arm: [MVE intrinsics] rework vmlaldavq vmlaldavxq vmlsldavq vmlsldavxq

Implement vmlaldavq, vmlaldavxq, vmlsldavq, vmlsldavxq using the new
MVE builtins framework.

2022-10-25  Christophe Lyon  <christophe.lyon@arm.com>

	gcc/
	* config/arm/arm-mve-builtins-base.cc (vmlaldavq, vmlaldavxq)
	(vmlsldavq, vmlsldavxq): New.
	* config/arm/arm-mve-builtins-base.def (vmlaldavq, vmlaldavxq)
	(vmlsldavq, vmlsldavxq): New.
	* config/arm/arm-mve-builtins-base.h (vmlaldavq, vmlaldavxq)
	(vmlsldavq, vmlsldavxq): New.
	* config/arm/arm_mve.h (vmlaldavq): Remove.
	(vmlsldavxq): Remove.
	(vmlsldavq): Remove.
	(vmlaldavxq): Remove.
	(vmlaldavq_p): Remove.
	(vmlaldavxq_p): Remove.
	(vmlsldavq_p): Remove.
	(vmlsldavxq_p): Remove.
	(vmlaldavq_u16): Remove.
	(vmlsldavxq_s16): Remove.
	(vmlsldavq_s16): Remove.
	(vmlaldavxq_s16): Remove.
	(vmlaldavq_s16): Remove.
	(vmlaldavq_u32): Remove.
	(vmlsldavxq_s32): Remove.
	(vmlsldavq_s32): Remove.
	(vmlaldavxq_s32): Remove.
	(vmlaldavq_s32): Remove.
	(vmlaldavq_p_s16): Remove.
	(vmlaldavxq_p_s16): Remove.
	(vmlsldavq_p_s16): Remove.
	(vmlsldavxq_p_s16): Remove.
	(vmlaldavq_p_u16): Remove.
	(vmlaldavq_p_s32): Remove.
	(vmlaldavxq_p_s32): Remove.
	(vmlsldavq_p_s32): Remove.
	(vmlsldavxq_p_s32): Remove.
	(vmlaldavq_p_u32): Remove.
	(__arm_vmlaldavq_u16): Remove.
	(__arm_vmlsldavxq_s16): Remove.
	(__arm_vmlsldavq_s16): Remove.
	(__arm_vmlaldavxq_s16): Remove.
	(__arm_vmlaldavq_s16): Remove.
	(__arm_vmlaldavq_u32): Remove.
	(__arm_vmlsldavxq_s32): Remove.
	(__arm_vmlsldavq_s32): Remove.
	(__arm_vmlaldavxq_s32): Remove.
	(__arm_vmlaldavq_s32): Remove.
	(__arm_vmlaldavq_p_s16): Remove.
	(__arm_vmlaldavxq_p_s16): Remove.
	(__arm_vmlsldavq_p_s16): Remove.
	(__arm_vmlsldavxq_p_s16): Remove.
	(__arm_vmlaldavq_p_u16): Remove.
	(__arm_vmlaldavq_p_s32): Remove.
	(__arm_vmlaldavxq_p_s32): Remove.
	(__arm_vmlsldavq_p_s32): Remove.
	(__arm_vmlsldavxq_p_s32): Remove.
	(__arm_vmlaldavq_p_u32): Remove.
	(__arm_vmlaldavq): Remove.
	(__arm_vmlsldavxq): Remove.
	(__arm_vmlsldavq): Remove.
	(__arm_vmlaldavxq): Remove.
	(__arm_vmlaldavq_p): Remove.
	(__arm_vmlaldavxq_p): Remove.
	(__arm_vmlsldavq_p): Remove.
	(__arm_vmlsldavxq_p): Remove.

