arm: [MVE intrinsics] rework vmlaldavaq vmlaldavaxq vmlsldavaq vmlsldavaxq

Implement vmlaldavaq, vmlaldavaxq, vmlsldavaq, vmlsldavaxq using the
new MVE builtins framework.

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

	gcc/
	* config/arm/arm-mve-builtins-base.cc (vmlaldavaq, vmlaldavaxq)
	(vmlsldavaq, vmlsldavaxq): New.
	* config/arm/arm-mve-builtins-base.def (vmlaldavaq, vmlaldavaxq)
	(vmlsldavaq, vmlsldavaxq): New.
	* config/arm/arm-mve-builtins-base.h (vmlaldavaq, vmlaldavaxq)
	(vmlsldavaq, vmlsldavaxq): New.
	* config/arm/arm_mve.h (vmlaldavaq): Remove.
	(vmlaldavaxq): Remove.
	(vmlsldavaq): Remove.
	(vmlsldavaxq): Remove.
	(vmlaldavaq_p): Remove.
	(vmlaldavaxq_p): Remove.
	(vmlsldavaq_p): Remove.
	(vmlsldavaxq_p): Remove.
	(vmlaldavaq_s16): Remove.
	(vmlaldavaxq_s16): Remove.
	(vmlsldavaq_s16): Remove.
	(vmlsldavaxq_s16): Remove.
	(vmlaldavaq_u16): Remove.
	(vmlaldavaq_s32): Remove.
	(vmlaldavaxq_s32): Remove.
	(vmlsldavaq_s32): Remove.
	(vmlsldavaxq_s32): Remove.
	(vmlaldavaq_u32): Remove.
	(vmlaldavaq_p_s32): Remove.
	(vmlaldavaq_p_s16): Remove.
	(vmlaldavaq_p_u32): Remove.
	(vmlaldavaq_p_u16): Remove.
	(vmlaldavaxq_p_s32): Remove.
	(vmlaldavaxq_p_s16): Remove.
	(vmlsldavaq_p_s32): Remove.
	(vmlsldavaq_p_s16): Remove.
	(vmlsldavaxq_p_s32): Remove.
	(vmlsldavaxq_p_s16): Remove.
	(__arm_vmlaldavaq_s16): Remove.
	(__arm_vmlaldavaxq_s16): Remove.
	(__arm_vmlsldavaq_s16): Remove.
	(__arm_vmlsldavaxq_s16): Remove.
	(__arm_vmlaldavaq_u16): Remove.
	(__arm_vmlaldavaq_s32): Remove.
	(__arm_vmlaldavaxq_s32): Remove.
	(__arm_vmlsldavaq_s32): Remove.
	(__arm_vmlsldavaxq_s32): Remove.
	(__arm_vmlaldavaq_u32): Remove.
	(__arm_vmlaldavaq_p_s32): Remove.
	(__arm_vmlaldavaq_p_s16): Remove.
	(__arm_vmlaldavaq_p_u32): Remove.
	(__arm_vmlaldavaq_p_u16): Remove.
	(__arm_vmlaldavaxq_p_s32): Remove.
	(__arm_vmlaldavaxq_p_s16): Remove.
	(__arm_vmlsldavaq_p_s32): Remove.
	(__arm_vmlsldavaq_p_s16): Remove.
	(__arm_vmlsldavaxq_p_s32): Remove.
	(__arm_vmlsldavaxq_p_s16): Remove.
	(__arm_vmlaldavaq): Remove.
	(__arm_vmlaldavaxq): Remove.
	(__arm_vmlsldavaq): Remove.
	(__arm_vmlsldavaxq): Remove.
	(__arm_vmlaldavaq_p): Remove.
	(__arm_vmlaldavaxq_p): Remove.
	(__arm_vmlsldavaq_p): Remove.
	(__arm_vmlsldavaxq_p): Remove.

