" "8")])
-
-;;
-;; [vldrwq_z_s vldrwq_z_u]
-;;
-(define_insn "mve_vldrwq_z_<supf>v4si"
- [(set (match_operand:V4SI 0 "s_register_operand" "=w")
- (unspec:V4SI [(match_operand:V4SI 1 "mve_memory_operand" "Ux")
- (match_operand:V4BI 2 "vpr_register_operand" "Up")]
- VLDRWQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[2];
- int regno = REGNO (operands[0]);
- ops[0] = gen_rtx_REG (TImode, regno);
- ops[1] = operands[1];
- output_asm_insn ("vpst\;vldrwt.32\t%q0, %E1",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrwq_<supf>v4si"))
- (set_attr "length" "8")])
-
-(define_expand "@mve_vld1q_f<mode>"
- [(match_operand:MVE_0 0 "s_register_operand")
- (unspec:MVE_0 [(match_operand:<MVE_CNVT> 1 "mve_memory_operand")] VLD1Q_F)
- ]
- "TARGET_HAVE_MVE || TARGET_HAVE_MVE_FLOAT"
-{
- emit_insn (gen_mve_vldr<V_sz_elem1>q_f<mode>(operands[0],operands[1]));
- DONE;
-})
-
-(define_expand "@mve_vld1q_<supf><mode>"
- [(match_operand:MVE_2 0 "s_register_operand")
- (unspec:MVE_2 [(match_operand:MVE_2 1 "mve_memory_operand")] VLD1Q)
- ]
- "TARGET_HAVE_MVE"
-{
- emit_insn (gen_mve_vldr<V_sz_elem1>q_<supf><mode>(operands[0],operands[1]));
- DONE;
-})
-
;;
;; [vldrdq_gather_base_s vldrdq_gather_base_u]
;;
[(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vldrwq_gather_shifted_offset_<supf>v4si"))
(set_attr "length" "8")])
-;;
-;; [vstrhq_f]
-;;
-(define_insn "mve_vstrhq_fv8hf"
- [(set (match_operand:V8HI 0 "mve_memory_operand" "=Ux")
- (unspec:V8HI [(match_operand:V8HF 1 "s_register_operand" "w")]
- VSTRHQ_F))
- ]
- "TARGET_HAVE_MVE && TARGET_HAVE_MVE_FLOAT"
-{
- rtx ops[2];
- int regno = REGNO (operands[1]);
- ops[1] = gen_rtx_REG (TImode, regno);
- ops[0] = operands[0];
- output_asm_insn ("vstrh.16\t%q1, %E0",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrhq_fv8hf"))
- (set_attr "length" "4")])
-
-;;
-;; [vstrhq_p_f]
-;;
-(define_insn "mve_vstrhq_p_fv8hf"
- [(set (match_operand:V8HI 0 "mve_memory_operand" "=Ux")
- (unspec:V8HI
- [(match_operand:V8HF 1 "s_register_operand" "w")
- (match_operand:V8BI 2 "vpr_register_operand" "Up")
- (match_dup 0)]
- VSTRHQ_F))]
- "TARGET_HAVE_MVE && TARGET_HAVE_MVE_FLOAT"
-{
- rtx ops[2];
- int regno = REGNO (operands[1]);
- ops[1] = gen_rtx_REG (TImode, regno);
- ops[0] = operands[0];
- output_asm_insn ("vpst\;vstrht.16\t%q1, %E0",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrhq_fv8hf"))
- (set_attr "length" "8")])
-
-;;
-;; [vstrhq_p_s vstrhq_p_u]
-;;
-(define_insn "mve_vstrhq_p_<supf><mode>"
- [(set (match_operand:<MVE_H_ELEM> 0 "mve_memory_operand" "=Ux")
- (unspec:<MVE_H_ELEM>
- [(match_operand:MVE_5 1 "s_register_operand" "w")
- (match_operand:<MVE_VPRED> 2 "vpr_register_operand" "Up")
- (match_dup 0)]
- VSTRHQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[2];
- int regno = REGNO (operands[1]);
- ops[1] = gen_rtx_REG (TImode, regno);
- ops[0] = operands[0];
- output_asm_insn ("vpst\;vstrht.<V_sz_elem>\t%q1, %E0",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrhq_<supf><mode>"))
- (set_attr "length" "8")])
-
;;
;; [vstrhq_scatter_offset_p_s vstrhq_scatter_offset_p_u]
;;
[(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrhq_scatter_shifted_offset_<supf><mode>_insn"))
(set_attr "length" "4")])
-;;
-;; [vstrhq_s, vstrhq_u]
-;;
-(define_insn "mve_vstrhq_<supf><mode>"
