RE: [PATCH][GCC-10 backport] arm: Remove coercion from scalar argument to vmin & vmax intrinsics.

[Kyrylo Tkachov via Gcc-patches]

> -----Original Message-----
> From: Srinath Parvathaneni <srinath.parvathan...@arm.com>
> Sent: 06 October 2020 14:37
> To: gcc-patches@gcc.gnu.org
> Cc: Kyrylo Tkachov <kyrylo.tkac...@arm.com>
> Subject: [PATCH][GCC-10 backport] arm: Remove coercion from scalar
> argument to vmin & vmax intrinsics.
> 
> Hello,
> 
> Straight backport of Joe's patch with no changes.
> 
> This patch fixes an issue with vmin* and vmax* intrinsics which accept
> a scalar argument. Previously when the scalar was of different width
> to the vector elements this would generate __ARM_undef. This change
> allows the scalar argument to be implicitly converted to the correct
> width. Also tidied up the relevant unit tests, some of which would
> have passed even if only one of two or three intrinsic calls had
> compiled correctly.
> 
> Bootstrapped and tested on arm-none-eabi, gcc and CMSIS_DSP
> testsuites are clean. OK for trunk?

Ok.
Thanks,
Kyrill

> 
> Thanks,
> Joe
> 
> gcc/ChangeLog:
> 
> 2020-08-10  Joe Ramsay  <joe.ram...@arm.com>
> 
>       * config/arm/arm_mve.h (__arm_vmaxnmavq): Remove coercion of
> scalar
>       argument.
>       (__arm_vmaxnmvq): Likewise.
>       (__arm_vminnmavq): Likewise.
>       (__arm_vminnmvq): Likewise.
>       (__arm_vmaxnmavq_p): Likewise.
>       (__arm_vmaxnmvq_p): Likewise (and delete duplicate definition).
>       (__arm_vminnmavq_p): Likewise.
>       (__arm_vminnmvq_p): Likewise.
>       (__arm_vmaxavq): Likewise.
>       (__arm_vmaxavq_p): Likewise.
>       (__arm_vmaxvq): Likewise.
>       (__arm_vmaxvq_p): Likewise.
>       (__arm_vminavq): Likewise.
>       (__arm_vminavq_p): Likewise.
>       (__arm_vminvq): Likewise.
>       (__arm_vminvq_p): Likewise.
> 
> gcc/testsuite/ChangeLog:
> 
>       * gcc.target/arm/mve/intrinsics/vmaxavq_p_s16.c: Add test for
> mismatched
>       width of scalar argument.
>       * gcc.target/arm/mve/intrinsics/vmaxavq_p_s32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxavq_p_s8.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxavq_s16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxavq_s32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxavq_s8.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxnmavq_f16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxnmavq_f32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxnmavq_p_f16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxnmavq_p_f32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxnmvq_f16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxnmvq_f32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxnmvq_p_f16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxnmvq_p_f32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxvq_p_s16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxvq_p_s32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxvq_p_s8.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxvq_p_u16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxvq_p_u32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxvq_p_u8.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxvq_s16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxvq_s32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxvq_s8.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxvq_u16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxvq_u32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vmaxvq_u8.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminavq_p_s16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminavq_p_s32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminavq_p_s8.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminavq_s16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminavq_s32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminavq_s8.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminnmavq_f16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminnmavq_f32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminnmavq_p_f16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminnmavq_p_f32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminnmvq_f16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminnmvq_f32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminnmvq_p_f16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminnmvq_p_f32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminvq_p_s16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminvq_p_s32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminvq_p_s8.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminvq_p_u16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminvq_p_u32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminvq_p_u8.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminvq_s16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminvq_s32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminvq_s8.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminvq_u16.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminvq_u32.c: Likewise.
>       * gcc.target/arm/mve/intrinsics/vminvq_u8.c: Likewise.
> 
> (cherry picked from commit 251950d899bc3c18b5775fe9fe20bebbdc8d15cb)
> 
> 
> ###############     Attachment also inlined for ease of reply
> ###############
> 
> 
> diff --git a/gcc/config/arm/arm_mve.h b/gcc/config/arm/arm_mve.h
> index
> 99cff41cccbe22f5f6bfe8db513092830885976c..26c83c7efefa34f17d8f8d30e
> 5a3141f680041df 100644
> --- a/gcc/config/arm/arm_mve.h
> +++ b/gcc/config/arm/arm_mve.h
> @@ -41682,16 +41682,16 @@ extern void *__ARM_undef;
>  #define __arm_vmaxavq(p0,p1) ({ __typeof(p0) __p0 = (p0); \
>    __typeof(p1) __p1 = (p1); \
>    _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)])0,
> \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vmaxavq_s8 (__ARM_mve_coerce(__p0, uint8_t),
> __ARM_mve_coerce(__p1, int8x16_t)), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vmaxavq_s16 (__ARM_mve_coerce(__p0, uint16_t),
> __ARM_mve_coerce(__p1, int16x8_t)), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vmaxavq_s32 (__ARM_mve_coerce(__p0, uint32_t),
> __ARM_mve_coerce(__p1, int32x4_t)));})
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vmaxavq_s8 (__p0, __ARM_mve_coerce(__p1, int8x16_t)), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vmaxavq_s16 (__p0, __ARM_mve_coerce(__p1, int16x8_t)), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vmaxavq_s32 (__p0, __ARM_mve_coerce(__p1, int32x4_t)));})
> 
>  #define __arm_vmaxavq_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_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vmaxavq_p_s8 (__ARM_mve_coerce(__p0, uint8_t),
> __ARM_mve_coerce(__p1, int8x16_t), p2), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vmaxavq_p_s16 (__ARM_mve_coerce(__p0, uint16_t),
> __ARM_mve_coerce(__p1, int16x8_t), p2), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vmaxavq_p_s32 (__ARM_mve_coerce(__p0, uint32_t),
> __ARM_mve_coerce(__p1, int32x4_t), p2));})
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vmaxavq_p_s8 (__p0, __ARM_mve_coerce(__p1, int8x16_t), p2), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vmaxavq_p_s16 (__p0, __ARM_mve_coerce(__p1, int16x8_t), p2), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vmaxavq_p_s32 (__p0, __ARM_mve_coerce(__p1, int32x4_t), p2));})
> 