diff --git a/gcc/config/arm/arm-mve-builtins-base.cc b/gcc/config/arm/arm-mve-builtins-base.cc
index 142ba9357a1bdd33b028ac8c5c817626df5c9fdb..2b0c800013c86cdac2009f3f96feabbbe3c64244 100644 (file)
--- a/gcc/config/arm/arm-mve-builtins-base.cc
+++ b/gcc/config/arm/arm-mve-builtins-base.cc
@@ -285,12 +285,16 @@ 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 (vmlaldavaq, VMLALDAVAQ)
+FUNCTION_PRED_P_S (vmlaldavaxq, VMLALDAVAXQ)
 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 (vmlsldavaq, VMLSLDAVAQ)
+FUNCTION_PRED_P_S (vmlsldavaxq, VMLSLDAVAXQ)
 FUNCTION_PRED_P_S (vmlsldavq, VMLSLDAVQ)
 FUNCTION_PRED_P_S (vmlsldavxq, VMLSLDAVXQ)
 FUNCTION_WITHOUT_N_NO_F (vmovlbq, VMOVLBQ)

diff --git a/gcc/config/arm/arm-mve-builtins-base.def b/gcc/config/arm/arm-mve-builtins-base.def
index 1dd3ad3489bdc9d80c84fbc1ded0482797e0c132..d61badb99d93f19ff60ff408bbea7652b168cce8 100644 (file)
--- a/gcc/config/arm/arm-mve-builtins-base.def
+++ b/gcc/config/arm/arm-mve-builtins-base.def
@@ -54,12 +54,16 @@ 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 (vmlaldavaq, binary_acca_int64, integer_16_32, p_or_none)
+DEF_MVE_FUNCTION (vmlaldavaxq, binary_acca_int64, signed_16_32, 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 (vmlsldavaq, binary_acca_int64, signed_16_32, p_or_none)
+DEF_MVE_FUNCTION (vmlsldavaxq, binary_acca_int64, signed_16_32, 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)

diff --git a/gcc/config/arm/arm-mve-builtins-base.h b/gcc/config/arm/arm-mve-builtins-base.h
index 9604991b16865883e969e33352891f862f322cd4..47bb8712fe5379f2453e0a8643c7b011a47b2fc8 100644 (file)
--- a/gcc/config/arm/arm-mve-builtins-base.h
+++ b/gcc/config/arm/arm-mve-builtins-base.h
@@ -67,12 +67,16 @@ 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 vmlaldavaq;
+extern const function_base *const vmlaldavaxq;
 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 vmlsldavaq;
+extern const function_base *const vmlsldavaxq;
 extern const function_base *const vmlsldavq;
 extern const function_base *const vmlsldavxq;
 extern const function_base *const vmovlbq;