- [(set (match_operand:<MVE_H_ELEM> 0 "mve_memory_operand" "=Ux")
- (unspec:<MVE_H_ELEM> [(match_operand:MVE_5 1 "s_register_operand" "w")]
- VSTRHQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[2];
- int regno = REGNO (operands[1]);
- ops[1] = gen_rtx_REG (TImode, regno);
- ops[0] = operands[0];
- output_asm_insn ("vstrh.<V_sz_elem>\t%q1, %E0",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrhq_<supf><mode>"))
- (set_attr "length" "4")])
-
-;;
-;; [vstrwq_f]
-;;
-(define_insn "mve_vstrwq_fv4sf"
- [(set (match_operand:V4SI 0 "mve_memory_operand" "=Ux")
- (unspec:V4SI [(match_operand:V4SF 1 "s_register_operand" "w")]
- VSTRWQ_F))
- ]
- "TARGET_HAVE_MVE && TARGET_HAVE_MVE_FLOAT"
-{
- rtx ops[2];
- int regno = REGNO (operands[1]);
- ops[1] = gen_rtx_REG (TImode, regno);
- ops[0] = operands[0];
- output_asm_insn ("vstrw.32\t%q1, %E0",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrwq_fv4sf"))
- (set_attr "length" "4")])
-
-;;
-;; [vstrwq_p_f]
-;;
-(define_insn "mve_vstrwq_p_fv4sf"
- [(set (match_operand:V4SI 0 "mve_memory_operand" "=Ux")
- (unspec:V4SI
- [(match_operand:V4SF 1 "s_register_operand" "w")
- (match_operand:V4BI 2 "vpr_register_operand" "Up")
- (match_dup 0)]
- VSTRWQ_F))]
- "TARGET_HAVE_MVE && TARGET_HAVE_MVE_FLOAT"
-{
- rtx ops[2];
- int regno = REGNO (operands[1]);
- ops[1] = gen_rtx_REG (TImode, regno);
- ops[0] = operands[0];
- output_asm_insn ("vpst\;vstrwt.32\t%q1, %E0",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrwq_fv4sf"))
- (set_attr "length" "8")])
-
-;;
-;; [vstrwq_p_s vstrwq_p_u]
-;;
-(define_insn "mve_vstrwq_p_<supf>v4si"
- [(set (match_operand:V4SI 0 "mve_memory_operand" "=Ux")
- (unspec:V4SI
- [(match_operand:V4SI 1 "s_register_operand" "w")
- (match_operand:V4BI 2 "vpr_register_operand" "Up")
- (match_dup 0)]
- VSTRWQ))]
- "TARGET_HAVE_MVE"
-{
- rtx ops[2];
- int regno = REGNO (operands[1]);
- ops[1] = gen_rtx_REG (TImode, regno);
- ops[0] = operands[0];
- output_asm_insn ("vpst\;vstrwt.32\t%q1, %E0",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrwq_<supf>v4si"))
- (set_attr "length" "8")])
-
-;;
-;; [vstrwq_s vstrwq_u]
-;;
-(define_insn "mve_vstrwq_<supf>v4si"
- [(set (match_operand:V4SI 0 "mve_memory_operand" "=Ux")
- (unspec:V4SI [(match_operand:V4SI 1 "s_register_operand" "w")]
- VSTRWQ))
- ]
- "TARGET_HAVE_MVE"
-{
- rtx ops[2];
- int regno = REGNO (operands[1]);
- ops[1] = gen_rtx_REG (TImode, regno);
- ops[0] = operands[0];
- output_asm_insn ("vstrw.32\t%q1, %E0",ops);
- return "";
-}
- [(set (attr "mve_unpredicated_insn") (symbol_ref "CODE_FOR_mve_vstrwq_<supf>v4si"))
- (set_attr "length" "4")])
-
-(define_expand "@mve_vst1q_f<mode>"
- [(match_operand:<MVE_CNVT> 0 "mve_memory_operand")
- (unspec:<MVE_CNVT> [(match_operand:MVE_0 1 "s_register_operand")] VST1Q_F)
- ]
- "TARGET_HAVE_MVE || TARGET_HAVE_MVE_FLOAT"
-{
- emit_insn (gen_mve_vstr<V_sz_elem1>q_f<mode>(operands[0],operands[1]));
- DONE;
-})
-
-(define_expand "@mve_vst1q_<supf><mode>"
- [(match_operand:MVE_2 0 "mve_memory_operand")
- (unspec:MVE_2 [(match_operand:MVE_2 1 "s_register_operand")] VST1Q)
- ]
- "TARGET_HAVE_MVE"
-{
- emit_insn (gen_mve_vstr<V_sz_elem1>q_<supf><mode>(operands[0],operands[1]));
- DONE;
-})
-
;;
;; [vstrdq_scatter_base_p_s vstrdq_scatter_base_p_u]
;;
projects / thirdparty / gcc.git / commitdiff