4 files changed, 12 insertions, 366 deletions

diff --git a/gcc/config/arm/arm-mve-builtins-base.cc b/gcc/config/arm/arm-mve-builtins-base.cc
index a81cf4c..af1a2c9 100644
--- a/gcc/config/arm/arm-mve-builtins-base.cc
+++ b/gcc/config/arm/arm-mve-builtins-base.cc
@@ -285,10 +285,14 @@ FUNCTION_PRED_P_S (vmladavaxq, VMLADAVAXQ)
 FUNCTION_PRED_P_S_U (vmladavaq, VMLADAVAQ)
 FUNCTION_PRED_P_S_U (vmladavq, VMLADAVQ)
 FUNCTION_PRED_P_S (vmladavxq, VMLADAVXQ)
+FUNCTION_PRED_P_S_U (vmlaldavq, VMLALDAVQ)
+FUNCTION_PRED_P_S (vmlaldavxq, VMLALDAVXQ)
 FUNCTION_PRED_P_S (vmlsdavaq, VMLSDAVAQ)
 FUNCTION_PRED_P_S (vmlsdavaxq, VMLSDAVAXQ)
 FUNCTION_PRED_P_S (vmlsdavq, VMLSDAVQ)
 FUNCTION_PRED_P_S (vmlsdavxq, VMLSDAVXQ)
+FUNCTION_PRED_P_S (vmlsldavq, VMLSLDAVQ)
+FUNCTION_PRED_P_S (vmlsldavxq, VMLSLDAVXQ)
 FUNCTION_WITHOUT_N_NO_F (vmovlbq, VMOVLBQ)
 FUNCTION_WITHOUT_N_NO_F (vmovltq, VMOVLTQ)
 FUNCTION_WITHOUT_N_NO_F (vmovnbq, VMOVNBQ)
diff --git a/gcc/config/arm/arm-mve-builtins-base.def b/gcc/config/arm/arm-mve-builtins-base.def
index 934f45b..f7f353b 100644
--- a/gcc/config/arm/arm-mve-builtins-base.def
+++ b/gcc/config/arm/arm-mve-builtins-base.def
@@ -54,10 +54,14 @@ DEF_MVE_FUNCTION (vmladavaq, binary_acca_int32, all_integer, p_or_none)
 DEF_MVE_FUNCTION (vmladavaxq, binary_acca_int32, all_signed, p_or_none)
 DEF_MVE_FUNCTION (vmladavq, binary_acc_int32, all_integer, p_or_none)
 DEF_MVE_FUNCTION (vmladavxq, binary_acc_int32, all_signed, p_or_none)
+DEF_MVE_FUNCTION (vmlaldavq, binary_acc_int64, integer_16_32, p_or_none)
+DEF_MVE_FUNCTION (vmlaldavxq, binary_acc_int64, signed_16_32, p_or_none)
 DEF_MVE_FUNCTION (vmlsdavaq, binary_acca_int32, all_signed, p_or_none)
 DEF_MVE_FUNCTION (vmlsdavaxq, binary_acca_int32, all_signed, p_or_none)
 DEF_MVE_FUNCTION (vmlsdavq, binary_acc_int32, all_integer, p_or_none)
 DEF_MVE_FUNCTION (vmlsdavxq, binary_acc_int32, all_signed, p_or_none)
+DEF_MVE_FUNCTION (vmlsldavq, binary_acc_int64, signed_16_32, p_or_none)
+DEF_MVE_FUNCTION (vmlsldavxq, binary_acc_int64, signed_16_32, p_or_none)
 DEF_MVE_FUNCTION (vmovlbq, unary_widen, integer_8_16, mx_or_none)
 DEF_MVE_FUNCTION (vmovltq, unary_widen, integer_8_16, mx_or_none)
 DEF_MVE_FUNCTION (vmovnbq, binary_move_narrow, integer_16_32, m_or_none)
diff --git a/gcc/config/arm/arm-mve-builtins-base.h b/gcc/config/arm/arm-mve-builtins-base.h
index 1d29a94..08d07a7 100644
--- a/gcc/config/arm/arm-mve-builtins-base.h
+++ b/gcc/config/arm/arm-mve-builtins-base.h
@@ -67,10 +67,14 @@ extern const function_base *const vmladavaq;
 extern const function_base *const vmladavaxq;
 extern const function_base *const vmladavq;
 extern const function_base *const vmladavxq;
+extern const function_base *const vmlaldavq;
+extern const function_base *const vmlaldavxq;
 extern const function_base *const vmlsdavaq;
 extern const function_base *const vmlsdavaxq;
 extern const function_base *const vmlsdavq;
 extern const function_base *const vmlsdavxq;
+extern const function_base *const vmlsldavq;
+extern const function_base *const vmlsldavxq;
 extern const function_base *const vmovlbq;
 extern const function_base *const vmovltq;
 extern const function_base *const vmovnbq;
diff --git a/gcc/config/arm/arm_mve.h b/gcc/config/arm/arm_mve.h
index f8afe19..50e9ecb 100644
--- a/gcc/config/arm/arm_mve.h
+++ b/gcc/config/arm/arm_mve.h
@@ -55,12 +55,8 @@
 #define vhcaddq_rot270(__a, __b) __arm_vhcaddq_rot270(__a, __b)