diff --git a/gcc/config/arm/arm_mve.h b/gcc/config/arm/arm_mve.h
index e662c0c46770bccb7aa8ccf0866c7b31e5540139..506ac3371e4793c85212fdd23f856e4e300f8c3c 100644 (file)
--- a/gcc/config/arm/arm_mve.h
+++ b/gcc/config/arm/arm_mve.h
@@ -81,10 +81,6 @@
 #define vrmlaldavhaxq(__a, __b, __c) __arm_vrmlaldavhaxq(__a, __b, __c)
 #define vrmlsldavhaq(__a, __b, __c) __arm_vrmlsldavhaq(__a, __b, __c)
 #define vrmlsldavhaxq(__a, __b, __c) __arm_vrmlsldavhaxq(__a, __b, __c)
-#define vmlaldavaq(__a, __b, __c) __arm_vmlaldavaq(__a, __b, __c)
-#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 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)
@@ -111,10 +107,6 @@
 #define vqrdmlsdhq_m(__inactive, __a, __b, __p) __arm_vqrdmlsdhq_m(__inactive, __a, __b, __p)
 #define vqrdmlsdhxq_m(__inactive, __a, __b, __p) __arm_vqrdmlsdhxq_m(__inactive, __a, __b, __p)
 #define vsliq_m(__a, __b, __imm, __p) __arm_vsliq_m(__a, __b, __imm, __p)
-#define vmlaldavaq_p(__a, __b, __c, __p) __arm_vmlaldavaq_p(__a, __b, __c, __p)
-#define vmlaldavaxq_p(__a, __b, __c, __p) __arm_vmlaldavaxq_p(__a, __b, __c, __p)
-#define vmlsldavaq_p(__a, __b, __c, __p) __arm_vmlsldavaq_p(__a, __b, __c, __p)
-#define vmlsldavaxq_p(__a, __b, __c, __p) __arm_vmlsldavaxq_p(__a, __b, __c, __p)
 #define vmullbq_poly_m(__inactive, __a, __b, __p) __arm_vmullbq_poly_m(__inactive, __a, __b, __p)
 #define vmulltq_poly_m(__inactive, __a, __b, __p) __arm_vmulltq_poly_m(__inactive, __a, __b, __p)
 #define vqdmullbq_m(__inactive, __a, __b, __p) __arm_vqdmullbq_m(__inactive, __a, __b, __p)
@@ -524,10 +516,6 @@
 #define vfmaq_n_f16(__a, __b, __c) __arm_vfmaq_n_f16(__a, __b, __c)
 #define vfmasq_n_f16(__a, __b, __c) __arm_vfmasq_n_f16(__a, __b, __c)
 #define vfmsq_f16(__a, __b, __c) __arm_vfmsq_f16(__a, __b, __c)
-#define vmlaldavaq_s16(__a, __b, __c) __arm_vmlaldavaq_s16(__a, __b, __c)
-#define vmlaldavaxq_s16(__a, __b, __c) __arm_vmlaldavaxq_s16(__a, __b, __c)
-#define vmlsldavaq_s16(__a, __b, __c) __arm_vmlsldavaq_s16(__a, __b, __c)
-#define vmlsldavaxq_s16(__a, __b, __c) __arm_vmlsldavaxq_s16(__a, __b, __c)
 #define vcvtmq_m_s16_f16(__inactive, __a, __p) __arm_vcvtmq_m_s16_f16(__inactive, __a, __p)
 #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)
@@ -538,7 +526,6 @@
 #define vcvtnq_m_u16_f16(__inactive, __a, __p) __arm_vcvtnq_m_u16_f16(__inactive, __a, __p)
 #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 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)
@@ -548,10 +535,6 @@
 #define vfmaq_n_f32(__a, __b, __c) __arm_vfmaq_n_f32(__a, __b, __c)
 #define vfmasq_n_f32(__a, __b, __c) __arm_vfmasq_n_f32(__a, __b, __c)
 #define vfmsq_f32(__a, __b, __c) __arm_vfmsq_f32(__a, __b, __c)
-#define vmlaldavaq_s32(__a, __b, __c) __arm_vmlaldavaq_s32(__a, __b, __c)
-#define vmlaldavaxq_s32(__a, __b, __c) __arm_vmlaldavaxq_s32(__a, __b, __c)
-#define vmlsldavaq_s32(__a, __b, __c) __arm_vmlsldavaq_s32(__a, __b, __c)
-#define vmlsldavaxq_s32(__a, __b, __c) __arm_vmlsldavaxq_s32(__a, __b, __c)
 #define vcvtmq_m_s32_f32(__inactive, __a, __p) __arm_vcvtmq_m_s32_f32(__inactive, __a, __p)
 #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)
@@ -562,7 +545,6 @@
 #define vcvtnq_m_u32_f32(__inactive, __a, __p) __arm_vcvtnq_m_u32_f32(__inactive, __a, __p)
 #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 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)
@@ -678,16 +660,6 @@
 #define vsliq_m_n_u8(__a, __b,  __imm, __p) __arm_vsliq_m_n_u8(__a, __b,  __imm, __p)
 #define vsliq_m_n_u32(__a, __b,  __imm, __p) __arm_vsliq_m_n_u32(__a, __b,  __imm, __p)
 #define vsliq_m_n_u16(__a, __b,  __imm, __p) __arm_vsliq_m_n_u16(__a, __b,  __imm, __p)
-#define vmlaldavaq_p_s32(__a, __b, __c, __p) __arm_vmlaldavaq_p_s32(__a, __b, __c, __p)
-#define vmlaldavaq_p_s16(__a, __b, __c, __p) __arm_vmlaldavaq_p_s16(__a, __b, __c, __p)
-#define vmlaldavaq_p_u32(__a, __b, __c, __p) __arm_vmlaldavaq_p_u32(__a, __b, __c, __p)
-#define vmlaldavaq_p_u16(__a, __b, __c, __p) __arm_vmlaldavaq_p_u16(__a, __b, __c, __p)
-#define vmlaldavaxq_p_s32(__a, __b, __c, __p) __arm_vmlaldavaxq_p_s32(__a, __b, __c, __p)
-#define vmlaldavaxq_p_s16(__a, __b, __c, __p) __arm_vmlaldavaxq_p_s16(__a, __b, __c, __p)
-#define vmlsldavaq_p_s32(__a, __b, __c, __p) __arm_vmlsldavaq_p_s32(__a, __b, __c, __p)
-#define vmlsldavaq_p_s16(__a, __b, __c, __p) __arm_vmlsldavaq_p_s16(__a, __b, __c, __p)
-#define vmlsldavaxq_p_s32(__a, __b, __c, __p) __arm_vmlsldavaxq_p_s32(__a, __b, __c, __p)
-#define vmlsldavaxq_p_s16(__a, __b, __c, __p) __arm_vmlsldavaxq_p_s16(__a, __b, __c, __p)
 #define vmullbq_poly_m_p8(__inactive, __a, __b, __p) __arm_vmullbq_poly_m_p8(__inactive, __a, __b, __p)
 #define vmullbq_poly_m_p16(__inactive, __a, __b, __p) __arm_vmullbq_poly_m_p16(__inactive, __a, __b, __p)
 #define vmulltq_poly_m_p8(__inactive, __a, __b, __p) __arm_vmulltq_poly_m_p8(__inactive, __a, __b, __p)
@@ -2571,34 +2543,6 @@ __arm_vmvnq_m_n_s16 (int16x8_t __inactive, const int __imm, mve_pred16_t __p)
   return __builtin_mve_vmvnq_m_n_sv8hi (__inactive, __imm, __p);
 }
 
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq_s16 (int64_t __a, int16x8_t __b, int16x8_t __c)
-{
-  return __builtin_mve_vmlaldavaq_sv8hi (__a, __b, __c);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaxq_s16 (int64_t __a, int16x8_t __b, int16x8_t __c)
-{
-  return __builtin_mve_vmlaldavaxq_sv8hi (__a, __b, __c);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaq_s16 (int64_t __a, int16x8_t __b, int16x8_t __c)
-{
-  return __builtin_mve_vmlsldavaq_sv8hi (__a, __b, __c);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaxq_s16 (int64_t __a, int16x8_t __b, int16x8_t __c)
-{
-  return __builtin_mve_vmlsldavaxq_sv8hi (__a, __b, __c);
-}
-
 __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)
@@ -2606,13 +2550,6 @@ __arm_vmvnq_m_n_u16 (uint16x8_t __inactive, const int __imm, mve_pred16_t __p)
   return __builtin_mve_vmvnq_m_n_uv8hi (__inactive, __imm, __p);
 }
 
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq_u16 (uint64_t __a, uint16x8_t __b, uint16x8_t __c)
-{
-  return __builtin_mve_vmlaldavaq_uv8hi (__a, __b, __c);
-}
-
 __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)
@@ -2620,34 +2557,6 @@ __arm_vmvnq_m_n_s32 (int32x4_t __inactive, const int __imm, mve_pred16_t __p)
   return __builtin_mve_vmvnq_m_n_sv4si (__inactive, __imm, __p);
 }
 
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq_s32 (int64_t __a, int32x4_t __b, int32x4_t __c)
-{
-  return __builtin_mve_vmlaldavaq_sv4si (__a, __b, __c);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaxq_s32 (int64_t __a, int32x4_t __b, int32x4_t __c)
-{
-  return __builtin_mve_vmlaldavaxq_sv4si (__a, __b, __c);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaq_s32 (int64_t __a, int32x4_t __b, int32x4_t __c)
-{
-  return __builtin_mve_vmlsldavaq_sv4si (__a, __b, __c);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaxq_s32 (int64_t __a, int32x4_t __b, int32x4_t __c)
-{
-  return __builtin_mve_vmlsldavaxq_sv4si (__a, __b, __c);
-}
-
 __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)
@@ -2655,13 +2564,6 @@ __arm_vmvnq_m_n_u32 (uint32x4_t __inactive, const int __imm, mve_pred16_t __p)
   return __builtin_mve_vmvnq_m_n_uv4si (__inactive, __imm, __p);
 }
 