>  #define __arm_vmaxq_x(p1,p2,p3) ({ __typeof(p1) __p1 = (p1); \
>    __typeof(p2) __p2 = (p2); \
> @@ -41706,36 +41706,36 @@ extern void *__ARM_undef;
>  #define __arm_vmaxvq(p0,p1) ({ __typeof(p0) __p0 = (p0); \
>    __typeof(p1) __p1 = (p1); \
>    _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)])0,
> \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vmaxvq_s8 (__ARM_mve_coerce(__p0, int8_t),
> __ARM_mve_coerce(__p1, int8x16_t)), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vmaxvq_s16 (__ARM_mve_coerce(__p0, int16_t),
> __ARM_mve_coerce(__p1, int16x8_t)), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vmaxvq_s32 (__ARM_mve_coerce(__p0, int32_t),
> __ARM_mve_coerce(__p1, int32x4_t)), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint8x16_t]:
> __arm_vmaxvq_u8 (__ARM_mve_coerce(__p0, uint8_t),
> __ARM_mve_coerce(__p1, uint8x16_t)), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint16x8_t]:
> __arm_vmaxvq_u16 (__ARM_mve_coerce(__p0, uint16_t),
> __ARM_mve_coerce(__p1, uint16x8_t)), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint32x4_t]:
> __arm_vmaxvq_u32 (__ARM_mve_coerce(__p0, uint32_t),
> __ARM_mve_coerce(__p1, uint32x4_t)));})
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vmaxvq_s8 (__p0, __ARM_mve_coerce(__p1, int8x16_t)), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vmaxvq_s16 (__p0, __ARM_mve_coerce(__p1, int16x8_t)), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vmaxvq_s32 (__p0, __ARM_mve_coerce(__p1, int32x4_t)), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint8x16_t]:
> __arm_vmaxvq_u8 (__p0, __ARM_mve_coerce(__p1, uint8x16_t)), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint16x8_t]:
> __arm_vmaxvq_u16 (__p0, __ARM_mve_coerce(__p1, uint16x8_t)), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint32x4_t]:
> __arm_vmaxvq_u32 (__p0,__ARM_mve_coerce(__p1, uint32x4_t)));})
> 
>  #define __arm_vmaxvq_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_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vmaxvq_p_s8 (__ARM_mve_coerce(__p0, int8_t),
> __ARM_mve_coerce(__p1, int8x16_t), p2), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vmaxvq_p_s16 (__ARM_mve_coerce(__p0, int16_t),
> __ARM_mve_coerce(__p1, int16x8_t), p2), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vmaxvq_p_s32 (__ARM_mve_coerce(__p0, int32_t),
> __ARM_mve_coerce(__p1, int32x4_t), p2), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint8x16_t]:
> __arm_vmaxvq_p_u8 (__ARM_mve_coerce(__p0, uint8_t),
> __ARM_mve_coerce(__p1, uint8x16_t), p2), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint16x8_t]:
> __arm_vmaxvq_p_u16 (__ARM_mve_coerce(__p0, uint16_t),
> __ARM_mve_coerce(__p1, uint16x8_t), p2), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint32x4_t]:
> __arm_vmaxvq_p_u32 (__ARM_mve_coerce(__p0, uint32_t),
> __ARM_mve_coerce(__p1, uint32x4_t), p2));})
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vmaxvq_p_s8 (__p0, __ARM_mve_coerce(__p1, int8x16_t), p2), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vmaxvq_p_s16 (__p0, __ARM_mve_coerce(__p1, int16x8_t), p2), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vmaxvq_p_s32 (__p0, __ARM_mve_coerce(__p1, int32x4_t), p2), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint8x16_t]:
> __arm_vmaxvq_p_u8 (__p0, __ARM_mve_coerce(__p1, uint8x16_t), p2), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint16x8_t]:
> __arm_vmaxvq_p_u16 (__p0, __ARM_mve_coerce(__p1, uint16x8_t), p2), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint32x4_t]:
> __arm_vmaxvq_p_u32 (__p0, __ARM_mve_coerce(__p1, uint32x4_t), p2));})
> 
>  #define __arm_vminavq(p0,p1) ({ __typeof(p0) __p0 = (p0); \
>    __typeof(p1) __p1 = (p1); \
>    _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)])0,
> \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vminavq_s8 (__ARM_mve_coerce(__p0, uint8_t),
> __ARM_mve_coerce(__p1, int8x16_t)), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vminavq_s16 (__ARM_mve_coerce(__p0, uint16_t),
> __ARM_mve_coerce(__p1, int16x8_t)), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vminavq_s32 (__ARM_mve_coerce(__p0, uint32_t),
> __ARM_mve_coerce(__p1, int32x4_t)));})
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vminavq_s8 (__p0, __ARM_mve_coerce(__p1, int8x16_t)), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vminavq_s16 (__p0, __ARM_mve_coerce(__p1, int16x8_t)), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vminavq_s32 (__p0, __ARM_mve_coerce(__p1, int32x4_t)));})
> 
>  #define __arm_vminavq_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_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vminavq_p_s8 (__ARM_mve_coerce(__p0, uint8_t),
> __ARM_mve_coerce(__p1, int8x16_t), p2), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vminavq_p_s16 (__ARM_mve_coerce(__p0, uint16_t),
> __ARM_mve_coerce(__p1, int16x8_t), p2), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vminavq_p_s32 (__ARM_mve_coerce(__p0, uint32_t),
> __ARM_mve_coerce(__p1, int32x4_t), p2));})
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vminavq_p_s8 (__p0, __ARM_mve_coerce(__p1, int8x16_t), p2), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vminavq_p_s16 (__p0, __ARM_mve_coerce(__p1, int16x8_t), p2), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vminavq_p_s32 (__p0, __ARM_mve_coerce(__p1, int32x4_t), p2));})
> 
>  #define __arm_vminq_x(p1,p2,p3) ({ __typeof(p1) __p1 = (p1); \
>    __typeof(p2) __p2 = (p2); \
> @@ -41750,22 +41750,22 @@ extern void *__ARM_undef;
>  #define __arm_vminvq(p0,p1) ({ __typeof(p0) __p0 = (p0); \
>    __typeof(p1) __p1 = (p1); \
>    _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)])0,
> \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vminvq_s8 (__ARM_mve_coerce(__p0, int8_t),
> __ARM_mve_coerce(__p1, int8x16_t)), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vminvq_s16 (__ARM_mve_coerce(__p0, int16_t),
> __ARM_mve_coerce(__p1, int16x8_t)), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vminvq_s32 (__ARM_mve_coerce(__p0, int32_t),
> __ARM_mve_coerce(__p1, int32x4_t)), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint8x16_t]:
> __arm_vminvq_u8 (__ARM_mve_coerce(__p0, uint8_t),
> __ARM_mve_coerce(__p1, uint8x16_t)), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint16x8_t]:
> __arm_vminvq_u16 (__ARM_mve_coerce(__p0, uint16_t),
> __ARM_mve_coerce(__p1, uint16x8_t)), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint32x4_t]:
> __arm_vminvq_u32 (__ARM_mve_coerce(__p0, uint32_t),
> __ARM_mve_coerce(__p1, uint32x4_t)));})
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vminvq_s8 (__p0, __ARM_mve_coerce(__p1, int8x16_t)), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vminvq_s16 (__p0, __ARM_mve_coerce(__p1, int16x8_t)), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vminvq_s32 (__p0, __ARM_mve_coerce(__p1, int32x4_t)), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint8x16_t]:
> __arm_vminvq_u8 (__p0, __ARM_mve_coerce(__p1, uint8x16_t)), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint16x8_t]:
> __arm_vminvq_u16 (__p0, __ARM_mve_coerce(__p1, uint16x8_t)), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint32x4_t]:
> __arm_vminvq_u32 (__p0, __ARM_mve_coerce(__p1, uint32x4_t)));})
> 
>  #define __arm_vminvq_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_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vminvq_p_s8 (__ARM_mve_coerce(__p0, int8_t),