 #define vmulltq_poly(__a, __b) __arm_vmulltq_poly(__a, __b)
 #define vmullbq_poly(__a, __b) __arm_vmullbq_poly(__a, __b)
-#define vmlaldavq(__a, __b) __arm_vmlaldavq(__a, __b)
 #define vqdmulltq(__a, __b) __arm_vqdmulltq(__a, __b)
 #define vqdmullbq(__a, __b) __arm_vqdmullbq(__a, __b)
-#define vmlsldavxq(__a, __b) __arm_vmlsldavxq(__a, __b)
-#define vmlsldavq(__a, __b) __arm_vmlsldavq(__a, __b)
-#define vmlaldavxq(__a, __b) __arm_vmlaldavxq(__a, __b)
 #define vrmlaldavhq(__a, __b) __arm_vrmlaldavhq(__a, __b)
 #define vrmlsldavhxq(__a, __b) __arm_vrmlsldavhxq(__a, __b)
 #define vrmlsldavhq(__a, __b) __arm_vrmlsldavhq(__a, __b)
@@ -97,10 +93,6 @@
 #define vmlaldavaxq(__a, __b, __c) __arm_vmlaldavaxq(__a, __b, __c)
 #define vmlsldavaq(__a, __b, __c) __arm_vmlsldavaq(__a, __b, __c)
 #define vmlsldavaxq(__a, __b, __c) __arm_vmlsldavaxq(__a, __b, __c)
-#define vmlaldavq_p(__a, __b, __p) __arm_vmlaldavq_p(__a, __b, __p)
-#define vmlaldavxq_p(__a, __b, __p) __arm_vmlaldavxq_p(__a, __b, __p)
-#define vmlsldavq_p(__a, __b, __p) __arm_vmlsldavq_p(__a, __b, __p)
-#define vmlsldavxq_p(__a, __b, __p) __arm_vmlsldavxq_p(__a, __b, __p)
 #define vsriq_m(__a, __b, __imm, __p) __arm_vsriq_m(__a, __b, __imm, __p)
 #define vqshluq_m(__inactive, __a, __imm, __p) __arm_vqshluq_m(__inactive, __a, __imm, __p)
 #define vbicq_m(__inactive, __a, __b, __p) __arm_vbicq_m(__inactive, __a, __b, __p)
@@ -394,17 +386,12 @@
 #define vbicq_s32(__a, __b) __arm_vbicq_s32(__a, __b)
 #define vmulltq_poly_p8(__a, __b) __arm_vmulltq_poly_p8(__a, __b)
 #define vmullbq_poly_p8(__a, __b) __arm_vmullbq_poly_p8(__a, __b)
-#define vmlaldavq_u16(__a, __b) __arm_vmlaldavq_u16(__a, __b)
 #define vbicq_n_u16(__a,  __imm) __arm_vbicq_n_u16(__a,  __imm)
 #define vqdmulltq_s16(__a, __b) __arm_vqdmulltq_s16(__a, __b)
 #define vqdmulltq_n_s16(__a, __b) __arm_vqdmulltq_n_s16(__a, __b)
 #define vqdmullbq_s16(__a, __b) __arm_vqdmullbq_s16(__a, __b)
 #define vqdmullbq_n_s16(__a, __b) __arm_vqdmullbq_n_s16(__a, __b)
 #define vornq_f16(__a, __b) __arm_vornq_f16(__a, __b)
-#define vmlsldavxq_s16(__a, __b) __arm_vmlsldavxq_s16(__a, __b)
-#define vmlsldavq_s16(__a, __b) __arm_vmlsldavq_s16(__a, __b)
-#define vmlaldavxq_s16(__a, __b) __arm_vmlaldavxq_s16(__a, __b)
-#define vmlaldavq_s16(__a, __b) __arm_vmlaldavq_s16(__a, __b)
 #define vcmulq_rot90_f16(__a, __b) __arm_vcmulq_rot90_f16(__a, __b)
 #define vcmulq_rot270_f16(__a, __b) __arm_vcmulq_rot270_f16(__a, __b)
 #define vcmulq_rot180_f16(__a, __b) __arm_vcmulq_rot180_f16(__a, __b)
@@ -415,17 +402,12 @@
 #define vbicq_n_s16(__a,  __imm) __arm_vbicq_n_s16(__a,  __imm)
 #define vmulltq_poly_p16(__a, __b) __arm_vmulltq_poly_p16(__a, __b)
 #define vmullbq_poly_p16(__a, __b) __arm_vmullbq_poly_p16(__a, __b)
-#define vmlaldavq_u32(__a, __b) __arm_vmlaldavq_u32(__a, __b)
 #define vbicq_n_u32(__a,  __imm) __arm_vbicq_n_u32(__a,  __imm)
 #define vqdmulltq_s32(__a, __b) __arm_vqdmulltq_s32(__a, __b)
 #define vqdmulltq_n_s32(__a, __b) __arm_vqdmulltq_n_s32(__a, __b)
 #define vqdmullbq_s32(__a, __b) __arm_vqdmullbq_s32(__a, __b)
 #define vqdmullbq_n_s32(__a, __b) __arm_vqdmullbq_n_s32(__a, __b)
 #define vornq_f32(__a, __b) __arm_vornq_f32(__a, __b)
-#define vmlsldavxq_s32(__a, __b) __arm_vmlsldavxq_s32(__a, __b)
-#define vmlsldavq_s32(__a, __b) __arm_vmlsldavq_s32(__a, __b)
-#define vmlaldavxq_s32(__a, __b) __arm_vmlaldavxq_s32(__a, __b)