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq_u32 (uint64_t __a, uint32x4_t __b, uint32x4_t __c)
-{
-  return __builtin_mve_vmlaldavaq_uv4si (__a, __b, __c);
-}
-
 __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)
@@ -3439,76 +3341,6 @@ __arm_vsliq_m_n_u16 (uint16x8_t __a, uint16x8_t __b, const int __imm, mve_pred16
   return __builtin_mve_vsliq_m_n_uv8hi (__a, __b, __imm, __p);
 }
 
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq_p_s32 (int64_t __a, int32x4_t __b, int32x4_t __c, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlaldavaq_p_sv4si (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq_p_s16 (int64_t __a, int16x8_t __b, int16x8_t __c, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlaldavaq_p_sv8hi (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq_p_u32 (uint64_t __a, uint32x4_t __b, uint32x4_t __c, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlaldavaq_p_uv4si (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq_p_u16 (uint64_t __a, uint16x8_t __b, uint16x8_t __c, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlaldavaq_p_uv8hi (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaxq_p_s32 (int64_t __a, int32x4_t __b, int32x4_t __c, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlaldavaxq_p_sv4si (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaxq_p_s16 (int64_t __a, int16x8_t __b, int16x8_t __c, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlaldavaxq_p_sv8hi (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaq_p_s32 (int64_t __a, int32x4_t __b, int32x4_t __c, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlsldavaq_p_sv4si (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaq_p_s16 (int64_t __a, int16x8_t __b, int16x8_t __c, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlsldavaq_p_sv8hi (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaxq_p_s32 (int64_t __a, int32x4_t __b, int32x4_t __c, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlsldavaxq_p_sv4si (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaxq_p_s16 (int64_t __a, int16x8_t __b, int16x8_t __c, mve_pred16_t __p)
-{
-  return __builtin_mve_vmlsldavaxq_p_sv8hi (__a, __b, __c, __p);
-}
-
 __extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vmullbq_poly_m_p8 (uint16x8_t __inactive, uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
@@ -9611,34 +9443,6 @@ __arm_vmvnq_m (int16x8_t __inactive, const int __imm, mve_pred16_t __p)
  return __arm_vmvnq_m_n_s16 (__inactive, __imm, __p);
 }
 
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq (int64_t __a, int16x8_t __b, int16x8_t __c)
-{
- return __arm_vmlaldavaq_s16 (__a, __b, __c);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaxq (int64_t __a, int16x8_t __b, int16x8_t __c)
-{
- return __arm_vmlaldavaxq_s16 (__a, __b, __c);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaq (int64_t __a, int16x8_t __b, int16x8_t __c)
-{
- return __arm_vmlsldavaq_s16 (__a, __b, __c);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaxq (int64_t __a, int16x8_t __b, int16x8_t __c)
-{
- return __arm_vmlsldavaxq_s16 (__a, __b, __c);
-}
-
 __extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vmvnq_m (uint16x8_t __inactive, const int __imm, mve_pred16_t __p)
@@ -9646,13 +9450,6 @@ __arm_vmvnq_m (uint16x8_t __inactive, const int __imm, mve_pred16_t __p)
  return __arm_vmvnq_m_n_u16 (__inactive, __imm, __p);
 }
 
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq (uint64_t __a, uint16x8_t __b, uint16x8_t __c)
-{
- return __arm_vmlaldavaq_u16 (__a, __b, __c);
-}
-
 __extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vmvnq_m (int32x4_t __inactive, const int __imm, mve_pred16_t __p)
@@ -9660,34 +9457,6 @@ __arm_vmvnq_m (int32x4_t __inactive, const int __imm, mve_pred16_t __p)
  return __arm_vmvnq_m_n_s32 (__inactive, __imm, __p);
 }
 
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq (int64_t __a, int32x4_t __b, int32x4_t __c)
-{
- return __arm_vmlaldavaq_s32 (__a, __b, __c);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaxq (int64_t __a, int32x4_t __b, int32x4_t __c)
-{
- return __arm_vmlaldavaxq_s32 (__a, __b, __c);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaq (int64_t __a, int32x4_t __b, int32x4_t __c)
-{
- return __arm_vmlsldavaq_s32 (__a, __b, __c);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaxq (int64_t __a, int32x4_t __b, int32x4_t __c)
-{
- return __arm_vmlsldavaxq_s32 (__a, __b, __c);
-}
-
 __extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vmvnq_m (uint32x4_t __inactive, const int __imm, mve_pred16_t __p)
@@ -9695,13 +9464,6 @@ __arm_vmvnq_m (uint32x4_t __inactive, const int __imm, mve_pred16_t __p)
  return __arm_vmvnq_m_n_u32 (__inactive, __imm, __p);
 }
 
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq (uint64_t __a, uint32x4_t __b, uint32x4_t __c)
-{
- return __arm_vmlaldavaq_u32 (__a, __b, __c);
-}
-
 __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)
@@ -10479,76 +10241,6 @@ __arm_vsliq_m (uint16x8_t __a, uint16x8_t __b, const int __imm, mve_pred16_t __p
  return __arm_vsliq_m_n_u16 (__a, __b, __imm, __p);
 }
 
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq_p (int64_t __a, int32x4_t __b, int32x4_t __c, mve_pred16_t __p)
-{
- return __arm_vmlaldavaq_p_s32 (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq_p (int64_t __a, int16x8_t __b, int16x8_t __c, mve_pred16_t __p)
-{
- return __arm_vmlaldavaq_p_s16 (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq_p (uint64_t __a, uint32x4_t __b, uint32x4_t __c, mve_pred16_t __p)
-{
- return __arm_vmlaldavaq_p_u32 (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline uint64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaq_p (uint64_t __a, uint16x8_t __b, uint16x8_t __c, mve_pred16_t __p)
-{
- return __arm_vmlaldavaq_p_u16 (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaxq_p (int64_t __a, int32x4_t __b, int32x4_t __c, mve_pred16_t __p)
-{
- return __arm_vmlaldavaxq_p_s32 (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlaldavaxq_p (int64_t __a, int16x8_t __b, int16x8_t __c, mve_pred16_t __p)
-{
- return __arm_vmlaldavaxq_p_s16 (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaq_p (int64_t __a, int32x4_t __b, int32x4_t __c, mve_pred16_t __p)
-{
- return __arm_vmlsldavaq_p_s32 (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaq_p (int64_t __a, int16x8_t __b, int16x8_t __c, mve_pred16_t __p)
-{
- return __arm_vmlsldavaq_p_s16 (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaxq_p (int64_t __a, int32x4_t __b, int32x4_t __c, mve_pred16_t __p)
-{
- return __arm_vmlsldavaxq_p_s32 (__a, __b, __c, __p);
-}
-
-__extension__ extern __inline int64_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vmlsldavaxq_p (int64_t __a, int16x8_t __b, int16x8_t __c, mve_pred16_t __p)
-{
- return __arm_vmlsldavaxq_p_s16 (__a, __b, __c, __p);
-}
-
 __extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vmullbq_poly_m (uint16x8_t __inactive, uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
@@ -16633,36 +16325,6 @@ extern void *__ARM_undef;
   int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vqrdmlsdhq_m_s16 (__ARM_mve_coerce(__p0, int16x8_t), __ARM_mve_coerce(__p1, int16x8_t), __ARM_mve_coerce(__p2, int16x8_t), p3), \
   int (*)[__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vqrdmlsdhq_m_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3));})
 