> __ARM_mve_coerce(__p1, int8x16_t), p2), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vminvq_p_s16 (__ARM_mve_coerce(__p0, int16_t),
> __ARM_mve_coerce(__p1, int16x8_t), p2), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vminvq_p_s32 (__ARM_mve_coerce(__p0, int32_t),
> __ARM_mve_coerce(__p1, int32x4_t), p2), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint8x16_t]:
> __arm_vminvq_p_u8 (__ARM_mve_coerce(__p0, uint8_t),
> __ARM_mve_coerce(__p1, uint8x16_t), p2), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint16x8_t]:
> __arm_vminvq_p_u16 (__ARM_mve_coerce(__p0, uint16_t),
> __ARM_mve_coerce(__p1, uint16x8_t), p2), \
> -  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint32x4_t]:
> __arm_vminvq_p_u32 (__ARM_mve_coerce(__p0, uint32_t),
> __ARM_mve_coerce(__p1, uint32x4_t), p2));})
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int8x16_t]:
> __arm_vminvq_p_s8 (__p0, __ARM_mve_coerce(__p1, int8x16_t), p2), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t]:
> __arm_vminvq_p_s16 (__p0, __ARM_mve_coerce(__p1, int16x8_t), p2), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t]:
> __arm_vminvq_p_s32 (__p0, __ARM_mve_coerce(__p1, int32x4_t), p2), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint8x16_t]:
> __arm_vminvq_p_u8 (__p0, __ARM_mve_coerce(__p1, uint8x16_t), p2), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint16x8_t]:
> __arm_vminvq_p_u16 (__p0, __ARM_mve_coerce(__p1, uint16x8_t), p2), \
> +  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint32x4_t]:
> __arm_vminvq_p_u32 (__p0, __ARM_mve_coerce(__p1, uint32x4_t), p2));})
> 
>  #define __arm_vmladavaq(p0,p1,p2) ({ __typeof(p0) __p0 = (p0); \
>    __typeof(p1) __p1 = (p1); \
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_p_s16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_p_s16.c
> index
> 02e02270d4dee2fa4065b35372959f9d5738ff5d..74ffad4e72695c761908876
> 062dc03cbf7037f00 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_p_s16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_p_s16.c
> @@ -10,7 +10,6 @@ foo (uint16_t a, int16x8_t b, mve_pred16_t p)
>    return vmaxavq_p_s16 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxavt.s16"  }  } */
> 
>  uint16_t
>  foo1 (uint16_t a, int16x8_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (uint16_t a, int16x8_t b, mve_pred16_t p)
>    return vmaxavq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxavt.s16"  }  } */
> +
> +int16_t
> +foo2 (uint8_t a, int16x8_t b, mve_pred16_t p)
> +{
> +  return vmaxavq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxavt.s16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_p_s32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_p_s32.c
> index
> 7ecd94abbe7752ff217b7852942f1a72d6ad5dd5..40800b0f12ef3a836692d25
> d7a50d4b5ff63e852 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_p_s32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_p_s32.c
> @@ -10,7 +10,6 @@ foo (uint32_t a, int32x4_t b, mve_pred16_t p)
>    return vmaxavq_p_s32 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxavt.s32"  }  } */
> 
>  uint32_t
>  foo1 (uint32_t a, int32x4_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (uint32_t a, int32x4_t b, mve_pred16_t p)
>    return vmaxavq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxavt.s32"  }  } */
> +
> +int32_t
> +foo2 (uint16_t a, int32x4_t b, mve_pred16_t p)
> +{
> +  return vmaxavq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxavt.s32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_p_s8.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_p_s8.c
> index
> 7a21de7815382ef00ad90ce150f875705ba44a01..7638737fb842de8fc22345b
> 94065abac3096a8d3 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_p_s8.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_p_s8.c
> @@ -10,7 +10,6 @@ foo (uint8_t a, int8x16_t b, mve_pred16_t p)
>    return vmaxavq_p_s8 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxavt.s8"  }  } */
> 
>  uint8_t
>  foo1 (uint8_t a, int8x16_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (uint8_t a, int8x16_t b, mve_pred16_t p)
>    return vmaxavq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxavt.s8"  }  } */
> +
> +int8_t
> +foo2 (uint32_t a, int8x16_t b, mve_pred16_t p)
> +{
> +  return vmaxavq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxavt.s8" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_s16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_s16.c
> index
> 4621eba63c756725746c09873aae4df24b9fd4d3..0dca149b3e8652cc3c2b5a
> dbb40067ed65eb10a9 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_s16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_s16.c
> @@ -10,7 +10,6 @@ foo (uint16_t a, int16x8_t b)
>    return vmaxavq_s16 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxav.s16"  }  } */
> 
>  uint16_t
>  foo1 (uint16_t a, int16x8_t b)
> @@ -18,4 +17,12 @@ foo1 (uint16_t a, int16x8_t b)
>    return vmaxavq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxav.s16"  }  } */
> +
> +int16_t
> +foo2 (uint8_t a, int16x8_t b)
> +{
> +  return vmaxavq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxav.s16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_s32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_s32.c
> index
> 8813d9da05347b4e0c4012140edd7e6af59853c6..f419a771017f74d1cf300db
> e9ed31723ce254ca8 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_s32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_s32.c
> @@ -10,7 +10,6 @@ foo (uint32_t a, int32x4_t b)
>    return vmaxavq_s32 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxav.s32"  }  } */
> 
>  uint32_t
>  foo1 (uint32_t a, int32x4_t b)
> @@ -18,4 +17,12 @@ foo1 (uint32_t a, int32x4_t b)
>    return vmaxavq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxav.s32"  }  } */
> +
> +int32_t
> +foo2 (uint16_t a, int32x4_t b)
> +{
> +  return vmaxavq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxav.s32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_s8.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_s8.c
> index
> 961f1d28ad5aa176ebbdc6312b24cb26f0f4fb7e..214ad88f4aa6641f8d2cec9
> 17c67a4a7b774badf 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_s8.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxavq_s8.c
> @@ -10,7 +10,6 @@ foo (uint8_t a, int8x16_t b)
>    return vmaxavq_s8 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxav.s8"  }  } */
> 
>  uint8_t
>  foo1 (uint8_t a, int8x16_t b)
> @@ -18,4 +17,12 @@ foo1 (uint8_t a, int8x16_t b)
>    return vmaxavq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxav.s8"  }  } */
> +
> +int8_t
> +foo2 (uint32_t a, int8x16_t b)
> +{
> +  return vmaxavq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxav.s8" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_f16.c
> index
> de48ea8f932d1ac01b1f771cfcc6c98cfd8ebf6f..6d8cf19a3415c094e2ede662
> b78b35b8aa152388 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_f16.c
> @@ -10,7 +10,6 @@ foo (float16_t a, float16x8_t b)
>    return vmaxnmavq_f16 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmav.f16"  }  } */
> 
>  float16_t
>  foo1 (float16_t a, float16x8_t b)
> @@ -18,4 +17,12 @@ foo1 (float16_t a, float16x8_t b)
>    return vmaxnmavq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmav.f16"  }  } */
> +
> +float16_t
> +foo2 (float32_t a, float16x8_t b)
> +{
> +  return vmaxnmavq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxnmav.f16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_f32.c