-#define vmlaldavq_s32(__a, __b) __arm_vmlaldavq_s32(__a, __b)
 #define vcmulq_rot90_f32(__a, __b) __arm_vcmulq_rot90_f32(__a, __b)
 #define vcmulq_rot270_f32(__a, __b) __arm_vcmulq_rot270_f32(__a, __b)
 #define vcmulq_rot180_f32(__a, __b) __arm_vcmulq_rot180_f32(__a, __b)
@@ -568,10 +550,6 @@
 #define vcvtnq_m_s16_f16(__inactive, __a, __p) __arm_vcvtnq_m_s16_f16(__inactive, __a, __p)
 #define vcvtpq_m_s16_f16(__inactive, __a, __p) __arm_vcvtpq_m_s16_f16(__inactive, __a, __p)
 #define vcvtq_m_s16_f16(__inactive, __a, __p) __arm_vcvtq_m_s16_f16(__inactive, __a, __p)
-#define vmlaldavq_p_s16(__a, __b, __p) __arm_vmlaldavq_p_s16(__a, __b, __p)
-#define vmlaldavxq_p_s16(__a, __b, __p) __arm_vmlaldavxq_p_s16(__a, __b, __p)
-#define vmlsldavq_p_s16(__a, __b, __p) __arm_vmlsldavq_p_s16(__a, __b, __p)
-#define vmlsldavxq_p_s16(__a, __b, __p) __arm_vmlsldavxq_p_s16(__a, __b, __p)
 #define vpselq_f16(__a, __b, __p) __arm_vpselq_f16(__a, __b, __p)
 #define vmvnq_m_n_u16(__inactive,  __imm, __p) __arm_vmvnq_m_n_u16(__inactive,  __imm, __p)
 #define vcvtmq_m_u16_f16(__inactive, __a, __p) __arm_vcvtmq_m_u16_f16(__inactive, __a, __p)
@@ -579,7 +557,6 @@
 #define vcvtpq_m_u16_f16(__inactive, __a, __p) __arm_vcvtpq_m_u16_f16(__inactive, __a, __p)
 #define vcvtq_m_u16_f16(__inactive, __a, __p) __arm_vcvtq_m_u16_f16(__inactive, __a, __p)
 #define vmlaldavaq_u16(__a, __b, __c) __arm_vmlaldavaq_u16(__a, __b, __c)
-#define vmlaldavq_p_u16(__a, __b, __p) __arm_vmlaldavq_p_u16(__a, __b, __p)
 #define vmvnq_m_n_s32(__inactive,  __imm, __p) __arm_vmvnq_m_n_s32(__inactive,  __imm, __p)
 #define vcmlaq_f32(__a, __b, __c) __arm_vcmlaq_f32(__a, __b, __c)
 #define vcmlaq_rot180_f32(__a, __b, __c) __arm_vcmlaq_rot180_f32(__a, __b, __c)
@@ -597,10 +574,6 @@
 #define vcvtnq_m_s32_f32(__inactive, __a, __p) __arm_vcvtnq_m_s32_f32(__inactive, __a, __p)
 #define vcvtpq_m_s32_f32(__inactive, __a, __p) __arm_vcvtpq_m_s32_f32(__inactive, __a, __p)
 #define vcvtq_m_s32_f32(__inactive, __a, __p) __arm_vcvtq_m_s32_f32(__inactive, __a, __p)
-#define vmlaldavq_p_s32(__a, __b, __p) __arm_vmlaldavq_p_s32(__a, __b, __p)
-#define vmlaldavxq_p_s32(__a, __b, __p) __arm_vmlaldavxq_p_s32(__a, __b, __p)
-#define vmlsldavq_p_s32(__a, __b, __p) __arm_vmlsldavq_p_s32(__a, __b, __p)
-#define vmlsldavxq_p_s32(__a, __b, __p) __arm_vmlsldavxq_p_s32(__a, __b, __p)
 #define vpselq_f32(__a, __b, __p) __arm_vpselq_f32(__a, __b, __p)
 #define vmvnq_m_n_u32(__inactive,  __imm, __p) __arm_vmvnq_m_n_u32(__inactive,  __imm, __p)
 #define vcvtmq_m_u32_f32(__inactive, __a, __p) __arm_vcvtmq_m_u32_f32(__inactive, __a, __p)
@@ -608,7 +581,6 @@
 #define vcvtpq_m_u32_f32(__inactive, __a, __p) __arm_vcvtpq_m_u32_f32(__inactive, __a, __p)
 #define vcvtq_m_u32_f32(__inactive, __a, __p) __arm_vcvtq_m_u32_f32(__inactive, __a, __p)
 #define vmlaldavaq_u32(__a, __b, __c) __arm_vmlaldavaq_u32(__a, __b, __c)
-#define vmlaldavq_p_u32(__a, __b, __p) __arm_vmlaldavq_p_u32(__a, __b, __p)
 #define vsriq_m_n_s8(__a, __b,  __imm, __p) __arm_vsriq_m_n_s8(__a, __b,  __imm, __p)
 #define vcvtq_m_n_f16_u16(__inactive, __a,  __imm6, __p) __arm_vcvtq_m_n_f16_u16(__inactive, __a,  __imm6, __p)
 #define vqshluq_m_n_s8(__inactive, __a,  __imm, __p) __arm_vqshluq_m_n_s8(__inactive, __a,  __imm, __p)
@@ -1849,13 +1821,6 @@ __arm_vmullbq_poly_p8 (uint8x16_t __a, uint8x16_t __b)
   return __builtin_mve_vmullbq_poly_pv16qi (__a, __b);
 }
 