-#define __arm_vmlaldavaq_p(p0,p1,p2,p3) ({ __typeof(p0) __p0 = (p0); \
-  __typeof(p1) __p1 = (p1); \
-  __typeof(p2) __p2 = (p2); \
-  _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
-  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlaldavaq_p_s16 (__ARM_mve_coerce3(p0, int), __ARM_mve_coerce(__p1, int16x8_t), __ARM_mve_coerce(__p2, int16x8_t), p3), \
-  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlaldavaq_p_s32 (__ARM_mve_coerce3(p0, int), __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3), \
-  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vmlaldavaq_p_u16 (__ARM_mve_coerce3(p0, int), __ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, uint16x8_t), p3), \
-  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vmlaldavaq_p_u32 (__ARM_mve_coerce3(p0, int), __ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t), p3));})
-
-#define __arm_vmlaldavaxq_p(p0,p1,p2,p3) ({ __typeof(p0) __p0 = (p0); \
-  __typeof(p1) __p1 = (p1); \
-  __typeof(p2) __p2 = (p2); \
-  _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
-  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlaldavaxq_p_s16 (__ARM_mve_coerce3(p0, int), __ARM_mve_coerce(__p1, int16x8_t), __ARM_mve_coerce(__p2, int16x8_t), p3), \
-  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlaldavaxq_p_s32 (__ARM_mve_coerce3(p0, int), __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3));})
-
-#define __arm_vmlsldavaq_p(p0,p1,p2,p3) ({ __typeof(p0) __p0 = (p0); \
-  __typeof(p1) __p1 = (p1); \
-  __typeof(p2) __p2 = (p2); \
-  _Generic( (int (*)[__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlsldavaq_p_s16(__p0, __ARM_mve_coerce(__p1, int16x8_t), __ARM_mve_coerce(__p2, int16x8_t), p3), \
-  int (*)[__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlsldavaq_p_s32(__p0, __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3));})
-
-#define __arm_vmlsldavaxq_p(p0,p1,p2,p3) ({ __typeof(p0) __p0 = (p0); \
-  __typeof(p1) __p1 = (p1); \
-  __typeof(p2) __p2 = (p2); \
-  _Generic( (int (*)[__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlsldavaxq_p_s16(__p0, __ARM_mve_coerce(__p1, int16x8_t), __ARM_mve_coerce(__p2, int16x8_t), p3), \
-  int (*)[__ARM_mve_type_int32x4_t][__ARM_mve_type_int32x4_t]: __arm_vmlsldavaxq_p_s32(__p0, __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3));})
-
 #define __arm_vrmlaldavhaxq_p(p0,p1,p2,p3) __arm_vrmlaldavhaxq_p_s32(p0,p1,p2,p3)
 
 #define __arm_vrmlsldavhaq_p(p0,p1,p2,p3) __arm_vrmlsldavhaq_p_s32(p0,p1,p2,p3)
@@ -16966,36 +16628,6 @@ extern void *__ARM_undef;
   int (*)[__ARM_mve_type_uint16x8_t]: __arm_vshlcq_m_u16 (__ARM_mve_coerce(__p0, uint16x8_t), p1, p2, p3), \
   int (*)[__ARM_mve_type_uint32x4_t]: __arm_vshlcq_m_u32 (__ARM_mve_coerce(__p0, uint32x4_t), p1, p2, p3));})
 
-#define __arm_vmlaldavaq(p0,p1,p2) ({ __typeof(p0) __p0 = (p0); \
-  __typeof(p1) __p1 = (p1); \
-  __typeof(p2) __p2 = (p2); \
-  _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
-  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlaldavaq_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_vmlaldavaq_s32 (__ARM_mve_coerce3(p0, int), __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t)), \
-  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vmlaldavaq_u16 (__ARM_mve_coerce3(p0, int), __ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, uint16x8_t)), \
-  int (*)[__ARM_mve_type_int_n][__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vmlaldavaq_u32 (__ARM_mve_coerce3(p0, int), __ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t)));})
-
-#define __arm_vmlaldavaxq(p0,p1,p2) ({ __typeof(p0) __p0 = (p0); \
-  __typeof(p1) __p1 = (p1); \
-  __typeof(p2) __p2 = (p2); \
-  _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
-  int (*)[__ARM_mve_type_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_vmlsldavaq(p0,p1,p2) ({ __typeof(p0) __p0 = (p0); \
-  __typeof(p1) __p1 = (p1); \
-  __typeof(p2) __p2 = (p2); \
-  _Generic( (int (*)[__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vmlsldavaq_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_vmlsldavaq_s32(__p0, __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t)));})
-
-#define __arm_vmlsldavaxq(p0,p1,p2) ({ __typeof(p0) __p0 = (p0); \
-  __typeof(p1) __p1 = (p1); \
-  __typeof(p2) __p2 = (p2); \
-  _Generic( (int (*)[__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
-  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_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, \