> index
> b4c7f8369694132724d7d778c6b3b972d93be6ab..ef79030d8ebd892781748c
> 8cdff497926189816b 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_f32.c
> @@ -10,7 +10,6 @@ foo (float32_t a, float32x4_t b)
>    return vmaxnmavq_f32 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmav.f32"  }  } */
> 
>  float32_t
>  foo1 (float32_t a, float32x4_t b)
> @@ -18,4 +17,12 @@ foo1 (float32_t a, float32x4_t b)
>    return vmaxnmavq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmav.f32"  }  } */
> +
> +float32_t
> +foo2 (float16_t a, float32x4_t b)
> +{
> +  return vmaxnmavq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxnmav.f32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_p_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_p_f16.c
> index
> 9c2eed08c37127c6db82647a4b74b04fcd616292..f7f39f59dade185134069da
> a6b6e3ac0241fc382 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_p_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_p_f16.c
> @@ -10,7 +10,6 @@ foo (float16_t a, float16x8_t b, mve_pred16_t p)
>    return vmaxnmavq_p_f16 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmavt.f16"  }  } */
> 
>  float16_t
>  foo1 (float16_t a, float16x8_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (float16_t a, float16x8_t b, mve_pred16_t p)
>    return vmaxnmavq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmavt.f16"  }  } */
> +
> +float16_t
> +foo2 (float32_t a, float16x8_t b, mve_pred16_t p)
> +{
> +  return vmaxnmavq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxnmavt.f16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_p_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_p_f32.c
> index
> 1cadccb6c8b909b1d980fb7e7b05258701c3f20b..341f6254a5a65cd5e77561f
> c398061f4343b1b9b 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_p_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmavq_p_f32.c
> @@ -10,7 +10,6 @@ foo (float32_t a, float32x4_t b, mve_pred16_t p)
>    return vmaxnmavq_p_f32 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmavt.f32"  }  } */
> 
>  float32_t
>  foo1 (float32_t a, float32x4_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (float32_t a, float32x4_t b, mve_pred16_t p)
>    return vmaxnmavq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmavt.f32"  }  } */
> +
> +float32_t
> +foo2 (float16_t a, float32x4_t b, mve_pred16_t p)
> +{
> +  return vmaxnmavq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxnmavt.f32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_f16.c
> index
> 81f4b9bd49a8c6661717e0938795f28745e392fe..80bd1d4cda100df86f60faa
> 04f4c452928f1c630 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_f16.c
> @@ -10,7 +10,6 @@ foo (float16_t a, float16x8_t b)
>    return vmaxnmvq_f16 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmv.f16"  }  } */
> 
>  float16_t
>  foo1 (float16_t a, float16x8_t b)
> @@ -18,4 +17,12 @@ foo1 (float16_t a, float16x8_t b)
>    return vmaxnmvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmv.f16"  }  } */
> +
> +float16_t
> +foo2 (float32_t a, float16x8_t b)
> +{
> +  return vmaxnmvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxnmv.f16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_f32.c
> index
> ab06c2b3de2a59ebb09cb1529e9f4ada9deef34d..bb2fc46f88a617facd28748
> c576e8556d6dc2a69 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_f32.c
> @@ -10,7 +10,6 @@ foo (float32_t a, float32x4_t b)
>    return vmaxnmvq_f32 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmv.f32"  }  } */
> 
>  float32_t
>  foo1 (float32_t a, float32x4_t b)
> @@ -18,4 +17,12 @@ foo1 (float32_t a, float32x4_t b)
>    return vmaxnmvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmv.f32"  }  } */
> +
> +float32_t
> +foo2 (float16_t a, float32x4_t b)
> +{
> +  return vmaxnmvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxnmv.f32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_p_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_p_f16.c
> index
> e37c5a107bb8c96b4073f93d2ae6e07b87c95c7b..3efe203007b2bd3f29dee5
> db0d62da50de21f1cd 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_p_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_p_f16.c
> @@ -10,7 +10,6 @@ foo (float16_t a, float16x8_t b, mve_pred16_t p)
>    return vmaxnmvq_p_f16 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmvt.f16"  }  } */
> 
>  float16_t
>  foo1 (float16_t a, float16x8_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (float16_t a, float16x8_t b, mve_pred16_t p)
>    return vmaxnmvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmvt.f16"  }  } */
> +
> +float16_t
> +foo2 (float32_t a, float16x8_t b, mve_pred16_t p)
> +{
> +  return vmaxnmvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxnmvt.f16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_p_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_p_f32.c
> index
> 884cd456614cc2299483898acd2c1282a3355f4d..6c13247f1f1db0eaa5752d9
> d19244f71a0fcb691 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_p_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxnmvq_p_f32.c
> @@ -10,7 +10,6 @@ foo (float32_t a, float32x4_t b, mve_pred16_t p)
>    return vmaxnmvq_p_f32 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmvt.f32"  }  } */
> 
>  float32_t
>  foo1 (float32_t a, float32x4_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (float32_t a, float32x4_t b, mve_pred16_t p)
>    return vmaxnmvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxnmvt.f32"  }  } */
> +
> +float32_t
> +foo2 (float16_t a, float32x4_t b, mve_pred16_t p)
> +{
> +  return vmaxnmvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxnmvt.f32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_s16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_s16.c
> index
> 79de370bfd76379df832c2c1242f372b3fe32b93..657efc51bea5856111fa3f1e
> ce361327e6566ccd 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_s16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_s16.c
> @@ -10,7 +10,6 @@ foo (int16_t a, int16x8_t b, mve_pred16_t p)
>    return vmaxvq_p_s16 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxvt.s16"  }  } */
> 
>  int16_t
>  foo1 (int16_t a, int16x8_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (int16_t a, int16x8_t b, mve_pred16_t p)
>    return vmaxvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxvt.s16"  }  } */
> +
> +int16_t
> +foo2 (int8_t a, int16x8_t b, mve_pred16_t p)
> +{
> +  return vmaxvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxvt.s16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_s32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_s32.c
> index
> e52674486eaabd75d2ed32f85a7fb483e78383b2..5882351c0fa4f6562bb768
> 18f7d2a666528105fc 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_s32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_s32.c
> @@ -10,7 +10,6 @@ foo (int32_t a, int32x4_t b, mve_pred16_t p)
>    return vmaxvq_p_s32 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxvt.s32"  }  } */
> 
>  int32_t
>  foo1 (int32_t a, int32x4_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (int32_t a, int32x4_t b, mve_pred16_t p)
>    return vmaxvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxvt.s32"  }  } */
> +
> +int32_t
> +foo2 (int16_t a, int32x4_t b, mve_pred16_t p)
> +{
> +  return vmaxvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxvt.s32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_s8.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_s8.c