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq_u16 (uint16x8_t __a, uint16x8_t __b)
-{
-  return __builtin_mve_vmlaldavq_uv8hi (__a, __b);
-}
-
 __extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq_n_u16 (uint16x8_t __a, const int __imm)
@@ -1891,34 +1856,6 @@ __arm_vqdmullbq_n_s16 (int16x8_t __a, int16_t __b)
   return __builtin_mve_vqdmullbq_n_sv8hi (__a, __b);
 }
 
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavxq_s16 (int16x8_t __a, int16x8_t __b)
-{
-  return __builtin_mve_vmlsldavxq_sv8hi (__a, __b);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavq_s16 (int16x8_t __a, int16x8_t __b)
-{
-  return __builtin_mve_vmlsldavq_sv8hi (__a, __b);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavxq_s16 (int16x8_t __a, int16x8_t __b)
-{
-  return __builtin_mve_vmlaldavxq_sv8hi (__a, __b);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq_s16 (int16x8_t __a, int16x8_t __b)
-{
-  return __builtin_mve_vmlaldavq_sv8hi (__a, __b);
-}
-
 __extension__ extern __inline int16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq_n_s16 (int16x8_t __a, const int __imm)
@@ -1940,13 +1877,6 @@ __arm_vmullbq_poly_p16 (uint16x8_t __a, uint16x8_t __b)
   return __builtin_mve_vmullbq_poly_pv8hi (__a, __b);
 }
 
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq_u32 (uint32x4_t __a, uint32x4_t __b)
-{
-  return __builtin_mve_vmlaldavq_uv4si (__a, __b);
-}
-
 __extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq_n_u32 (uint32x4_t __a, const int __imm)
@@ -1982,34 +1912,6 @@ __arm_vqdmullbq_n_s32 (int32x4_t __a, int32_t __b)
   return __builtin_mve_vqdmullbq_n_sv4si (__a, __b);
 }
 
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavxq_s32 (int32x4_t __a, int32x4_t __b)
-{
-  return __builtin_mve_vmlsldavxq_sv4si (__a, __b);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavq_s32 (int32x4_t __a, int32x4_t __b)
-{
-  return __builtin_mve_vmlsldavq_sv4si (__a, __b);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavxq_s32 (int32x4_t __a, int32x4_t __b)
-{
-  return __builtin_mve_vmlaldavxq_sv4si (__a, __b);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq_s32 (int32x4_t __a, int32x4_t __b)
-{
-  return __builtin_mve_vmlaldavq_sv4si (__a, __b);
-}
-
 __extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq_n_s32 (int32x4_t __a, const int __imm)
@@ -2785,34 +2687,6 @@ __arm_vmlsldavaxq_s16 (int64_t __a, int16x8_t __b, int16x8_t __c)
   return __builtin_mve_vmlsldavaxq_sv8hi (__a, __b, __c);
 }
 
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq_p_s16 (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlaldavq_p_sv8hi (__a, __b, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavxq_p_s16 (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlaldavxq_p_sv8hi (__a, __b, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavq_p_s16 (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlsldavq_p_sv8hi (__a, __b, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavxq_p_s16 (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlsldavxq_p_sv8hi (__a, __b, __p);
-}
-
 __extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vmvnq_m_n_u16 (uint16x8_t __inactive, const int __imm, mve_pred16_t __p)
@@ -2827,13 +2701,6 @@ __arm_vmlaldavaq_u16 (uint64_t __a, uint16x8_t __b, uint16x8_t __c)
   return __builtin_mve_vmlaldavaq_uv8hi (__a, __b, __c);
 }
 
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq_p_u16 (uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlaldavq_p_uv8hi (__a, __b, __p);
-}
-
 __extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vmvnq_m_n_s32 (int32x4_t __inactive, const int __imm, mve_pred16_t __p)
@@ -2869,34 +2736,6 @@ __arm_vmlsldavaxq_s32 (int64_t __a, int32x4_t __b, int32x4_t __c)
   return __builtin_mve_vmlsldavaxq_sv4si (__a, __b, __c);
 }
 
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq_p_s32 (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlaldavq_p_sv4si (__a, __b, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavxq_p_s32 (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlaldavxq_p_sv4si (__a, __b, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavq_p_s32 (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlsldavq_p_sv4si (__a, __b, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavxq_p_s32 (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlsldavxq_p_sv4si (__a, __b, __p);
-}
-
 __extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vmvnq_m_n_u32 (uint32x4_t __inactive, const int __imm, mve_pred16_t __p)
@@ -2911,13 +2750,6 @@ __arm_vmlaldavaq_u32 (uint64_t __a, uint32x4_t __b, uint32x4_t __c)
   return __builtin_mve_vmlaldavaq_uv4si (__a, __b, __c);
 }
 
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq_p_u32 (uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlaldavq_p_uv4si (__a, __b, __p);
-}
-
 __extension__ extern __inline int8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vsriq_m_n_s8 (int8x16_t __a, int8x16_t __b, const int __imm, mve_pred16_t __p)
@@ -9139,13 +8971,6 @@ __arm_vmullbq_poly (uint8x16_t __a, uint8x16_t __b)
  return __arm_vmullbq_poly_p8 (__a, __b);
 }
 
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq (uint16x8_t __a, uint16x8_t __b)
-{
- return __arm_vmlaldavq_u16 (__a, __b);
-}
-
 __extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq (uint16x8_t __a, const int __imm)
@@ -9181,34 +9006,6 @@ __arm_vqdmullbq (int16x8_t __a, int16_t __b)
  return __arm_vqdmullbq_n_s16 (__a, __b);
 }
 
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavxq (int16x8_t __a, int16x8_t __b)
-{
- return __arm_vmlsldavxq_s16 (__a, __b);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavq (int16x8_t __a, int16x8_t __b)
-{
- return __arm_vmlsldavq_s16 (__a, __b);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavxq (int16x8_t __a, int16x8_t __b)
-{
- return __arm_vmlaldavxq_s16 (__a, __b);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq (int16x8_t __a, int16x8_t __b)
-{
- return __arm_vmlaldavq_s16 (__a, __b);
-}
-
 __extension__ extern __inline int16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq (int16x8_t __a, const int __imm)
@@ -9230,13 +9027,6 @@ __arm_vmullbq_poly (uint16x8_t __a, uint16x8_t __b)
  return __arm_vmullbq_poly_p16 (__a, __b);
 }
 