> index
> d3cedd4cd33977bf17c5db1bd818081511bcaa30..3737ecd3307fb5bd931e8fc
> 9962b21209e1bfbd4 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_s8.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_s8.c
> @@ -10,7 +10,6 @@ foo (int8_t a, int8x16_t b, mve_pred16_t p)
>    return vmaxvq_p_s8 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxvt.s8"  }  } */
> 
>  int8_t
>  foo1 (int8_t a, int8x16_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (int8_t a, int8x16_t b, mve_pred16_t p)
>    return vmaxvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxvt.s8"  }  } */
> +
> +int8_t
> +foo2 (int32_t a, int8x16_t b, mve_pred16_t p)
> +{
> +  return vmaxvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxvt.s8" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_u16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_u16.c
> index
> 79572f7a246547f0fd4f0192ccdd2f43aa994442..348cf39caa0c32d283bd6d00
> 3c2b69334b13f611 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_u16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_u16.c
> @@ -10,7 +10,6 @@ foo (uint16_t a, uint16x8_t b, mve_pred16_t p)
>    return vmaxvq_p_u16 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxvt.u16"  }  } */
> 
>  uint16_t
>  foo1 (uint16_t a, uint16x8_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (uint16_t a, uint16x8_t b, mve_pred16_t p)
>    return vmaxvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxvt.u16"  }  } */
> +
> +uint16_t
> +foo2 (uint32_t a, uint16x8_t b, mve_pred16_t p)
> +{
> +  return vmaxvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxvt.u16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_u32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_u32.c
> index
> e2f7a6f332989108e674cdadf62ca5c33271a343..f2e976216c58e6cedb773fe
> 6fc0c8ad70ffdcb67 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_u32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_u32.c
> @@ -10,7 +10,6 @@ foo (uint32_t a, uint32x4_t b, mve_pred16_t p)
>    return vmaxvq_p_u32 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxvt.u32"  }  } */
> 
>  uint32_t
>  foo1 (uint32_t a, uint32x4_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (uint32_t a, uint32x4_t b, mve_pred16_t p)
>    return vmaxvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxvt.u32"  }  } */
> +
> +uint32_t
> +foo2 (uint8_t a, uint32x4_t b, mve_pred16_t p)
> +{
> +  return vmaxvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxvt.u32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_u8.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_u8.c
> index
> f977806bf2ee0c5a224f075d0f920ba496b28e8c..7df5b63c9bc68d48715e707
> ef34dc949eca2cc99 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_u8.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_p_u8.c
> @@ -10,7 +10,6 @@ foo (uint8_t a, uint8x16_t b, mve_pred16_t p)
>    return vmaxvq_p_u8 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxvt.u8"  }  } */
> 
>  uint8_t
>  foo1 (uint8_t a, uint8x16_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (uint8_t a, uint8x16_t b, mve_pred16_t p)
>    return vmaxvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxvt.u8"  }  } */
> +
> +uint8_t
> +foo2 (uint16_t a, uint8x16_t b, mve_pred16_t p)
> +{
> +  return vmaxvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxvt.u8" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_s16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_s16.c
> index
> 90f10b54bff3e92a957c48d35217a29f70efc32e..8412452cf3316a4db378672
> 467a89795f5a0bcea 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_s16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_s16.c
> @@ -10,7 +10,6 @@ foo (int16_t a, int16x8_t b)
>    return vmaxvq_s16 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxv.s16"  }  } */
> 
>  int16_t
>  foo1 (int16_t a, int16x8_t b)
> @@ -18,4 +17,12 @@ foo1 (int16_t a, int16x8_t b)
>    return vmaxvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxv.s16"  }  } */
> +
> +int16_t
> +foo2 (int8_t a, int16x8_t b)
> +{
> +  return vmaxvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxv.s16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_s32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_s32.c
> index
> aa0e88b94835066d6d7acd9c133f247019eb1c18..09f4909c9a8557ba1aac54
> cce9dfab77630072ae 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_s32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_s32.c
> @@ -10,7 +10,6 @@ foo (int32_t a, int32x4_t b)
>    return vmaxvq_s32 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxv.s32"  }  } */
> 
>  int32_t
>  foo1 (int32_t a, int32x4_t b)
> @@ -18,4 +17,12 @@ foo1 (int32_t a, int32x4_t b)
>    return vmaxvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxv.s32"  }  } */
> +
> +int32_t
> +foo2 (int16_t a, int32x4_t b)
> +{
> +  return vmaxvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxv.s32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_s8.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_s8.c
> index
> 884b84d8e39dd4ea3ffaf66e8e8c2f23f6ca0467..a087bbc6b6466ad06e33534
> 308a68b64d8dc05c2 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_s8.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_s8.c
> @@ -10,7 +10,6 @@ foo (int8_t a, int8x16_t b)
>    return vmaxvq_s8 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxv.s8"  }  } */
> 
>  int8_t
>  foo1 (int8_t a, int8x16_t b)
> @@ -18,4 +17,12 @@ foo1 (int8_t a, int8x16_t b)
>    return vmaxvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxv.s8"  }  } */
> +
> +int8_t
> +foo2 (int32_t a, int8x16_t b)
> +{
> +  return vmaxvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxv.s8" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_u16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_u16.c
> index
> 2813ebdd3c672b8bc86492b440eb7bc6b83702f0..47fe0d1cf0f89b04b74aa58
> afd9c4acbae69137f 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_u16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_u16.c
> @@ -10,7 +10,6 @@ foo (uint16_t a, uint16x8_t b)
>    return vmaxvq_u16 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxv.u16"  }  } */
> 
>  uint16_t
>  foo1 (uint16_t a, uint16x8_t b)
> @@ -18,4 +17,12 @@ foo1 (uint16_t a, uint16x8_t b)
>    return vmaxvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxv.u16"  }  } */
> +
> +uint16_t
> +foo2 (uint32_t a, uint16x8_t b)
> +{
> +  return vmaxvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxv.u16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_u32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_u32.c
> index
> ab51b1e108872c59c33fd6ac3d64d3387ea7e762..aa723daf5ddffeb20809c18
> 51d18c232783e3725 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_u32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_u32.c
> @@ -10,7 +10,6 @@ foo (uint32_t a, uint32x4_t b)
>    return vmaxvq_u32 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxv.u32"  }  } */
> 
>  uint32_t
>  foo1 (uint32_t a, uint32x4_t b)
> @@ -18,4 +17,12 @@ foo1 (uint32_t a, uint32x4_t b)
>    return vmaxvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxv.u32"  }  } */
> +
> +uint32_t
> +foo2 (uint8_t a, uint32x4_t b)
> +{
> +  return vmaxvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxv.u32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_u8.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_u8.c
> index
> 3326cfb4b660654157ba4527d38385bbed75c373..3aae785040c172ef86f6c8
> 18943eedb4e3fa11bb 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_u8.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vmaxvq_u8.c
> @@ -10,7 +10,6 @@ foo (uint8_t a, uint8x16_t b)
>    return vmaxvq_u8 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxv.u8"  }  } */
> 
>  uint8_t
>  foo1 (uint8_t a, uint8x16_t b)
> @@ -18,4 +17,12 @@ foo1 (uint8_t a, uint8x16_t b)
>    return vmaxvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vmaxv.u8"  }  } */
> +
> +uint8_t
> +foo2 (uint16_t a, uint8x16_t b)
> +{
> +  return vmaxvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vmaxv.u8" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_p_s16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_p_s16.c
> index
> 6b876483bc48b7c31e639319adedfad9a2130f0b..9303ae02e39f38c0152f927
> 9f7823b11e14486f0 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_p_s16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_p_s16.c
> @@ -10,7 +10,6 @@ foo (uint16_t a, int16x8_t b, mve_pred16_t p)
>    return vminavq_p_s16 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminavt.s16"  }  } */
> 
>  uint16_t
>  foo1 (uint16_t a, int16x8_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (uint16_t a, int16x8_t b, mve_pred16_t p)
>    return vminavq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminavt.s16"  }  } */
> +
> +int16_t
> +foo2 (uint8_t a, int16x8_t b, mve_pred16_t p)
> +{
> +  return vminavq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminavt.s16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_p_s32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_p_s32.c
> index
> 086ff56b93572911afb09c791997380ba2d0f2d2..36247f68b2cbc192926a76a
> 27eaf607ea88d2764 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_p_s32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_p_s32.c
> @@ -10,7 +10,6 @@ foo (uint32_t a, int32x4_t b, mve_pred16_t p)
>    return vminavq_p_s32 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminavt.s32"  }  } */
> 
>  uint32_t
>  foo1 (uint32_t a, int32x4_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (uint32_t a, int32x4_t b, mve_pred16_t p)
>    return vminavq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminavt.s32"  }  } */
> +
> +int32_t
> +foo2 (uint16_t a, int32x4_t b, mve_pred16_t p)
> +{
> +  return vminavq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminavt.s32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_p_s8.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_p_s8.c
> index
> 999c11ca73cf0fde0b6588048c15becce0a46782..d3361615dcc983e3d42ce27
> 72441f74a00e13d5e 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_p_s8.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_p_s8.c
> @@ -10,7 +10,6 @@ foo (uint8_t a, int8x16_t b, mve_pred16_t p)
>    return vminavq_p_s8 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminavt.s8"  }  } */
> 
>  uint8_t
>  foo1 (uint8_t a, int8x16_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (uint8_t a, int8x16_t b, mve_pred16_t p)
>    return vminavq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminavt.s8"  }  } */
> +
> +int8_t
> +foo2 (uint32_t a, int8x16_t b, mve_pred16_t p)
> +{
> +  return vminavq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminavt.s8" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_s16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_s16.c
> index
> a626e3122914f0c730f501bf05a94e76c8f74d3f..17e4edca2f1acf8f94a22a9e
> 9dae4c7868c0b455 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_s16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_s16.c
> @@ -10,7 +10,6 @@ foo (uint16_t a, int16x8_t b)
>    return vminavq_s16 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminav.s16"  }  } */
> 
>  uint16_t
>  foo1 (uint16_t a, int16x8_t b)
> @@ -18,4 +17,12 @@ foo1 (uint16_t a, int16x8_t b)
>    return vminavq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminav.s16"  }  } */
> +
> +int16_t
> +foo2 (uint8_t a, int16x8_t b)
> +{
> +  return vminavq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminav.s16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_s32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_s32.c
> index
> be575cb5da3ba37fdfaf9df7f37b92d726a3834f..032d02b885750c9b7077681
> ac7618c063b160ae0 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_s32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_s32.c
> @@ -10,7 +10,6 @@ foo (uint32_t a, int32x4_t b)
>    return vminavq_s32 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminav.s32"  }  } */
> 
>  uint32_t
>  foo1 (uint32_t a, int32x4_t b)
> @@ -18,4 +17,12 @@ foo1 (uint32_t a, int32x4_t b)
>    return vminavq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminav.s32"  }  } */
> +
> +int32_t
> +foo2 (uint16_t a, int32x4_t b)
> +{
> +  return vminavq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminav.s32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_s8.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_s8.c
> index
> c3dfe4bfeba8a361aa16cad0966c15af7e9473b6..2a2bb3d614674652ac3a8a
> 7956a2d394f36f27a8 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_s8.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminavq_s8.c
> @@ -10,7 +10,6 @@ foo (uint8_t a, int8x16_t b)
>    return vminavq_s8 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminav.s8"  }  } */
> 
>  uint8_t
>  foo1 (uint8_t a, int8x16_t b)
> @@ -18,4 +17,12 @@ foo1 (uint8_t a, int8x16_t b)
>    return vminavq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminav.s8"  }  } */
> +
> +int8_t
> +foo2 (uint32_t a, int8x16_t b)
> +{
> +  return vminavq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminav.s8" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_f16.c
> index
> 2111681cea4d1ba86146816ef74c004b932a412e..fadb23e05c847a8f15b9ac
> 1f15d8e6c89a1b34bc 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_f16.c
> @@ -10,7 +10,6 @@ foo (float16_t a, float16x8_t b)
>    return vminnmavq_f16 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmav.f16"  }  } */
> 
>  float16_t
>  foo1 (float16_t a, float16x8_t b)
> @@ -18,4 +17,12 @@ foo1 (float16_t a, float16x8_t b)
>    return vminnmavq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmav.f16"  }  } */
> +
> +float16_t
> +foo2 (float32_t a, float16x8_t b)
> +{
> +  return vminnmavq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminnmav.f16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_f32.c
> index
> bd87b85ec4a610674af1b4f5179c69ca634c5da0..84714a96b9f54c3aba7aad7
> bfa6e6ff7758e5679 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_f32.c
> @@ -10,7 +10,6 @@ foo (float32_t a, float32x4_t b)
>    return vminnmavq_f32 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmav.f32"  }  } */
> 
>  float32_t
>  foo1 (float32_t a, float32x4_t b)
> @@ -18,4 +17,12 @@ foo1 (float32_t a, float32x4_t b)
>    return vminnmavq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmav.f32"  }  } */
> +
> +float32_t
> +foo2 (float16_t a, float32x4_t b)
> +{
> +  return vminnmavq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminnmav.f32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_p_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_p_f16.c
> index
> e6d0bb5a8f011383cd92b635240624b822ca139e..c79fa307ae024e3c8d8697
> 654af580f460bf4415 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_p_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_p_f16.c
> @@ -10,7 +10,6 @@ foo (float16_t a, float16x8_t b, mve_pred16_t p)
>    return vminnmavq_p_f16 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmavt.f16"  }  } */
> 
>  float16_t
>  foo1 (float16_t a, float16x8_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (float16_t a, float16x8_t b, mve_pred16_t p)
>    return vminnmavq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmavt.f16"  }  } */
> +
> +float16_t
> +foo2 (float32_t a, float16x8_t b, mve_pred16_t p)
> +{
> +  return vminnmavq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminnmavt.f16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_p_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_p_f32.c
> index
> 6b56b67b7ba1f85446b6f6b652c5af4b82b659f6..bea04c7aac699dad5f21adc
> 8aa3f8be070557855 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_p_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmavq_p_f32.c
> @@ -10,7 +10,6 @@ foo (float32_t a, float32x4_t b, mve_pred16_t p)
>    return vminnmavq_p_f32 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmavt.f32"  }  } */
> 
>  float32_t
>  foo1 (float32_t a, float32x4_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (float32_t a, float32x4_t b, mve_pred16_t p)
>    return vminnmavq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmavt.f32"  }  } */
> +
> +float32_t
> +foo2 (float16_t a, float32x4_t b, mve_pred16_t p)
> +{
> +  return vminnmavq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminnmavt.f32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_f16.c
> index
> 4d4caaea992e7ac5b507199f743bef87cee71278..0eb3a4af14ec57d60ec34bf
> c2d7210d8db60ef52 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_f16.c
> @@ -10,7 +10,6 @@ foo (float16_t a, float16x8_t b)
>    return vminnmvq_f16 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmv.f16"  }  } */
> 
>  float16_t
>  foo1 (float16_t a, float16x8_t b)
> @@ -18,4 +17,12 @@ foo1 (float16_t a, float16x8_t b)
>    return vminnmvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmv.f16"  }  } */
> +
> +float16_t
> +foo2 (float32_t a, float16x8_t b)
> +{
> +  return vminnmvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminnmv.f16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_f32.c
> index
> dab04d93f0f7275aad46264a518806a0ccdceb1c..f3183508f8e5088e7fa37aa
> d0cca77c1014065d8 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_f32.c
> @@ -10,7 +10,6 @@ foo (float32_t a, float32x4_t b)
>    return vminnmvq_f32 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmv.f32"  }  } */
> 
>  float32_t
>  foo1 (float32_t a, float32x4_t b)
> @@ -18,4 +17,12 @@ foo1 (float32_t a, float32x4_t b)
>    return vminnmvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmv.f32"  }  } */
> +
> +float32_t
> +foo2 (float16_t a, float32x4_t b)
> +{
> +  return vminnmvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminnmv.f32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_p_f16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_p_f16.c
> index
> f5eafb1012261b1dcbee82c984e9997bf2938a60..16f6ac514c86d99c87708a6
> a69eca984f15f769b 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_p_f16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_p_f16.c
> @@ -10,7 +10,6 @@ foo (float16_t a, float16x8_t b, mve_pred16_t p)
>    return vminnmvq_p_f16 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmvt.f16"  }  } */
> 
>  float16_t
>  foo1 (float16_t a, float16x8_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (float16_t a, float16x8_t b, mve_pred16_t p)
>    return vminnmvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmvt.f16"  }  } */
> +
> +float16_t
> +foo2 (float32_t a, float16x8_t b, mve_pred16_t p)
> +{
> +  return vminnmvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminnmvt.f16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_p_f32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_p_f32.c
> index
> 5ac20bf305f66e50a919f0eff0a26b3d3dfbea7d..a8e4f9ffba7286a4b77c97ac
> 156d79e3be3355d4 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_p_f32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminnmvq_p_f32.c
> @@ -10,7 +10,6 @@ foo (float32_t a, float32x4_t b, mve_pred16_t p)
>    return vminnmvq_p_f32 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmvt.f32"  }  } */
> 
>  float32_t
>  foo1 (float32_t a, float32x4_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (float32_t a, float32x4_t b, mve_pred16_t p)
>    return vminnmvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminnmvt.f32"  }  } */
> +
> +float32_t
> +foo2 (float16_t a, float32x4_t b, mve_pred16_t p)
> +{
> +  return vminnmvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminnmvt.f32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_s16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_s16.c
> index
> c2edb622d5a5c77376e0c5d6486998832786c37e..91bb63f6ba6ea0dcafb22d
> b612aa42d44a58de67 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_s16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_s16.c
> @@ -10,7 +10,6 @@ foo (int16_t a, int16x8_t b, mve_pred16_t p)
>    return vminvq_p_s16 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminvt.s16"  }  } */
> 
>  int16_t
>  foo1 (int16_t a, int16x8_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (int16_t a, int16x8_t b, mve_pred16_t p)
>    return vminvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminvt.s16"  }  } */
> +
> +int16_t
> +foo2 (int8_t a, int16x8_t b, mve_pred16_t p)
> +{
> +  return vminvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminvt.s16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_s32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_s32.c
> index
> ba8921715a41636774fc583382dc152e9dab4a6b..a846701312c858cf2cb6d9
> d9c113ce9c5a31fb53 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_s32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_s32.c
> @@ -10,7 +10,6 @@ foo (int32_t a, int32x4_t b, mve_pred16_t p)
>    return vminvq_p_s32 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminvt.s32"  }  } */
> 
>  int32_t
>  foo1 (int32_t a, int32x4_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (int32_t a, int32x4_t b, mve_pred16_t p)
>    return vminvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminvt.s32"  }  } */
> +
> +int32_t
> +foo2 (int16_t a, int32x4_t b, mve_pred16_t p)
> +{
> +  return vminvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminvt.s32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_s8.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_s8.c
> index
> 1665c53f96e91c44e4b6d213bfaf33e96eae50c0..716d414f3a7555de8738349
> b35df01d2a6f08832 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_s8.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_s8.c
> @@ -10,7 +10,6 @@ foo (int8_t a, int8x16_t b, mve_pred16_t p)
>    return vminvq_p_s8 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminvt.s8"  }  } */
> 
>  int8_t
>  foo1 (int8_t a, int8x16_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (int8_t a, int8x16_t b, mve_pred16_t p)
>    return vminvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminvt.s8"  }  } */
> +
> +int8_t
> +foo2 (int32_t a, int8x16_t b, mve_pred16_t p)
> +{
> +  return vminvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminvt.s8" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_u16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_u16.c
> index
> 5bade0ac334f0591922d7c8c14e58a8f8af511af..cc7f8fe8933d3e773f91a860
> 5287709f94492097 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_u16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_u16.c
> @@ -10,7 +10,6 @@ foo (uint16_t a, uint16x8_t b, mve_pred16_t p)
>    return vminvq_p_u16 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminvt.u16"  }  } */
> 
>  uint16_t
>  foo1 (uint16_t a, uint16x8_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (uint16_t a, uint16x8_t b, mve_pred16_t p)
>    return vminvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminvt.u16"  }  } */
> +
> +uint16_t
> +foo2 (uint32_t a, uint16x8_t b, mve_pred16_t p)
> +{
> +  return vminvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminvt.u16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_u32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_u32.c
> index
> c4c574882a384213b774ea12387ecffb7029eb14..6bde0be29ccf73a1cb36535
> a3d2677d4338c3da7 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_u32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_u32.c
> @@ -10,7 +10,6 @@ foo (uint32_t a, uint32x4_t b, mve_pred16_t p)
>    return vminvq_p_u32 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminvt.u32"  }  } */
> 
>  uint32_t
>  foo1 (uint32_t a, uint32x4_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (uint32_t a, uint32x4_t b, mve_pred16_t p)
>    return vminvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminvt.u32"  }  } */
> +
> +uint32_t
> +foo2 (uint8_t a, uint32x4_t b, mve_pred16_t p)
> +{
> +  return vminvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminvt.u32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_u8.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_u8.c
> index
> dc890dc34725816c5ec163a4ee05b044f68e4682..bb894904f3cfeee8c881bf7
> f2b13fb6d7e0f4f7a 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_u8.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_p_u8.c
> @@ -10,7 +10,6 @@ foo (uint8_t a, uint8x16_t b, mve_pred16_t p)
>    return vminvq_p_u8 (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminvt.u8"  }  } */
> 
>  uint8_t
>  foo1 (uint8_t a, uint8x16_t b, mve_pred16_t p)
> @@ -18,4 +17,12 @@ foo1 (uint8_t a, uint8x16_t b, mve_pred16_t p)
>    return vminvq_p (a, b, p);
>  }
> 
> -/* { dg-final { scan-assembler "vminvt.u8"  }  } */
> +
> +uint8_t
> +foo2 (uint16_t a, uint8x16_t b, mve_pred16_t p)
> +{
> +  return vminvq_p (a, b, p);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminvt.u8" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_s16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_s16.c
> index
> f6eed633f03f4ef7058f4579c4d36db0e2494f14..6d589aa4a0506cd18fb4942e
> 5511b14fe15a2811 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_s16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_s16.c
> @@ -10,7 +10,6 @@ foo (int16_t a, int16x8_t b)
>    return vminvq_s16 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminv.s16"  }  } */
> 
>  int16_t
>  foo1 (int16_t a, int16x8_t b)
> @@ -18,4 +17,11 @@ foo1 (int16_t a, int16x8_t b)
>    return vminvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminv.s16"  }  } */
> +int16_t
> +foo2 (int8_t a, int16x8_t b)
> +{
> +  return vminvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminv.s16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_s32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_s32.c
> index
> 4077c32fffec22e528cede2c4fcecc60a98ae008..7c727d6d92b05424aa8441bf
> 8980355bee72f168 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_s32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_s32.c
> @@ -10,7 +10,6 @@ foo (int32_t a, int32x4_t b)
>    return vminvq_s32 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminv.s32"  }  } */
> 
>  int32_t
>  foo1 (int32_t a, int32x4_t b)
> @@ -18,4 +17,11 @@ foo1 (int32_t a, int32x4_t b)
>    return vminvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminv.s32"  }  } */
> +int32_t
> +foo2 (int8_t a, int32x4_t b)
> +{
> +  return vminvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminv.s32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_s8.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_s8.c
> index
> bdf15f45acc98ff14969a7d311deac7dd45a898f..76309482fc56d6166d91f084
> 5c3fb60dfdaaddfd 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_s8.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_s8.c
> @@ -10,7 +10,6 @@ foo (int8_t a, int8x16_t b)
>    return vminvq_s8 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminv.s8"  }  } */
> 
>  int8_t
>  foo1 (int8_t a, int8x16_t b)
> @@ -18,4 +17,11 @@ foo1 (int8_t a, int8x16_t b)
>    return vminvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminv.s8"  }  } */
> +int8_t
> +foo2 (int32_t a, int8x16_t b)
> +{
> +  return vminvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminv.s8" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_u16.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_u16.c
> index
> 5c0935c78cd48bebf0f565ffcccde623dc8e8aab..698975f456c708697f4c4820
> 970b08c6321d81fc 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_u16.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_u16.c
> @@ -10,7 +10,6 @@ foo (uint16_t a, uint16x8_t b)
>    return vminvq_u16 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminv.u16"  }  } */
> 
>  uint16_t
>  foo1 (uint16_t a, uint16x8_t b)
> @@ -18,4 +17,12 @@ foo1 (uint16_t a, uint16x8_t b)
>    return vminvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminv.u16"  }  } */
> +
> +uint8_t
> +foo2 (uint32_t a, uint16x8_t b)
> +{
> +  return vminvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminv.u16" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_u32.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_u32.c
> index
> 1580c87fbb1bfe67aea0fa1c29ad89adbc9aafe8..7489f81debfeda816fe29379
> b695afa50df3ac0e 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_u32.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_u32.c
> @@ -10,7 +10,6 @@ foo (uint32_t a, uint32x4_t b)
>    return vminvq_u32 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminv.u32"  }  } */
> 
>  uint32_t
>  foo1 (uint32_t a, uint32x4_t b)
> @@ -18,4 +17,11 @@ foo1 (uint32_t a, uint32x4_t b)
>    return vminvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminv.u32"  }  } */
> +uint32_t
> +foo2 (uint16_t a, uint32x4_t b)
> +{
> +  return vminvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminv.u32" 3 } } */
> diff --git a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_u8.c
> b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_u8.c
> index
> 95919b412e0aeced0b6c2f4ab95cfa805e655d22..aa2b986d5587c2c7a79e86
> 370cf179d5f8751db7 100644
> --- a/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_u8.c
> +++ b/gcc/testsuite/gcc.target/arm/mve/intrinsics/vminvq_u8.c
> @@ -10,7 +10,6 @@ foo (uint8_t a, uint8x16_t b)
>    return vminvq_u8 (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminv.u8"  }  } */
> 
>  uint8_t
>  foo1 (uint8_t a, uint8x16_t b)
> @@ -18,4 +17,12 @@ foo1 (uint8_t a, uint8x16_t b)
>    return vminvq (a, b);
>  }
> 
> -/* { dg-final { scan-assembler "vminv.u8"  }  } */
> +
> +uint16_t
> +foo2 (uint32_t a, uint8x16_t b)
> +{
> +  return vminvq (a, b);
> +}
> +
> +/* { dg-final { scan-assembler-not "__ARM_undef" } } */
> +/* { dg-final { scan-assembler-times "vminv.u8" 3 } } */