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq (uint32x4_t __a, uint32x4_t __b)
-{
- return __arm_vmlaldavq_u32 (__a, __b);
-}
-
 __extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq (uint32x4_t __a, const int __imm)
@@ -9272,34 +9062,6 @@ __arm_vqdmullbq (int32x4_t __a, int32_t __b)
  return __arm_vqdmullbq_n_s32 (__a, __b);
 }
 
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavxq (int32x4_t __a, int32x4_t __b)
-{
- return __arm_vmlsldavxq_s32 (__a, __b);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavq (int32x4_t __a, int32x4_t __b)
-{
- return __arm_vmlsldavq_s32 (__a, __b);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavxq (int32x4_t __a, int32x4_t __b)
-{
- return __arm_vmlaldavxq_s32 (__a, __b);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq (int32x4_t __a, int32x4_t __b)
-{
- return __arm_vmlaldavq_s32 (__a, __b);
-}
-
 __extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq (int32x4_t __a, const int __imm)
@@ -10035,34 +9797,6 @@ __arm_vmlsldavaxq (int64_t __a, int16x8_t __b, int16x8_t __c)
  return __arm_vmlsldavaxq_s16 (__a, __b, __c);
 }
 
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq_p (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vmlaldavq_p_s16 (__a, __b, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavxq_p (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vmlaldavxq_p_s16 (__a, __b, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavq_p (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vmlsldavq_p_s16 (__a, __b, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavxq_p (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vmlsldavxq_p_s16 (__a, __b, __p);
-}
-
 __extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vmvnq_m (uint16x8_t __inactive, const int __imm, mve_pred16_t __p)
@@ -10077,13 +9811,6 @@ __arm_vmlaldavaq (uint64_t __a, uint16x8_t __b, uint16x8_t __c)
  return __arm_vmlaldavaq_u16 (__a, __b, __c);
 }
 
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq_p (uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vmlaldavq_p_u16 (__a, __b, __p);
-}
-
 __extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vmvnq_m (int32x4_t __inactive, const int __imm, mve_pred16_t __p)
@@ -10119,34 +9846,6 @@ __arm_vmlsldavaxq (int64_t __a, int32x4_t __b, int32x4_t __c)
  return __arm_vmlsldavaxq_s32 (__a, __b, __c);
 }
 
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq_p (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vmlaldavq_p_s32 (__a, __b, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavxq_p (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vmlaldavxq_p_s32 (__a, __b, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavq_p (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vmlsldavq_p_s32 (__a, __b, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavxq_p (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vmlsldavxq_p_s32 (__a, __b, __p);
-}
-
 __extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vmvnq_m (uint32x4_t __inactive, const int __imm, mve_pred16_t __p)
@@ -10161,13 +9860,6 @@ __arm_vmlaldavaq (uint64_t __a, uint32x4_t __b, uint32x4_t __c)
  return __arm_vmlaldavaq_u32 (__a, __b, __c);
 }
 
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavq_p (uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vmlaldavq_p_u32 (__a, __b, __p);
-}
-
 __extension__ extern __inline int8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vsriq_m (int8x16_t __a, int8x16_t __b, const int __imm, mve_pred16_t __p)
@@ -15203,12 +14895,6 @@ extern void *__ARM_undef;
   int (*)[__ARM_mve_type_int16x8_t]: __arm_vqshluq_n_s16 (__ARM_mve_coerce(__p0, int16x8_t), p1), \
   int (*)[__ARM_mve_type_int32x4_t]: __arm_vqshluq_n_s32 (__ARM_mve_coerce(__p0, int32x4_t), p1));})
 
-#define __arm_vmlaldavxq(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_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlaldavxq_s16 (__ARM_mve_coerce(__p0, int16x8_t), __ARM_mve_coerce(__p1, int16x8_t)), \
-  int (*)[__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlaldavxq_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t)));})
-
 #define __arm_vqdmulltq(p0,p1) ({ __typeof(p0) __p0 = (p0); \
   __typeof(p1) __p1 = (p1); \
   _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)])0, \
@@ -16272,12 +15958,6 @@ extern void *__ARM_undef;
   int (*)[__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t]: __arm_vmullbq_poly_p8 (__ARM_mve_coerce(__p0, uint8x16_t), __ARM_mve_coerce(__p1, uint8x16_t)), \
   int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vmullbq_poly_p16 (__ARM_mve_coerce(__p0, uint16x8_t), __ARM_mve_coerce(__p1, uint16x8_t)));})
 
-#define __arm_vmlaldavxq(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_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlaldavxq_s16 (__ARM_mve_coerce(__p0, int16x8_t), __ARM_mve_coerce(__p1, int16x8_t)), \
-  int (*)[__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlaldavxq_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t)));})
-
 #define __arm_vqdmulltq(p0,p1) ({ __typeof(p0) __p0 = (p0); \
   __typeof(p1) __p1 = (p1); \
   _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)])0, \
@@ -17460,28 +17140,6 @@ extern void *__ARM_undef;
   int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlaldavaxq_s16 (__ARM_mve_coerce3(p0, int), __ARM_mve_coerce(__p1, int16x8_t), __ARM_mve_coerce(__p2, int16x8_t)), \
   int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlaldavaxq_s32 (__ARM_mve_coerce3(p0, int), __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t)));})
 
-#define __arm_vmlaldavq(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_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlaldavq_s16 (__ARM_mve_coerce(__p0, int16x8_t), __ARM_mve_coerce(__p1, int16x8_t)), \
-  int (*)[__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlaldavq_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t)), \
-  int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vmlaldavq_u16 (__ARM_mve_coerce(__p0, uint16x8_t), __ARM_mve_coerce(__p1, uint16x8_t)), \
-  int (*)[__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vmlaldavq_u32 (__ARM_mve_coerce(__p0, uint32x4_t), __ARM_mve_coerce(__p1, uint32x4_t)));})
-
-#define __arm_vmlaldavq_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_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlaldavq_p_s16 (__ARM_mve_coerce(__p0, int16x8_t), __ARM_mve_coerce(__p1, int16x8_t), p2), \
-  int (*)[__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlaldavq_p_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t), p2), \
-  int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vmlaldavq_p_u16 (__ARM_mve_coerce(__p0, uint16x8_t), __ARM_mve_coerce(__p1, uint16x8_t), p2), \
-  int (*)[__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vmlaldavq_p_u32 (__ARM_mve_coerce(__p0, uint32x4_t), __ARM_mve_coerce(__p1, uint32x4_t), p2));})
-
-#define __arm_vmlaldavxq_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_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlaldavxq_p_s16 (__ARM_mve_coerce(__p0, int16x8_t), __ARM_mve_coerce(__p1, int16x8_t), p2), \
-  int (*)[__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlaldavxq_p_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t), p2));})
-
 #define __arm_vmlsldavaq(p0,p1,p2) ({ __typeof(p0) __p0 = (p0); \
   __typeof(p1) __p1 = (p1); \
   __typeof(p2) __p2 = (p2); \
@@ -17496,30 +17154,6 @@ extern void *__ARM_undef;
   int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlsldavaxq_s16(__p0, __ARM_mve_coerce(__p1, int16x8_t), __ARM_mve_coerce(__p2, int16x8_t)), \
   int (*)[__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlsldavaxq_s32(__p0, __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t)));})
 
-#define __arm_vmlsldavq(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_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlsldavq_s16 (__ARM_mve_coerce(__p0, int16x8_t), __ARM_mve_coerce(__p1, int16x8_t)), \
-  int (*)[__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlsldavq_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t)));})
-
-#define __arm_vmlsldavq_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_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlsldavq_p_s16 (__ARM_mve_coerce(__p0, int16x8_t), __ARM_mve_coerce(__p1, int16x8_t), p2), \
-  int (*)[__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlsldavq_p_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t), p2));})
-
-#define __arm_vmlsldavxq(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_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlsldavxq_s16 (__ARM_mve_coerce(__p0, int16x8_t), __ARM_mve_coerce(__p1, int16x8_t)), \
-  int (*)[__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlsldavxq_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t)));})
-
-#define __arm_vmlsldavxq_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_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlsldavxq_p_s16 (__ARM_mve_coerce(__p0, int16x8_t), __ARM_mve_coerce(__p1, int16x8_t), p2), \
-  int (*)[__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlsldavxq_p_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t), p2));})
-
 #define __arm_vmullbq_int_x(p1,p2,p3) ({ __typeof(p1) __p1 = (p1); \
   __typeof(p2) __p2 = (p2); \
   _Generic( (int (*)[__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \