diff mbox series

[2/6] arm: [MVE intrinsics] rework vcaddq vhcaddq

Message ID	20230713102224.1161596-2-christophe.lyon@linaro.org
State	Unresolved
Headers	Received-SPF: pass (google.com: domain of gcc-patches-bounces+ouuuleilei=gmail.com@gcc.gnu.org designates 8.43.85.97 as permitted sender) client-ip=8.43.85.97; DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org BDADF3858035 To: gcc-patches@gcc.gnu.org, Kyrylo.Tkachov@arm.com, richard.earnshaw@arm.com, richard.sandiford@arm.com Cc: Christophe Lyon <christophe.lyon@linaro.org> Subject: [PATCH 2/6] arm: [MVE intrinsics] rework vcaddq vhcaddq Date: Thu, 13 Jul 2023 10:22:20 +0000 Message-Id: <20230713102224.1161596-2-christophe.lyon@linaro.org> In-Reply-To: <20230713102224.1161596-1-christophe.lyon@linaro.org> References: <20230713102224.1161596-1-christophe.lyon@linaro.org> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Precedence: list From: Christophe Lyon via Gcc-patches <gcc-patches@gcc.gnu.org> Reply-To: Christophe Lyon <christophe.lyon@linaro.org> Errors-To: gcc-patches-bounces+ouuuleilei=gmail.com@gcc.gnu.org Sender: "Gcc-patches" <gcc-patches-bounces+ouuuleilei=gmail.com@gcc.gnu.org> X-getmail-retrieved-from-mailbox: INBOX
Series	[1/6] arm: [MVE intrinsics] Factorize vcaddq vhcaddq \| [1/6] arm: [MVE intrinsics] Factorize vcaddq vhcaddq [2/6] arm: [MVE intrinsics] rework vcaddq vhcaddq [3/6] arm: [MVE intrinsics factorize vcmulq [4/6] arm: [MVE intrinsics] rework vcmulq [5/6] arm: [MVE intrinsics] factorize vcmlaq [6/6] arm: [MVE intrinsics] rework vcmlaq

Checks

Context	Check	Description
snail/gcc-patch-check	warning	Git am fail log

Commit Message

Christophe Lyon July 13, 2023, 10:22 a.m. UTC

  Implement vcaddq, vhcaddq using the new MVE builtins framework.

2023-07-13  Christophe Lyon  <christophe.lyon@linaro.org>

	gcc/
	* config/arm/arm-mve-builtins-base.cc (vcaddq_rot90)
	(vcaddq_rot270, vhcaddq_rot90, vhcaddq_rot270): New.
	* config/arm/arm-mve-builtins-base.def (vcaddq_rot90)
	(vcaddq_rot270, vhcaddq_rot90, vhcaddq_rot270): New.
	* config/arm/arm-mve-builtins-base.h: (vcaddq_rot90)
	(vcaddq_rot270, vhcaddq_rot90, vhcaddq_rot270): New.
	* config/arm/arm-mve-builtins-functions.h (class
	unspec_mve_function_exact_insn_rot): New.
	* config/arm/arm_mve.h (vcaddq_rot90): Delete.
	(vcaddq_rot270): Delete.
	(vhcaddq_rot90): Delete.
	(vhcaddq_rot270): Delete.
	(vcaddq_rot270_m): Delete.
	(vcaddq_rot90_m): Delete.
	(vhcaddq_rot270_m): Delete.
	(vhcaddq_rot90_m): Delete.
	(vcaddq_rot90_x): Delete.
	(vcaddq_rot270_x): Delete.
	(vhcaddq_rot90_x): Delete.
	(vhcaddq_rot270_x): Delete.
	(vcaddq_rot90_u8): Delete.
	(vcaddq_rot270_u8): Delete.
	(vhcaddq_rot90_s8): Delete.
	(vhcaddq_rot270_s8): Delete.
	(vcaddq_rot90_s8): Delete.
	(vcaddq_rot270_s8): Delete.
	(vcaddq_rot90_u16): Delete.
	(vcaddq_rot270_u16): Delete.
	(vhcaddq_rot90_s16): Delete.
	(vhcaddq_rot270_s16): Delete.
	(vcaddq_rot90_s16): Delete.
	(vcaddq_rot270_s16): Delete.
	(vcaddq_rot90_u32): Delete.
	(vcaddq_rot270_u32): Delete.
	(vhcaddq_rot90_s32): Delete.
	(vhcaddq_rot270_s32): Delete.
	(vcaddq_rot90_s32): Delete.
	(vcaddq_rot270_s32): Delete.
	(vcaddq_rot90_f16): Delete.
	(vcaddq_rot270_f16): Delete.
	(vcaddq_rot90_f32): Delete.
	(vcaddq_rot270_f32): Delete.
	(vcaddq_rot270_m_s8): Delete.
	(vcaddq_rot270_m_s32): Delete.
	(vcaddq_rot270_m_s16): Delete.
	(vcaddq_rot270_m_u8): Delete.
	(vcaddq_rot270_m_u32): Delete.
	(vcaddq_rot270_m_u16): Delete.
	(vcaddq_rot90_m_s8): Delete.
	(vcaddq_rot90_m_s32): Delete.
	(vcaddq_rot90_m_s16): Delete.
	(vcaddq_rot90_m_u8): Delete.
	(vcaddq_rot90_m_u32): Delete.
	(vcaddq_rot90_m_u16): Delete.
	(vhcaddq_rot270_m_s8): Delete.
	(vhcaddq_rot270_m_s32): Delete.
	(vhcaddq_rot270_m_s16): Delete.
	(vhcaddq_rot90_m_s8): Delete.
	(vhcaddq_rot90_m_s32): Delete.
	(vhcaddq_rot90_m_s16): Delete.
	(vcaddq_rot270_m_f32): Delete.
	(vcaddq_rot270_m_f16): Delete.
	(vcaddq_rot90_m_f32): Delete.
	(vcaddq_rot90_m_f16): Delete.
	(vcaddq_rot90_x_s8): Delete.
	(vcaddq_rot90_x_s16): Delete.
	(vcaddq_rot90_x_s32): Delete.
	(vcaddq_rot90_x_u8): Delete.
	(vcaddq_rot90_x_u16): Delete.
	(vcaddq_rot90_x_u32): Delete.
	(vcaddq_rot270_x_s8): Delete.
	(vcaddq_rot270_x_s16): Delete.
	(vcaddq_rot270_x_s32): Delete.
	(vcaddq_rot270_x_u8): Delete.
	(vcaddq_rot270_x_u16): Delete.
	(vcaddq_rot270_x_u32): Delete.
	(vhcaddq_rot90_x_s8): Delete.
	(vhcaddq_rot90_x_s16): Delete.
	(vhcaddq_rot90_x_s32): Delete.
	(vhcaddq_rot270_x_s8): Delete.
	(vhcaddq_rot270_x_s16): Delete.
	(vhcaddq_rot270_x_s32): Delete.
	(vcaddq_rot90_x_f16): Delete.
	(vcaddq_rot90_x_f32): Delete.
	(vcaddq_rot270_x_f16): Delete.
	(vcaddq_rot270_x_f32): Delete.
	(__arm_vcaddq_rot90_u8): Delete.
	(__arm_vcaddq_rot270_u8): Delete.
	(__arm_vhcaddq_rot90_s8): Delete.
	(__arm_vhcaddq_rot270_s8): Delete.
	(__arm_vcaddq_rot90_s8): Delete.
	(__arm_vcaddq_rot270_s8): Delete.
	(__arm_vcaddq_rot90_u16): Delete.
	(__arm_vcaddq_rot270_u16): Delete.
	(__arm_vhcaddq_rot90_s16): Delete.
	(__arm_vhcaddq_rot270_s16): Delete.
	(__arm_vcaddq_rot90_s16): Delete.
	(__arm_vcaddq_rot270_s16): Delete.
	(__arm_vcaddq_rot90_u32): Delete.
	(__arm_vcaddq_rot270_u32): Delete.
	(__arm_vhcaddq_rot90_s32): Delete.
	(__arm_vhcaddq_rot270_s32): Delete.
	(__arm_vcaddq_rot90_s32): Delete.
	(__arm_vcaddq_rot270_s32): Delete.
	(__arm_vcaddq_rot270_m_s8): Delete.
	(__arm_vcaddq_rot270_m_s32): Delete.
	(__arm_vcaddq_rot270_m_s16): Delete.
	(__arm_vcaddq_rot270_m_u8): Delete.
	(__arm_vcaddq_rot270_m_u32): Delete.
	(__arm_vcaddq_rot270_m_u16): Delete.
	(__arm_vcaddq_rot90_m_s8): Delete.
	(__arm_vcaddq_rot90_m_s32): Delete.
	(__arm_vcaddq_rot90_m_s16): Delete.
	(__arm_vcaddq_rot90_m_u8): Delete.
	(__arm_vcaddq_rot90_m_u32): Delete.
	(__arm_vcaddq_rot90_m_u16): Delete.
	(__arm_vhcaddq_rot270_m_s8): Delete.
	(__arm_vhcaddq_rot270_m_s32): Delete.
	(__arm_vhcaddq_rot270_m_s16): Delete.
	(__arm_vhcaddq_rot90_m_s8): Delete.
	(__arm_vhcaddq_rot90_m_s32): Delete.
	(__arm_vhcaddq_rot90_m_s16): Delete.
	(__arm_vcaddq_rot90_x_s8): Delete.
	(__arm_vcaddq_rot90_x_s16): Delete.
	(__arm_vcaddq_rot90_x_s32): Delete.
	(__arm_vcaddq_rot90_x_u8): Delete.
	(__arm_vcaddq_rot90_x_u16): Delete.
	(__arm_vcaddq_rot90_x_u32): Delete.
	(__arm_vcaddq_rot270_x_s8): Delete.
	(__arm_vcaddq_rot270_x_s16): Delete.
	(__arm_vcaddq_rot270_x_s32): Delete.
	(__arm_vcaddq_rot270_x_u8): Delete.
	(__arm_vcaddq_rot270_x_u16): Delete.
	(__arm_vcaddq_rot270_x_u32): Delete.
	(__arm_vhcaddq_rot90_x_s8): Delete.
	(__arm_vhcaddq_rot90_x_s16): Delete.
	(__arm_vhcaddq_rot90_x_s32): Delete.
	(__arm_vhcaddq_rot270_x_s8): Delete.
	(__arm_vhcaddq_rot270_x_s16): Delete.
	(__arm_vhcaddq_rot270_x_s32): Delete.
	(__arm_vcaddq_rot90_f16): Delete.
	(__arm_vcaddq_rot270_f16): Delete.
	(__arm_vcaddq_rot90_f32): Delete.
	(__arm_vcaddq_rot270_f32): Delete.
	(__arm_vcaddq_rot270_m_f32): Delete.
	(__arm_vcaddq_rot270_m_f16): Delete.
	(__arm_vcaddq_rot90_m_f32): Delete.
	(__arm_vcaddq_rot90_m_f16): Delete.
	(__arm_vcaddq_rot90_x_f16): Delete.
	(__arm_vcaddq_rot90_x_f32): Delete.
	(__arm_vcaddq_rot270_x_f16): Delete.
	(__arm_vcaddq_rot270_x_f32): Delete.
	(__arm_vcaddq_rot90): Delete.
	(__arm_vcaddq_rot270): Delete.
	(__arm_vhcaddq_rot90): Delete.
	(__arm_vhcaddq_rot270): Delete.
	(__arm_vcaddq_rot270_m): Delete.
	(__arm_vcaddq_rot90_m): Delete.
	(__arm_vhcaddq_rot270_m): Delete.
	(__arm_vhcaddq_rot90_m): Delete.
	(__arm_vcaddq_rot90_x): Delete.
	(__arm_vcaddq_rot270_x): Delete.
	(__arm_vhcaddq_rot90_x): Delete.
	(__arm_vhcaddq_rot270_x): Delete.
---
 gcc/config/arm/arm-mve-builtins-base.cc     |    4 +
 gcc/config/arm/arm-mve-builtins-base.def    |    6 +
 gcc/config/arm/arm-mve-builtins-base.h      |    4 +
 gcc/config/arm/arm-mve-builtins-functions.h |  103 ++
 gcc/config/arm/arm_mve.h                    | 1398 ++-----------------
 5 files changed, 215 insertions(+), 1300 deletions(-)

diff mbox series

Patch

diff --git a/gcc/config/arm/arm-mve-builtins-base.cc b/gcc/config/arm/arm-mve-builtins-base.cc
index af02397f1c4..f15bb926147 100644
--- a/gcc/config/arm/arm-mve-builtins-base.cc
+++ b/gcc/config/arm/arm-mve-builtins-base.cc
@@ -260,6 +260,10 @@  FUNCTION_PRED_P_S_U (vaddvq, VADDVQ)
 FUNCTION_PRED_P_S_U (vaddvaq, VADDVAQ)
 FUNCTION_WITH_RTX_M (vandq, AND, VANDQ)
 FUNCTION_ONLY_N (vbrsrq, VBRSRQ)
+FUNCTION (vcaddq_rot90, unspec_mve_function_exact_insn_rot, (UNSPEC_VCADD90, UNSPEC_VCADD90, UNSPEC_VCADD90, VCADDQ_ROT90_M_S, VCADDQ_ROT90_M_U, VCADDQ_ROT90_M_F))
+FUNCTION (vcaddq_rot270, unspec_mve_function_exact_insn_rot, (UNSPEC_VCADD270, UNSPEC_VCADD270, UNSPEC_VCADD270, VCADDQ_ROT270_M_S, VCADDQ_ROT270_M_U, VCADDQ_ROT270_M_F))
+FUNCTION (vhcaddq_rot90, unspec_mve_function_exact_insn_rot, (VHCADDQ_ROT90_S, -1, -1, VHCADDQ_ROT90_M_S, -1, -1))
+FUNCTION (vhcaddq_rot270, unspec_mve_function_exact_insn_rot, (VHCADDQ_ROT270_S, -1, -1, VHCADDQ_ROT270_M_S, -1, -1))
 FUNCTION_WITHOUT_N_NO_U_F (vclsq, VCLSQ)
 FUNCTION (vclzq, unspec_based_mve_function_exact_insn, (CLZ, CLZ, CLZ, -1, -1, -1, VCLZQ_M_S, VCLZQ_M_U, -1, -1, -1 ,-1))
 FUNCTION (vcmpeqq, unspec_based_mve_function_exact_insn_vcmp, (EQ, EQ, EQ, VCMPEQQ_M_S, VCMPEQQ_M_U, VCMPEQQ_M_F, VCMPEQQ_M_N_S, VCMPEQQ_M_N_U, VCMPEQQ_M_N_F))
diff --git a/gcc/config/arm/arm-mve-builtins-base.def b/gcc/config/arm/arm-mve-builtins-base.def
index ee08d063407..9a793147960 100644
--- a/gcc/config/arm/arm-mve-builtins-base.def
+++ b/gcc/config/arm/arm-mve-builtins-base.def
@@ -28,6 +28,8 @@  DEF_MVE_FUNCTION (vaddvaq, unary_int32_acc, all_integer, p_or_none)
 DEF_MVE_FUNCTION (vaddvq, unary_int32, all_integer, p_or_none)
 DEF_MVE_FUNCTION (vandq, binary, all_integer, mx_or_none)
 DEF_MVE_FUNCTION (vbrsrq, binary_imm32, all_integer, mx_or_none)
+DEF_MVE_FUNCTION (vcaddq_rot90, binary, all_integer, mx_or_none)
+DEF_MVE_FUNCTION (vcaddq_rot270, binary, all_integer, mx_or_none)
 DEF_MVE_FUNCTION (vclsq, unary, all_signed, mx_or_none)
 DEF_MVE_FUNCTION (vclzq, unary, all_integer, mx_or_none)
 DEF_MVE_FUNCTION (vcmpcsq, cmp, all_unsigned, m_or_none)
@@ -42,6 +44,8 @@  DEF_MVE_FUNCTION (vcreateq, create, all_integer_with_64, none)
 DEF_MVE_FUNCTION (vdupq, unary_n, all_integer, mx_or_none)
 DEF_MVE_FUNCTION (veorq, binary, all_integer, mx_or_none)
 DEF_MVE_FUNCTION (vhaddq, binary_opt_n, all_integer, mx_or_none)
+DEF_MVE_FUNCTION (vhcaddq_rot90, binary, all_signed, mx_or_none)
+DEF_MVE_FUNCTION (vhcaddq_rot270, binary, all_signed, mx_or_none)
 DEF_MVE_FUNCTION (vhsubq, binary_opt_n, all_integer, mx_or_none)
 DEF_MVE_FUNCTION (vmaxaq, binary_maxamina, all_signed, m_or_none)
 DEF_MVE_FUNCTION (vmaxavq, binary_maxavminav, all_signed, p_or_none)
@@ -152,6 +156,8 @@  DEF_MVE_FUNCTION (vabsq, unary, all_float, mx_or_none)
 DEF_MVE_FUNCTION (vaddq, binary_opt_n, all_float, mx_or_none)
 DEF_MVE_FUNCTION (vandq, binary, all_float, mx_or_none)
 DEF_MVE_FUNCTION (vbrsrq, binary_imm32, all_float, mx_or_none)
+DEF_MVE_FUNCTION (vcaddq_rot90, binary, all_float, mx_or_none)
+DEF_MVE_FUNCTION (vcaddq_rot270, binary, all_float, mx_or_none)
 DEF_MVE_FUNCTION (vcmpeqq, cmp, all_float, m_or_none)
 DEF_MVE_FUNCTION (vcmpgeq, cmp, all_float, m_or_none)
 DEF_MVE_FUNCTION (vcmpgtq, cmp, all_float, m_or_none)
diff --git a/gcc/config/arm/arm-mve-builtins-base.h b/gcc/config/arm/arm-mve-builtins-base.h
index 942c8587446..8dac72970e0 100644
--- a/gcc/config/arm/arm-mve-builtins-base.h
+++ b/gcc/config/arm/arm-mve-builtins-base.h
@@ -33,6 +33,8 @@  extern const function_base *const vaddvaq;
 extern const function_base *const vaddvq;
 extern const function_base *const vandq;
 extern const function_base *const vbrsrq;
+extern const function_base *const vcaddq_rot90;
+extern const function_base *const vcaddq_rot270;
 extern const function_base *const vclsq;
 extern const function_base *const vclzq;
 extern const function_base *const vcmpcsq;
@@ -50,6 +52,8 @@  extern const function_base *const vfmaq;
 extern const function_base *const vfmasq;
 extern const function_base *const vfmsq;
 extern const function_base *const vhaddq;
+extern const function_base *const vhcaddq_rot90;
+extern const function_base *const vhcaddq_rot270;
 extern const function_base *const vhsubq;
 extern const function_base *const vmaxaq;
 extern const function_base *const vmaxavq;
diff --git a/gcc/config/arm/arm-mve-builtins-functions.h b/gcc/config/arm/arm-mve-builtins-functions.h
index 8f3bae4b7da..a6573844319 100644
--- a/gcc/config/arm/arm-mve-builtins-functions.h
+++ b/gcc/config/arm/arm-mve-builtins-functions.h
@@ -735,6 +735,109 @@  public:
   }
 };
 
+/* Map the function directly to CODE (UNSPEC, UNSPEC, UNSPEC, M) where
+   M is the vector mode associated with type suffix 0.  USed for the
+   operations where there is a "rot90" or "rot270" suffix, depending
+   on the UNSPEC.  */
+class unspec_mve_function_exact_insn_rot : public function_base
+{
+public:
+  CONSTEXPR unspec_mve_function_exact_insn_rot (int unspec_for_sint,
+						int unspec_for_uint,
+						int unspec_for_fp,
+						int unspec_for_m_sint,
+						int unspec_for_m_uint,
+						int unspec_for_m_fp)
+    : m_unspec_for_sint (unspec_for_sint),
+      m_unspec_for_uint (unspec_for_uint),
+      m_unspec_for_fp (unspec_for_fp),
+      m_unspec_for_m_sint (unspec_for_m_sint),
+      m_unspec_for_m_uint (unspec_for_m_uint),
+      m_unspec_for_m_fp (unspec_for_m_fp)
+  {}
+
+  /* The unspec code associated with signed-integer, unsigned-integer
+     and floating-point operations respectively.  It covers the cases
+     with and without the _m predicate.  */
+  int m_unspec_for_sint;
+  int m_unspec_for_uint;
+  int m_unspec_for_fp;
+  int m_unspec_for_m_sint;
+  int m_unspec_for_m_uint;
+  int m_unspec_for_m_fp;
+
+  rtx
+  expand (function_expander &e) const override
+  {
+    insn_code code;
+
+    switch (e.pred)
+      {
+      case PRED_none:
+	switch (e.mode_suffix_id)
+	  {
+	  case MODE_none:
+	    /* No predicate, no suffix.  */
+	    if (e.type_suffix (0).integer_p)
+	      if (e.type_suffix (0).unsigned_p)
+		code = code_for_mve_q (m_unspec_for_uint, m_unspec_for_uint, m_unspec_for_uint, e.vector_mode (0));
+	      else
+		code = code_for_mve_q (m_unspec_for_sint, m_unspec_for_sint, m_unspec_for_sint, e.vector_mode (0));
+	    else
+	      code = code_for_mve_q_f (m_unspec_for_fp, m_unspec_for_fp, e.vector_mode (0));
+	    break;
+
+	  default:
+	    gcc_unreachable ();
+	  }
+	return e.use_exact_insn (code);
+
+      case PRED_m:
+	switch (e.mode_suffix_id)
+	  {
+	  case MODE_none:
+	    /* No suffix, "m" predicate.  */
+	    if (e.type_suffix (0).integer_p)
+	      if (e.type_suffix (0).unsigned_p)
+		code = code_for_mve_q_m (m_unspec_for_m_uint, m_unspec_for_m_uint, m_unspec_for_m_uint, e.vector_mode (0));
+	      else
+		code = code_for_mve_q_m (m_unspec_for_m_sint, m_unspec_for_m_sint, m_unspec_for_m_sint, e.vector_mode (0));
+	    else
+	      code = code_for_mve_q_m_f (m_unspec_for_m_fp, m_unspec_for_m_fp, e.vector_mode (0));
+	    break;
+
+	  default:
+	    gcc_unreachable ();
+	  }
+	return e.use_cond_insn (code, 0);
+
+      case PRED_x:
+	switch (e.mode_suffix_id)
+	  {
+	  case MODE_none:
+	    /* No suffix, "x" predicate.  */
+	    if (e.type_suffix (0).integer_p)
+	      if (e.type_suffix (0).unsigned_p)
+		code = code_for_mve_q_m (m_unspec_for_m_uint, m_unspec_for_m_uint, m_unspec_for_m_uint, e.vector_mode (0));
+	      else
+		code = code_for_mve_q_m (m_unspec_for_m_sint, m_unspec_for_m_sint, m_unspec_for_m_sint, e.vector_mode (0));
+	    else
+	      code = code_for_mve_q_m_f (m_unspec_for_m_fp, m_unspec_for_m_fp, e.vector_mode (0));
+	    break;
+
+	  default:
+	    gcc_unreachable ();
+	  }
+	return e.use_pred_x_insn (code);
+
+      default:
+	gcc_unreachable ();
+      }
+
+    gcc_unreachable ();
+  }
+};
+
 } /* end namespace arm_mve */
 
 /* Declare the global function base NAME, creating it from an instance
diff --git a/gcc/config/arm/arm_mve.h b/gcc/config/arm/arm_mve.h
index 30485e74cdd..9297ad757ab 100644
--- a/gcc/config/arm/arm_mve.h
+++ b/gcc/config/arm/arm_mve.h
@@ -45,20 +45,12 @@ 
 #define vornq(__a, __b) __arm_vornq(__a, __b)
 #define vmulltq_int(__a, __b) __arm_vmulltq_int(__a, __b)
 #define vmullbq_int(__a, __b) __arm_vmullbq_int(__a, __b)
-#define vcaddq_rot90(__a, __b) __arm_vcaddq_rot90(__a, __b)
-#define vcaddq_rot270(__a, __b) __arm_vcaddq_rot270(__a, __b)
 #define vbicq(__a, __b) __arm_vbicq(__a, __b)
-#define vhcaddq_rot90(__a, __b) __arm_vhcaddq_rot90(__a, __b)
-#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 vbicq_m_n(__a, __imm, __p) __arm_vbicq_m_n(__a, __imm, __p)
 #define vshlcq(__a, __b, __imm) __arm_vshlcq(__a, __b, __imm)
 #define vbicq_m(__inactive, __a, __b, __p) __arm_vbicq_m(__inactive, __a, __b, __p)
-#define vcaddq_rot270_m(__inactive, __a, __b, __p) __arm_vcaddq_rot270_m(__inactive, __a, __b, __p)
-#define vcaddq_rot90_m(__inactive, __a, __b, __p) __arm_vcaddq_rot90_m(__inactive, __a, __b, __p)
-#define vhcaddq_rot270_m(__inactive, __a, __b, __p) __arm_vhcaddq_rot270_m(__inactive, __a, __b, __p)
-#define vhcaddq_rot90_m(__inactive, __a, __b, __p) __arm_vhcaddq_rot90_m(__inactive, __a, __b, __p)
 #define vmullbq_int_m(__inactive, __a, __b, __p) __arm_vmullbq_int_m(__inactive, __a, __b, __p)
 #define vmulltq_int_m(__inactive, __a, __b, __p) __arm_vmulltq_int_m(__inactive, __a, __b, __p)
 #define vornq_m(__inactive, __a, __b, __p) __arm_vornq_m(__inactive, __a, __b, __p)
@@ -141,10 +133,6 @@ 
 #define vmullbq_int_x(__a, __b, __p) __arm_vmullbq_int_x(__a, __b, __p)
 #define vmulltq_poly_x(__a, __b, __p) __arm_vmulltq_poly_x(__a, __b, __p)
 #define vmulltq_int_x(__a, __b, __p) __arm_vmulltq_int_x(__a, __b, __p)
-#define vcaddq_rot90_x(__a, __b, __p) __arm_vcaddq_rot90_x(__a, __b, __p)
-#define vcaddq_rot270_x(__a, __b, __p) __arm_vcaddq_rot270_x(__a, __b, __p)
-#define vhcaddq_rot90_x(__a, __b, __p) __arm_vhcaddq_rot90_x(__a, __b, __p)
-#define vhcaddq_rot270_x(__a, __b, __p) __arm_vhcaddq_rot270_x(__a, __b, __p)
 #define vbicq_x(__a, __b, __p) __arm_vbicq_x(__a, __b, __p)
 #define vornq_x(__a, __b, __p) __arm_vornq_x(__a, __b, __p)
 #define vadciq(__a, __b, __carry_out) __arm_vadciq(__a, __b, __carry_out)
@@ -249,44 +237,26 @@ 
 #define vornq_u8(__a, __b) __arm_vornq_u8(__a, __b)
 #define vmulltq_int_u8(__a, __b) __arm_vmulltq_int_u8(__a, __b)
 #define vmullbq_int_u8(__a, __b) __arm_vmullbq_int_u8(__a, __b)
-#define vcaddq_rot90_u8(__a, __b) __arm_vcaddq_rot90_u8(__a, __b)
-#define vcaddq_rot270_u8(__a, __b) __arm_vcaddq_rot270_u8(__a, __b)
 #define vbicq_u8(__a, __b) __arm_vbicq_u8(__a, __b)
 #define vornq_s8(__a, __b) __arm_vornq_s8(__a, __b)
 #define vmulltq_int_s8(__a, __b) __arm_vmulltq_int_s8(__a, __b)
 #define vmullbq_int_s8(__a, __b) __arm_vmullbq_int_s8(__a, __b)
-#define vhcaddq_rot90_s8(__a, __b) __arm_vhcaddq_rot90_s8(__a, __b)
-#define vhcaddq_rot270_s8(__a, __b) __arm_vhcaddq_rot270_s8(__a, __b)
-#define vcaddq_rot90_s8(__a, __b) __arm_vcaddq_rot90_s8(__a, __b)
-#define vcaddq_rot270_s8(__a, __b) __arm_vcaddq_rot270_s8(__a, __b)
 #define vbicq_s8(__a, __b) __arm_vbicq_s8(__a, __b)
 #define vornq_u16(__a, __b) __arm_vornq_u16(__a, __b)
 #define vmulltq_int_u16(__a, __b) __arm_vmulltq_int_u16(__a, __b)
 #define vmullbq_int_u16(__a, __b) __arm_vmullbq_int_u16(__a, __b)
-#define vcaddq_rot90_u16(__a, __b) __arm_vcaddq_rot90_u16(__a, __b)
-#define vcaddq_rot270_u16(__a, __b) __arm_vcaddq_rot270_u16(__a, __b)
 #define vbicq_u16(__a, __b) __arm_vbicq_u16(__a, __b)
 #define vornq_s16(__a, __b) __arm_vornq_s16(__a, __b)
 #define vmulltq_int_s16(__a, __b) __arm_vmulltq_int_s16(__a, __b)
 #define vmullbq_int_s16(__a, __b) __arm_vmullbq_int_s16(__a, __b)
-#define vhcaddq_rot90_s16(__a, __b) __arm_vhcaddq_rot90_s16(__a, __b)
-#define vhcaddq_rot270_s16(__a, __b) __arm_vhcaddq_rot270_s16(__a, __b)
-#define vcaddq_rot90_s16(__a, __b) __arm_vcaddq_rot90_s16(__a, __b)
-#define vcaddq_rot270_s16(__a, __b) __arm_vcaddq_rot270_s16(__a, __b)
 #define vbicq_s16(__a, __b) __arm_vbicq_s16(__a, __b)
 #define vornq_u32(__a, __b) __arm_vornq_u32(__a, __b)
 #define vmulltq_int_u32(__a, __b) __arm_vmulltq_int_u32(__a, __b)
 #define vmullbq_int_u32(__a, __b) __arm_vmullbq_int_u32(__a, __b)
-#define vcaddq_rot90_u32(__a, __b) __arm_vcaddq_rot90_u32(__a, __b)
-#define vcaddq_rot270_u32(__a, __b) __arm_vcaddq_rot270_u32(__a, __b)
 #define vbicq_u32(__a, __b) __arm_vbicq_u32(__a, __b)
 #define vornq_s32(__a, __b) __arm_vornq_s32(__a, __b)
 #define vmulltq_int_s32(__a, __b) __arm_vmulltq_int_s32(__a, __b)
 #define vmullbq_int_s32(__a, __b) __arm_vmullbq_int_s32(__a, __b)
-#define vhcaddq_rot90_s32(__a, __b) __arm_vhcaddq_rot90_s32(__a, __b)
-#define vhcaddq_rot270_s32(__a, __b) __arm_vhcaddq_rot270_s32(__a, __b)
-#define vcaddq_rot90_s32(__a, __b) __arm_vcaddq_rot90_s32(__a, __b)
-#define vcaddq_rot270_s32(__a, __b) __arm_vcaddq_rot270_s32(__a, __b)
 #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)
@@ -296,8 +266,6 @@ 
 #define vcmulq_rot270_f16(__a, __b) __arm_vcmulq_rot270_f16(__a, __b)
 #define vcmulq_rot180_f16(__a, __b) __arm_vcmulq_rot180_f16(__a, __b)
 #define vcmulq_f16(__a, __b) __arm_vcmulq_f16(__a, __b)
-#define vcaddq_rot90_f16(__a, __b) __arm_vcaddq_rot90_f16(__a, __b)
-#define vcaddq_rot270_f16(__a, __b) __arm_vcaddq_rot270_f16(__a, __b)
 #define vbicq_f16(__a, __b) __arm_vbicq_f16(__a, __b)
 #define vbicq_n_s16(__a,  __imm) __arm_vbicq_n_s16(__a,  __imm)
 #define vmulltq_poly_p16(__a, __b) __arm_vmulltq_poly_p16(__a, __b)
@@ -308,8 +276,6 @@ 
 #define vcmulq_rot270_f32(__a, __b) __arm_vcmulq_rot270_f32(__a, __b)
 #define vcmulq_rot180_f32(__a, __b) __arm_vcmulq_rot180_f32(__a, __b)
 #define vcmulq_f32(__a, __b) __arm_vcmulq_f32(__a, __b)
-#define vcaddq_rot90_f32(__a, __b) __arm_vcaddq_rot90_f32(__a, __b)
-#define vcaddq_rot270_f32(__a, __b) __arm_vcaddq_rot270_f32(__a, __b)
 #define vbicq_f32(__a, __b) __arm_vbicq_f32(__a, __b)
 #define vbicq_n_s32(__a,  __imm) __arm_vbicq_n_s32(__a,  __imm)
 #define vctp8q_m(__a, __p) __arm_vctp8q_m(__a, __p)
@@ -374,24 +340,6 @@ 
 #define vbicq_m_u8(__inactive, __a, __b, __p) __arm_vbicq_m_u8(__inactive, __a, __b, __p)
 #define vbicq_m_u32(__inactive, __a, __b, __p) __arm_vbicq_m_u32(__inactive, __a, __b, __p)
 #define vbicq_m_u16(__inactive, __a, __b, __p) __arm_vbicq_m_u16(__inactive, __a, __b, __p)
-#define vcaddq_rot270_m_s8(__inactive, __a, __b, __p) __arm_vcaddq_rot270_m_s8(__inactive, __a, __b, __p)
-#define vcaddq_rot270_m_s32(__inactive, __a, __b, __p) __arm_vcaddq_rot270_m_s32(__inactive, __a, __b, __p)
-#define vcaddq_rot270_m_s16(__inactive, __a, __b, __p) __arm_vcaddq_rot270_m_s16(__inactive, __a, __b, __p)
-#define vcaddq_rot270_m_u8(__inactive, __a, __b, __p) __arm_vcaddq_rot270_m_u8(__inactive, __a, __b, __p)
-#define vcaddq_rot270_m_u32(__inactive, __a, __b, __p) __arm_vcaddq_rot270_m_u32(__inactive, __a, __b, __p)
-#define vcaddq_rot270_m_u16(__inactive, __a, __b, __p) __arm_vcaddq_rot270_m_u16(__inactive, __a, __b, __p)
-#define vcaddq_rot90_m_s8(__inactive, __a, __b, __p) __arm_vcaddq_rot90_m_s8(__inactive, __a, __b, __p)
-#define vcaddq_rot90_m_s32(__inactive, __a, __b, __p) __arm_vcaddq_rot90_m_s32(__inactive, __a, __b, __p)
-#define vcaddq_rot90_m_s16(__inactive, __a, __b, __p) __arm_vcaddq_rot90_m_s16(__inactive, __a, __b, __p)
-#define vcaddq_rot90_m_u8(__inactive, __a, __b, __p) __arm_vcaddq_rot90_m_u8(__inactive, __a, __b, __p)
-#define vcaddq_rot90_m_u32(__inactive, __a, __b, __p) __arm_vcaddq_rot90_m_u32(__inactive, __a, __b, __p)
-#define vcaddq_rot90_m_u16(__inactive, __a, __b, __p) __arm_vcaddq_rot90_m_u16(__inactive, __a, __b, __p)
-#define vhcaddq_rot270_m_s8(__inactive, __a, __b, __p) __arm_vhcaddq_rot270_m_s8(__inactive, __a, __b, __p)
-#define vhcaddq_rot270_m_s32(__inactive, __a, __b, __p) __arm_vhcaddq_rot270_m_s32(__inactive, __a, __b, __p)
-#define vhcaddq_rot270_m_s16(__inactive, __a, __b, __p) __arm_vhcaddq_rot270_m_s16(__inactive, __a, __b, __p)
-#define vhcaddq_rot90_m_s8(__inactive, __a, __b, __p) __arm_vhcaddq_rot90_m_s8(__inactive, __a, __b, __p)
-#define vhcaddq_rot90_m_s32(__inactive, __a, __b, __p) __arm_vhcaddq_rot90_m_s32(__inactive, __a, __b, __p)
-#define vhcaddq_rot90_m_s16(__inactive, __a, __b, __p) __arm_vhcaddq_rot90_m_s16(__inactive, __a, __b, __p)
 #define vmullbq_int_m_s8(__inactive, __a, __b, __p) __arm_vmullbq_int_m_s8(__inactive, __a, __b, __p)
 #define vmullbq_int_m_s32(__inactive, __a, __b, __p) __arm_vmullbq_int_m_s32(__inactive, __a, __b, __p)
 #define vmullbq_int_m_s16(__inactive, __a, __b, __p) __arm_vmullbq_int_m_s16(__inactive, __a, __b, __p)
@@ -416,10 +364,6 @@ 
 #define vmulltq_poly_m_p16(__inactive, __a, __b, __p) __arm_vmulltq_poly_m_p16(__inactive, __a, __b, __p)
 #define vbicq_m_f32(__inactive, __a, __b, __p) __arm_vbicq_m_f32(__inactive, __a, __b, __p)
 #define vbicq_m_f16(__inactive, __a, __b, __p) __arm_vbicq_m_f16(__inactive, __a, __b, __p)
-#define vcaddq_rot270_m_f32(__inactive, __a, __b, __p) __arm_vcaddq_rot270_m_f32(__inactive, __a, __b, __p)
-#define vcaddq_rot270_m_f16(__inactive, __a, __b, __p) __arm_vcaddq_rot270_m_f16(__inactive, __a, __b, __p)
-#define vcaddq_rot90_m_f32(__inactive, __a, __b, __p) __arm_vcaddq_rot90_m_f32(__inactive, __a, __b, __p)
-#define vcaddq_rot90_m_f16(__inactive, __a, __b, __p) __arm_vcaddq_rot90_m_f16(__inactive, __a, __b, __p)
 #define vcmlaq_m_f32(__a, __b, __c, __p) __arm_vcmlaq_m_f32(__a, __b, __c, __p)
 #define vcmlaq_m_f16(__a, __b, __c, __p) __arm_vcmlaq_m_f16(__a, __b, __c, __p)
 #define vcmlaq_rot180_m_f32(__a, __b, __c, __p) __arm_vcmlaq_rot180_m_f32(__a, __b, __c, __p)
@@ -756,24 +700,6 @@ 
 #define vmulltq_int_x_u8(__a, __b, __p) __arm_vmulltq_int_x_u8(__a, __b, __p)
 #define vmulltq_int_x_u16(__a, __b, __p) __arm_vmulltq_int_x_u16(__a, __b, __p)
 #define vmulltq_int_x_u32(__a, __b, __p) __arm_vmulltq_int_x_u32(__a, __b, __p)
-#define vcaddq_rot90_x_s8(__a, __b, __p) __arm_vcaddq_rot90_x_s8(__a, __b, __p)
-#define vcaddq_rot90_x_s16(__a, __b, __p) __arm_vcaddq_rot90_x_s16(__a, __b, __p)
-#define vcaddq_rot90_x_s32(__a, __b, __p) __arm_vcaddq_rot90_x_s32(__a, __b, __p)
-#define vcaddq_rot90_x_u8(__a, __b, __p) __arm_vcaddq_rot90_x_u8(__a, __b, __p)
-#define vcaddq_rot90_x_u16(__a, __b, __p) __arm_vcaddq_rot90_x_u16(__a, __b, __p)
-#define vcaddq_rot90_x_u32(__a, __b, __p) __arm_vcaddq_rot90_x_u32(__a, __b, __p)
-#define vcaddq_rot270_x_s8(__a, __b, __p) __arm_vcaddq_rot270_x_s8(__a, __b, __p)
-#define vcaddq_rot270_x_s16(__a, __b, __p) __arm_vcaddq_rot270_x_s16(__a, __b, __p)
-#define vcaddq_rot270_x_s32(__a, __b, __p) __arm_vcaddq_rot270_x_s32(__a, __b, __p)
-#define vcaddq_rot270_x_u8(__a, __b, __p) __arm_vcaddq_rot270_x_u8(__a, __b, __p)
-#define vcaddq_rot270_x_u16(__a, __b, __p) __arm_vcaddq_rot270_x_u16(__a, __b, __p)
-#define vcaddq_rot270_x_u32(__a, __b, __p) __arm_vcaddq_rot270_x_u32(__a, __b, __p)
-#define vhcaddq_rot90_x_s8(__a, __b, __p) __arm_vhcaddq_rot90_x_s8(__a, __b, __p)
-#define vhcaddq_rot90_x_s16(__a, __b, __p) __arm_vhcaddq_rot90_x_s16(__a, __b, __p)
-#define vhcaddq_rot90_x_s32(__a, __b, __p) __arm_vhcaddq_rot90_x_s32(__a, __b, __p)
-#define vhcaddq_rot270_x_s8(__a, __b, __p) __arm_vhcaddq_rot270_x_s8(__a, __b, __p)
-#define vhcaddq_rot270_x_s16(__a, __b, __p) __arm_vhcaddq_rot270_x_s16(__a, __b, __p)
-#define vhcaddq_rot270_x_s32(__a, __b, __p) __arm_vhcaddq_rot270_x_s32(__a, __b, __p)
 #define vbicq_x_s8(__a, __b, __p) __arm_vbicq_x_s8(__a, __b, __p)
 #define vbicq_x_s16(__a, __b, __p) __arm_vbicq_x_s16(__a, __b, __p)
 #define vbicq_x_s32(__a, __b, __p) __arm_vbicq_x_s32(__a, __b, __p)
@@ -786,10 +712,6 @@ 
 #define vornq_x_u8(__a, __b, __p) __arm_vornq_x_u8(__a, __b, __p)
 #define vornq_x_u16(__a, __b, __p) __arm_vornq_x_u16(__a, __b, __p)
 #define vornq_x_u32(__a, __b, __p) __arm_vornq_x_u32(__a, __b, __p)
-#define vcaddq_rot90_x_f16(__a, __b, __p) __arm_vcaddq_rot90_x_f16(__a, __b, __p)
-#define vcaddq_rot90_x_f32(__a, __b, __p) __arm_vcaddq_rot90_x_f32(__a, __b, __p)
-#define vcaddq_rot270_x_f16(__a, __b, __p) __arm_vcaddq_rot270_x_f16(__a, __b, __p)
-#define vcaddq_rot270_x_f32(__a, __b, __p) __arm_vcaddq_rot270_x_f32(__a, __b, __p)
 #define vcmulq_x_f16(__a, __b, __p) __arm_vcmulq_x_f16(__a, __b, __p)
 #define vcmulq_x_f32(__a, __b, __p) __arm_vcmulq_x_f32(__a, __b, __p)
 #define vcmulq_rot90_x_f16(__a, __b, __p) __arm_vcmulq_rot90_x_f16(__a, __b, __p)
@@ -1058,22 +980,6 @@  __arm_vmullbq_int_u8 (uint8x16_t __a, uint8x16_t __b)
   return __builtin_mve_vmullbq_int_uv16qi (__a, __b);
 }
 
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_u8 (uint8x16_t __a, uint8x16_t __b)
-{
-  return (uint8x16_t)
-    __builtin_mve_vcaddq_rot90v16qi ((int8x16_t)__a, (int8x16_t)__b);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_u8 (uint8x16_t __a, uint8x16_t __b)
-{
-  return (uint8x16_t)
-    __builtin_mve_vcaddq_rot270v16qi ((int8x16_t)__a, (int8x16_t)__b);
-}
-
 __extension__ extern __inline uint8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq_u8 (uint8x16_t __a, uint8x16_t __b)
@@ -1102,34 +1008,6 @@  __arm_vmullbq_int_s8 (int8x16_t __a, int8x16_t __b)
   return __builtin_mve_vmullbq_int_sv16qi (__a, __b);
 }
 
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_s8 (int8x16_t __a, int8x16_t __b)
-{
-  return __builtin_mve_vhcaddq_rot90_sv16qi (__a, __b);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_s8 (int8x16_t __a, int8x16_t __b)
-{
-  return __builtin_mve_vhcaddq_rot270_sv16qi (__a, __b);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_s8 (int8x16_t __a, int8x16_t __b)
-{
-  return __builtin_mve_vcaddq_rot90v16qi (__a, __b);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_s8 (int8x16_t __a, int8x16_t __b)
-{
-  return __builtin_mve_vcaddq_rot270v16qi (__a, __b);
-}
-
 __extension__ extern __inline int8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq_s8 (int8x16_t __a, int8x16_t __b)
@@ -1158,22 +1036,6 @@  __arm_vmullbq_int_u16 (uint16x8_t __a, uint16x8_t __b)
   return __builtin_mve_vmullbq_int_uv8hi (__a, __b);
 }
 
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_u16 (uint16x8_t __a, uint16x8_t __b)
-{
-  return (uint16x8_t)
-    __builtin_mve_vcaddq_rot90v8hi ((int16x8_t)__a, (int16x8_t)__b);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_u16 (uint16x8_t __a, uint16x8_t __b)
-{
-  return (uint16x8_t)
-    __builtin_mve_vcaddq_rot270v8hi ((int16x8_t)__a, (int16x8_t)__b);
-}
-
 __extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq_u16 (uint16x8_t __a, uint16x8_t __b)
@@ -1202,34 +1064,6 @@  __arm_vmullbq_int_s16 (int16x8_t __a, int16x8_t __b)
   return __builtin_mve_vmullbq_int_sv8hi (__a, __b);
 }
 
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_s16 (int16x8_t __a, int16x8_t __b)
-{
-  return __builtin_mve_vhcaddq_rot90_sv8hi (__a, __b);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_s16 (int16x8_t __a, int16x8_t __b)
-{
-  return __builtin_mve_vhcaddq_rot270_sv8hi (__a, __b);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_s16 (int16x8_t __a, int16x8_t __b)
-{
-  return __builtin_mve_vcaddq_rot90v8hi (__a, __b);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_s16 (int16x8_t __a, int16x8_t __b)
-{
-  return __builtin_mve_vcaddq_rot270v8hi (__a, __b);
-}
-
 __extension__ extern __inline int16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq_s16 (int16x8_t __a, int16x8_t __b)
@@ -1258,22 +1092,6 @@  __arm_vmullbq_int_u32 (uint32x4_t __a, uint32x4_t __b)
   return __builtin_mve_vmullbq_int_uv4si (__a, __b);
 }
 
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_u32 (uint32x4_t __a, uint32x4_t __b)
-{
-  return (uint32x4_t)
-    __builtin_mve_vcaddq_rot90v4si ((int32x4_t)__a, (int32x4_t)__b);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_u32 (uint32x4_t __a, uint32x4_t __b)
-{
-  return (uint32x4_t)
-    __builtin_mve_vcaddq_rot270v4si ((int32x4_t)__a, (int32x4_t)__b);
-}
-
 __extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq_u32 (uint32x4_t __a, uint32x4_t __b)
@@ -1302,34 +1120,6 @@  __arm_vmullbq_int_s32 (int32x4_t __a, int32x4_t __b)
   return __builtin_mve_vmullbq_int_sv4si (__a, __b);
 }
 
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_s32 (int32x4_t __a, int32x4_t __b)
-{
-  return __builtin_mve_vhcaddq_rot90_sv4si (__a, __b);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_s32 (int32x4_t __a, int32x4_t __b)
-{
-  return __builtin_mve_vhcaddq_rot270_sv4si (__a, __b);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_s32 (int32x4_t __a, int32x4_t __b)
-{
-  return __builtin_mve_vcaddq_rot90v4si (__a, __b);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_s32 (int32x4_t __a, int32x4_t __b)
-{
-  return __builtin_mve_vcaddq_rot270v4si (__a, __b);
-}
-
 __extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq_s32 (int32x4_t __a, int32x4_t __b)
@@ -1545,132 +1335,6 @@  __arm_vbicq_m_u16 (uint16x8_t __inactive, uint16x8_t __a, uint16x8_t __b, mve_pr
   return __builtin_mve_vbicq_m_uv8hi (__inactive, __a, __b, __p);
 }
 
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m_s8 (int8x16_t __inactive, int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot270_m_sv16qi (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m_s32 (int32x4_t __inactive, int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot270_m_sv4si (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m_s16 (int16x8_t __inactive, int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot270_m_sv8hi (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m_u8 (uint8x16_t __inactive, uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot270_m_uv16qi (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m_u32 (uint32x4_t __inactive, uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot270_m_uv4si (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m_u16 (uint16x8_t __inactive, uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot270_m_uv8hi (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m_s8 (int8x16_t __inactive, int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot90_m_sv16qi (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m_s32 (int32x4_t __inactive, int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot90_m_sv4si (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m_s16 (int16x8_t __inactive, int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot90_m_sv8hi (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m_u8 (uint8x16_t __inactive, uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot90_m_uv16qi (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m_u32 (uint32x4_t __inactive, uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot90_m_uv4si (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m_u16 (uint16x8_t __inactive, uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot90_m_uv8hi (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_m_s8 (int8x16_t __inactive, int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vhcaddq_rot270_m_sv16qi (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_m_s32 (int32x4_t __inactive, int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vhcaddq_rot270_m_sv4si (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_m_s16 (int16x8_t __inactive, int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vhcaddq_rot270_m_sv8hi (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_m_s8 (int8x16_t __inactive, int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vhcaddq_rot90_m_sv16qi (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_m_s32 (int32x4_t __inactive, int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vhcaddq_rot90_m_sv4si (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_m_s16 (int16x8_t __inactive, int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vhcaddq_rot90_m_sv8hi (__inactive, __a, __b, __p);
-}
-
 __extension__ extern __inline int16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vmullbq_int_m_s8 (int16x8_t __inactive, int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
@@ -3850,281 +3514,155 @@  __arm_vmulltq_int_x_u32 (uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
 
 __extension__ extern __inline int8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x_s8 (int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
+__arm_vbicq_x_s8 (int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
 {
-  return __builtin_mve_vcaddq_rot90_m_sv16qi (__arm_vuninitializedq_s8 (), __a, __b, __p);
+  return __builtin_mve_vbicq_m_sv16qi (__arm_vuninitializedq_s8 (), __a, __b, __p);
 }
 
 __extension__ extern __inline int16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x_s16 (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
+__arm_vbicq_x_s16 (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
 {
-  return __builtin_mve_vcaddq_rot90_m_sv8hi (__arm_vuninitializedq_s16 (), __a, __b, __p);
+  return __builtin_mve_vbicq_m_sv8hi (__arm_vuninitializedq_s16 (), __a, __b, __p);
 }
 
 __extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x_s32 (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
+__arm_vbicq_x_s32 (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
 {
-  return __builtin_mve_vcaddq_rot90_m_sv4si (__arm_vuninitializedq_s32 (), __a, __b, __p);
+  return __builtin_mve_vbicq_m_sv4si (__arm_vuninitializedq_s32 (), __a, __b, __p);
 }
 
 __extension__ extern __inline uint8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x_u8 (uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
+__arm_vbicq_x_u8 (uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
 {
-  return __builtin_mve_vcaddq_rot90_m_uv16qi (__arm_vuninitializedq_u8 (), __a, __b, __p);
+  return __builtin_mve_vbicq_m_uv16qi (__arm_vuninitializedq_u8 (), __a, __b, __p);
 }
 
 __extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x_u16 (uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
+__arm_vbicq_x_u16 (uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
 {
-  return __builtin_mve_vcaddq_rot90_m_uv8hi (__arm_vuninitializedq_u16 (), __a, __b, __p);
+  return __builtin_mve_vbicq_m_uv8hi (__arm_vuninitializedq_u16 (), __a, __b, __p);
 }
 
 __extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x_u32 (uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
+__arm_vbicq_x_u32 (uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
 {
-  return __builtin_mve_vcaddq_rot90_m_uv4si (__arm_vuninitializedq_u32 (), __a, __b, __p);
+  return __builtin_mve_vbicq_m_uv4si (__arm_vuninitializedq_u32 (), __a, __b, __p);
 }
 
 __extension__ extern __inline int8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x_s8 (int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
+__arm_vornq_x_s8 (int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
 {
-  return __builtin_mve_vcaddq_rot270_m_sv16qi (__arm_vuninitializedq_s8 (), __a, __b, __p);
+  return __builtin_mve_vornq_m_sv16qi (__arm_vuninitializedq_s8 (), __a, __b, __p);
 }
 
 __extension__ extern __inline int16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x_s16 (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
+__arm_vornq_x_s16 (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
 {
-  return __builtin_mve_vcaddq_rot270_m_sv8hi (__arm_vuninitializedq_s16 (), __a, __b, __p);
+  return __builtin_mve_vornq_m_sv8hi (__arm_vuninitializedq_s16 (), __a, __b, __p);
 }
 
 __extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x_s32 (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
+__arm_vornq_x_s32 (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
 {
-  return __builtin_mve_vcaddq_rot270_m_sv4si (__arm_vuninitializedq_s32 (), __a, __b, __p);
+  return __builtin_mve_vornq_m_sv4si (__arm_vuninitializedq_s32 (), __a, __b, __p);
 }
 
 __extension__ extern __inline uint8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x_u8 (uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
+__arm_vornq_x_u8 (uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
 {
-  return __builtin_mve_vcaddq_rot270_m_uv16qi (__arm_vuninitializedq_u8 (), __a, __b, __p);
+  return __builtin_mve_vornq_m_uv16qi (__arm_vuninitializedq_u8 (), __a, __b, __p);
 }
 
 __extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x_u16 (uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
+__arm_vornq_x_u16 (uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
 {
-  return __builtin_mve_vcaddq_rot270_m_uv8hi (__arm_vuninitializedq_u16 (), __a, __b, __p);
+  return __builtin_mve_vornq_m_uv8hi (__arm_vuninitializedq_u16 (), __a, __b, __p);
 }
 
 __extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x_u32 (uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
+__arm_vornq_x_u32 (uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
 {
-  return __builtin_mve_vcaddq_rot270_m_uv4si (__arm_vuninitializedq_u32 (), __a, __b, __p);
+  return __builtin_mve_vornq_m_uv4si (__arm_vuninitializedq_u32 (), __a, __b, __p);
 }
 
-__extension__ extern __inline int8x16_t
+__extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_x_s8 (int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
+__arm_vadciq_s32 (int32x4_t __a, int32x4_t __b, unsigned * __carry_out)
 {
-  return __builtin_mve_vhcaddq_rot90_m_sv16qi (__arm_vuninitializedq_s8 (), __a, __b, __p);
+  int32x4_t __res = __builtin_mve_vadciq_sv4si (__a, __b);
+  *__carry_out = (__builtin_arm_get_fpscr_nzcvqc () >> 29) & 0x1u;
+  return __res;
 }
 
-__extension__ extern __inline int16x8_t
+__extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_x_s16 (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
+__arm_vadciq_u32 (uint32x4_t __a, uint32x4_t __b, unsigned * __carry_out)
 {
-  return __builtin_mve_vhcaddq_rot90_m_sv8hi (__arm_vuninitializedq_s16 (), __a, __b, __p);
+  uint32x4_t __res = __builtin_mve_vadciq_uv4si (__a, __b);
+  *__carry_out = (__builtin_arm_get_fpscr_nzcvqc () >> 29) & 0x1u;
+  return __res;
 }
 
 __extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_x_s32 (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vhcaddq_rot90_m_sv4si (__arm_vuninitializedq_s32 (), __a, __b, __p);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_x_s8 (int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
+__arm_vadciq_m_s32 (int32x4_t __inactive, int32x4_t __a, int32x4_t __b, unsigned * __carry_out, mve_pred16_t __p)
 {
-  return __builtin_mve_vhcaddq_rot270_m_sv16qi (__arm_vuninitializedq_s8 (), __a, __b, __p);
+  int32x4_t __res =  __builtin_mve_vadciq_m_sv4si (__inactive, __a, __b, __p);
+  *__carry_out = (__builtin_arm_get_fpscr_nzcvqc () >> 29) & 0x1u;
+  return __res;
 }
 
-__extension__ extern __inline int16x8_t
+__extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_x_s16 (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
+__arm_vadciq_m_u32 (uint32x4_t __inactive, uint32x4_t __a, uint32x4_t __b, unsigned * __carry_out, mve_pred16_t __p)
 {
-  return __builtin_mve_vhcaddq_rot270_m_sv8hi (__arm_vuninitializedq_s16 (), __a, __b, __p);
+  uint32x4_t __res = __builtin_mve_vadciq_m_uv4si (__inactive, __a, __b, __p);
+  *__carry_out = (__builtin_arm_get_fpscr_nzcvqc () >> 29) & 0x1u;
+  return __res;
 }
 
 __extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_x_s32 (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
+__arm_vadcq_s32 (int32x4_t __a, int32x4_t __b, unsigned * __carry)
 {
-  return __builtin_mve_vhcaddq_rot270_m_sv4si (__arm_vuninitializedq_s32 (), __a, __b, __p);
+  __builtin_arm_set_fpscr_nzcvqc((__builtin_arm_get_fpscr_nzcvqc () & ~0x20000000u) | ((*__carry & 0x1u) << 29));
+  int32x4_t __res = __builtin_mve_vadcq_sv4si (__a, __b);
+  *__carry = (__builtin_arm_get_fpscr_nzcvqc () >> 29) & 0x1u;
+  return __res;
 }
 
-__extension__ extern __inline int8x16_t
+__extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vbicq_x_s8 (int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
+__arm_vadcq_u32 (uint32x4_t __a, uint32x4_t __b, unsigned * __carry)
 {
-  return __builtin_mve_vbicq_m_sv16qi (__arm_vuninitializedq_s8 (), __a, __b, __p);
+  __builtin_arm_set_fpscr_nzcvqc((__builtin_arm_get_fpscr_nzcvqc () & ~0x20000000u) | ((*__carry & 0x1u) << 29));
+  uint32x4_t __res = __builtin_mve_vadcq_uv4si (__a, __b);
+  *__carry = (__builtin_arm_get_fpscr_nzcvqc () >> 29) & 0x1u;
+  return __res;
 }
 
-__extension__ extern __inline int16x8_t
+__extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vbicq_x_s16 (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
+__arm_vadcq_m_s32 (int32x4_t __inactive, int32x4_t __a, int32x4_t __b, unsigned * __carry, mve_pred16_t __p)
 {
-  return __builtin_mve_vbicq_m_sv8hi (__arm_vuninitializedq_s16 (), __a, __b, __p);
+  __builtin_arm_set_fpscr_nzcvqc((__builtin_arm_get_fpscr_nzcvqc () & ~0x20000000u) | ((*__carry & 0x1u) << 29));
+  int32x4_t __res = __builtin_mve_vadcq_m_sv4si (__inactive, __a, __b, __p);
+  *__carry = (__builtin_arm_get_fpscr_nzcvqc () >> 29) & 0x1u;
+  return __res;
 }
 
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vbicq_x_s32 (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vbicq_m_sv4si (__arm_vuninitializedq_s32 (), __a, __b, __p);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vbicq_x_u8 (uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vbicq_m_uv16qi (__arm_vuninitializedq_u8 (), __a, __b, __p);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vbicq_x_u16 (uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vbicq_m_uv8hi (__arm_vuninitializedq_u16 (), __a, __b, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vbicq_x_u32 (uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vbicq_m_uv4si (__arm_vuninitializedq_u32 (), __a, __b, __p);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vornq_x_s8 (int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vornq_m_sv16qi (__arm_vuninitializedq_s8 (), __a, __b, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vornq_x_s16 (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vornq_m_sv8hi (__arm_vuninitializedq_s16 (), __a, __b, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vornq_x_s32 (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vornq_m_sv4si (__arm_vuninitializedq_s32 (), __a, __b, __p);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vornq_x_u8 (uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vornq_m_uv16qi (__arm_vuninitializedq_u8 (), __a, __b, __p);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vornq_x_u16 (uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vornq_m_uv8hi (__arm_vuninitializedq_u16 (), __a, __b, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vornq_x_u32 (uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vornq_m_uv4si (__arm_vuninitializedq_u32 (), __a, __b, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vadciq_s32 (int32x4_t __a, int32x4_t __b, unsigned * __carry_out)
-{
-  int32x4_t __res = __builtin_mve_vadciq_sv4si (__a, __b);
-  *__carry_out = (__builtin_arm_get_fpscr_nzcvqc () >> 29) & 0x1u;
-  return __res;
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vadciq_u32 (uint32x4_t __a, uint32x4_t __b, unsigned * __carry_out)
-{
-  uint32x4_t __res = __builtin_mve_vadciq_uv4si (__a, __b);
-  *__carry_out = (__builtin_arm_get_fpscr_nzcvqc () >> 29) & 0x1u;
-  return __res;
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vadciq_m_s32 (int32x4_t __inactive, int32x4_t __a, int32x4_t __b, unsigned * __carry_out, mve_pred16_t __p)
-{
-  int32x4_t __res =  __builtin_mve_vadciq_m_sv4si (__inactive, __a, __b, __p);
-  *__carry_out = (__builtin_arm_get_fpscr_nzcvqc () >> 29) & 0x1u;
-  return __res;
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vadciq_m_u32 (uint32x4_t __inactive, uint32x4_t __a, uint32x4_t __b, unsigned * __carry_out, mve_pred16_t __p)
-{
-  uint32x4_t __res = __builtin_mve_vadciq_m_uv4si (__inactive, __a, __b, __p);
-  *__carry_out = (__builtin_arm_get_fpscr_nzcvqc () >> 29) & 0x1u;
-  return __res;
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vadcq_s32 (int32x4_t __a, int32x4_t __b, unsigned * __carry)
-{
-  __builtin_arm_set_fpscr_nzcvqc((__builtin_arm_get_fpscr_nzcvqc () & ~0x20000000u) | ((*__carry & 0x1u) << 29));
-  int32x4_t __res = __builtin_mve_vadcq_sv4si (__a, __b);
-  *__carry = (__builtin_arm_get_fpscr_nzcvqc () >> 29) & 0x1u;
-  return __res;
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vadcq_u32 (uint32x4_t __a, uint32x4_t __b, unsigned * __carry)
-{
-  __builtin_arm_set_fpscr_nzcvqc((__builtin_arm_get_fpscr_nzcvqc () & ~0x20000000u) | ((*__carry & 0x1u) << 29));
-  uint32x4_t __res = __builtin_mve_vadcq_uv4si (__a, __b);
-  *__carry = (__builtin_arm_get_fpscr_nzcvqc () >> 29) & 0x1u;
-  return __res;
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vadcq_m_s32 (int32x4_t __inactive, int32x4_t __a, int32x4_t __b, unsigned * __carry, mve_pred16_t __p)
-{
-  __builtin_arm_set_fpscr_nzcvqc((__builtin_arm_get_fpscr_nzcvqc () & ~0x20000000u) | ((*__carry & 0x1u) << 29));
-  int32x4_t __res = __builtin_mve_vadcq_m_sv4si (__inactive, __a, __b, __p);
-  *__carry = (__builtin_arm_get_fpscr_nzcvqc () >> 29) & 0x1u;
-  return __res;
-}
-
-__extension__ extern __inline uint32x4_t
+__extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vadcq_m_u32 (uint32x4_t __inactive, uint32x4_t __a, uint32x4_t __b, unsigned * __carry, mve_pred16_t __p)
 {
@@ -5051,20 +4589,6 @@  __arm_vcmulq_f16 (float16x8_t __a, float16x8_t __b)
   return __builtin_mve_vcmulqv8hf (__a, __b);
 }
 
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_f16 (float16x8_t __a, float16x8_t __b)
-{
-  return __builtin_mve_vcaddq_rot90v8hf (__a, __b);
-}
-
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_f16 (float16x8_t __a, float16x8_t __b)
-{
-  return __builtin_mve_vcaddq_rot270v8hf (__a, __b);
-}
-
 __extension__ extern __inline float16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq_f16 (float16x8_t __a, float16x8_t __b)
@@ -5107,20 +4631,6 @@  __arm_vcmulq_f32 (float32x4_t __a, float32x4_t __b)
   return __builtin_mve_vcmulqv4sf (__a, __b);
 }
 
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_f32 (float32x4_t __a, float32x4_t __b)
-{
-  return __builtin_mve_vcaddq_rot90v4sf (__a, __b);
-}
-
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_f32 (float32x4_t __a, float32x4_t __b)
-{
-  return __builtin_mve_vcaddq_rot270v4sf (__a, __b);
-}
-
 __extension__ extern __inline float32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq_f32 (float32x4_t __a, float32x4_t __b)
@@ -5437,34 +4947,6 @@  __arm_vbicq_m_f16 (float16x8_t __inactive, float16x8_t __a, float16x8_t __b, mve
   return __builtin_mve_vbicq_m_fv8hf (__inactive, __a, __b, __p);
 }
 
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m_f32 (float32x4_t __inactive, float32x4_t __a, float32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot270_m_fv4sf (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m_f16 (float16x8_t __inactive, float16x8_t __a, float16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot270_m_fv8hf (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m_f32 (float32x4_t __inactive, float32x4_t __a, float32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot90_m_fv4sf (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m_f16 (float16x8_t __inactive, float16x8_t __a, float16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot90_m_fv8hf (__inactive, __a, __b, __p);
-}
-
 __extension__ extern __inline float32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vcmlaq_m_f32 (float32x4_t __a, float32x4_t __b, float32x4_t __c, mve_pred16_t __p)
@@ -5877,34 +5359,6 @@  __arm_vstrwq_scatter_base_wb_p_f32 (uint32x4_t * __addr, const int __offset, flo
   *__addr = __builtin_mve_vstrwq_scatter_base_wb_p_fv4sf (*__addr, __offset, __value, __p);
 }
 
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x_f16 (float16x8_t __a, float16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot90_m_fv8hf (__arm_vuninitializedq_f16 (), __a, __b, __p);
-}
-
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x_f32 (float32x4_t __a, float32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot90_m_fv4sf (__arm_vuninitializedq_f32 (), __a, __b, __p);
-}
-
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x_f16 (float16x8_t __a, float16x8_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot270_m_fv8hf (__arm_vuninitializedq_f16 (), __a, __b, __p);
-}
-
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x_f32 (float32x4_t __a, float32x4_t __b, mve_pred16_t __p)
-{
-  return __builtin_mve_vcaddq_rot270_m_fv4sf (__arm_vuninitializedq_f32 (), __a, __b, __p);
-}
-
 __extension__ extern __inline float16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vcmulq_x_f16 (float16x8_t __a, float16x8_t __b, mve_pred16_t __p)
@@ -6408,20 +5862,6 @@  __arm_vmullbq_int (uint8x16_t __a, uint8x16_t __b)
  return __arm_vmullbq_int_u8 (__a, __b);
 }
 
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90 (uint8x16_t __a, uint8x16_t __b)
-{
- return __arm_vcaddq_rot90_u8 (__a, __b);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270 (uint8x16_t __a, uint8x16_t __b)
-{
- return __arm_vcaddq_rot270_u8 (__a, __b);
-}
-
 __extension__ extern __inline uint8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq (uint8x16_t __a, uint8x16_t __b)
@@ -6450,34 +5890,6 @@  __arm_vmullbq_int (int8x16_t __a, int8x16_t __b)
  return __arm_vmullbq_int_s8 (__a, __b);
 }
 
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90 (int8x16_t __a, int8x16_t __b)
-{
- return __arm_vhcaddq_rot90_s8 (__a, __b);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270 (int8x16_t __a, int8x16_t __b)
-{
- return __arm_vhcaddq_rot270_s8 (__a, __b);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90 (int8x16_t __a, int8x16_t __b)
-{
- return __arm_vcaddq_rot90_s8 (__a, __b);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270 (int8x16_t __a, int8x16_t __b)
-{
- return __arm_vcaddq_rot270_s8 (__a, __b);
-}
-
 __extension__ extern __inline int8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq (int8x16_t __a, int8x16_t __b)
@@ -6506,20 +5918,6 @@  __arm_vmullbq_int (uint16x8_t __a, uint16x8_t __b)
  return __arm_vmullbq_int_u16 (__a, __b);
 }
 
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90 (uint16x8_t __a, uint16x8_t __b)
-{
- return __arm_vcaddq_rot90_u16 (__a, __b);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270 (uint16x8_t __a, uint16x8_t __b)
-{
- return __arm_vcaddq_rot270_u16 (__a, __b);
-}
-
 __extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq (uint16x8_t __a, uint16x8_t __b)
@@ -6548,34 +5946,6 @@  __arm_vmullbq_int (int16x8_t __a, int16x8_t __b)
  return __arm_vmullbq_int_s16 (__a, __b);
 }
 
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90 (int16x8_t __a, int16x8_t __b)
-{
- return __arm_vhcaddq_rot90_s16 (__a, __b);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270 (int16x8_t __a, int16x8_t __b)
-{
- return __arm_vhcaddq_rot270_s16 (__a, __b);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90 (int16x8_t __a, int16x8_t __b)
-{
- return __arm_vcaddq_rot90_s16 (__a, __b);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270 (int16x8_t __a, int16x8_t __b)
-{
- return __arm_vcaddq_rot270_s16 (__a, __b);
-}
-
 __extension__ extern __inline int16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq (int16x8_t __a, int16x8_t __b)
@@ -6604,20 +5974,6 @@  __arm_vmullbq_int (uint32x4_t __a, uint32x4_t __b)
  return __arm_vmullbq_int_u32 (__a, __b);
 }
 
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90 (uint32x4_t __a, uint32x4_t __b)
-{
- return __arm_vcaddq_rot90_u32 (__a, __b);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270 (uint32x4_t __a, uint32x4_t __b)
-{
- return __arm_vcaddq_rot270_u32 (__a, __b);
-}
-
 __extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq (uint32x4_t __a, uint32x4_t __b)
@@ -6646,34 +6002,6 @@  __arm_vmullbq_int (int32x4_t __a, int32x4_t __b)
  return __arm_vmullbq_int_s32 (__a, __b);
 }
 
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90 (int32x4_t __a, int32x4_t __b)
-{
- return __arm_vhcaddq_rot90_s32 (__a, __b);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270 (int32x4_t __a, int32x4_t __b)
-{
- return __arm_vhcaddq_rot270_s32 (__a, __b);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90 (int32x4_t __a, int32x4_t __b)
-{
- return __arm_vcaddq_rot90_s32 (__a, __b);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270 (int32x4_t __a, int32x4_t __b)
-{
- return __arm_vcaddq_rot270_s32 (__a, __b);
-}
-
 __extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq (int32x4_t __a, int32x4_t __b)
@@ -6751,228 +6079,102 @@  __arm_vbicq_m_n (int32x4_t __a, const int __imm, mve_pred16_t __p)
  return __arm_vbicq_m_n_s32 (__a, __imm, __p);
 }
 
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vbicq_m_n (uint16x8_t __a, const int __imm, mve_pred16_t __p)
-{
- return __arm_vbicq_m_n_u16 (__a, __imm, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vbicq_m_n (uint32x4_t __a, const int __imm, mve_pred16_t __p)
-{
- return __arm_vbicq_m_n_u32 (__a, __imm, __p);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vshlcq (int8x16_t __a, uint32_t * __b, const int __imm)
-{
- return __arm_vshlcq_s8 (__a, __b, __imm);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vshlcq (uint8x16_t __a, uint32_t * __b, const int __imm)
-{
- return __arm_vshlcq_u8 (__a, __b, __imm);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vshlcq (int16x8_t __a, uint32_t * __b, const int __imm)
-{
- return __arm_vshlcq_s16 (__a, __b, __imm);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vshlcq (uint16x8_t __a, uint32_t * __b, const int __imm)
-{
- return __arm_vshlcq_u16 (__a, __b, __imm);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vshlcq (int32x4_t __a, uint32_t * __b, const int __imm)
-{
- return __arm_vshlcq_s32 (__a, __b, __imm);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vshlcq (uint32x4_t __a, uint32_t * __b, const int __imm)
-{
- return __arm_vshlcq_u32 (__a, __b, __imm);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vbicq_m (int8x16_t __inactive, int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
-{
- return __arm_vbicq_m_s8 (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vbicq_m (int32x4_t __inactive, int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vbicq_m_s32 (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vbicq_m (int16x8_t __inactive, int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vbicq_m_s16 (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vbicq_m (uint8x16_t __inactive, uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
-{
- return __arm_vbicq_m_u8 (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vbicq_m (uint32x4_t __inactive, uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vbicq_m_u32 (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vbicq_m (uint16x8_t __inactive, uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vbicq_m_u16 (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m (int8x16_t __inactive, int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot270_m_s8 (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m (int32x4_t __inactive, int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot270_m_s32 (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m (int16x8_t __inactive, int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot270_m_s16 (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m (uint8x16_t __inactive, uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot270_m_u8 (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline uint32x4_t
+__extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m (uint32x4_t __inactive, uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
+__arm_vbicq_m_n (uint16x8_t __a, const int __imm, mve_pred16_t __p)
 {
- return __arm_vcaddq_rot270_m_u32 (__inactive, __a, __b, __p);
+ return __arm_vbicq_m_n_u16 (__a, __imm, __p);
 }
 
-__extension__ extern __inline uint16x8_t
+__extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m (uint16x8_t __inactive, uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
+__arm_vbicq_m_n (uint32x4_t __a, const int __imm, mve_pred16_t __p)
 {
- return __arm_vcaddq_rot270_m_u16 (__inactive, __a, __b, __p);
+ return __arm_vbicq_m_n_u32 (__a, __imm, __p);
 }
 
 __extension__ extern __inline int8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m (int8x16_t __inactive, int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
+__arm_vshlcq (int8x16_t __a, uint32_t * __b, const int __imm)
 {
- return __arm_vcaddq_rot90_m_s8 (__inactive, __a, __b, __p);
+ return __arm_vshlcq_s8 (__a, __b, __imm);
 }
 
-__extension__ extern __inline int32x4_t
+__extension__ extern __inline uint8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m (int32x4_t __inactive, int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
+__arm_vshlcq (uint8x16_t __a, uint32_t * __b, const int __imm)
 {
- return __arm_vcaddq_rot90_m_s32 (__inactive, __a, __b, __p);
+ return __arm_vshlcq_u8 (__a, __b, __imm);
 }
 
 __extension__ extern __inline int16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m (int16x8_t __inactive, int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
+__arm_vshlcq (int16x8_t __a, uint32_t * __b, const int __imm)
 {
- return __arm_vcaddq_rot90_m_s16 (__inactive, __a, __b, __p);
+ return __arm_vshlcq_s16 (__a, __b, __imm);
 }
 
-__extension__ extern __inline uint8x16_t
+__extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m (uint8x16_t __inactive, uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
+__arm_vshlcq (uint16x8_t __a, uint32_t * __b, const int __imm)
 {
- return __arm_vcaddq_rot90_m_u8 (__inactive, __a, __b, __p);
+ return __arm_vshlcq_u16 (__a, __b, __imm);
 }
 
-__extension__ extern __inline uint32x4_t
+__extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m (uint32x4_t __inactive, uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
+__arm_vshlcq (int32x4_t __a, uint32_t * __b, const int __imm)
 {
- return __arm_vcaddq_rot90_m_u32 (__inactive, __a, __b, __p);
+ return __arm_vshlcq_s32 (__a, __b, __imm);
 }
 
-__extension__ extern __inline uint16x8_t
+__extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m (uint16x8_t __inactive, uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
+__arm_vshlcq (uint32x4_t __a, uint32_t * __b, const int __imm)
 {
- return __arm_vcaddq_rot90_m_u16 (__inactive, __a, __b, __p);
+ return __arm_vshlcq_u32 (__a, __b, __imm);
 }
 
 __extension__ extern __inline int8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_m (int8x16_t __inactive, int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
+__arm_vbicq_m (int8x16_t __inactive, int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
 {
- return __arm_vhcaddq_rot270_m_s8 (__inactive, __a, __b, __p);
+ return __arm_vbicq_m_s8 (__inactive, __a, __b, __p);
 }
 
 __extension__ extern __inline int32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_m (int32x4_t __inactive, int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
+__arm_vbicq_m (int32x4_t __inactive, int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
 {
- return __arm_vhcaddq_rot270_m_s32 (__inactive, __a, __b, __p);
+ return __arm_vbicq_m_s32 (__inactive, __a, __b, __p);
 }
 
 __extension__ extern __inline int16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_m (int16x8_t __inactive, int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
+__arm_vbicq_m (int16x8_t __inactive, int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
 {
- return __arm_vhcaddq_rot270_m_s16 (__inactive, __a, __b, __p);
+ return __arm_vbicq_m_s16 (__inactive, __a, __b, __p);
 }
 
-__extension__ extern __inline int8x16_t
+__extension__ extern __inline uint8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_m (int8x16_t __inactive, int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
+__arm_vbicq_m (uint8x16_t __inactive, uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
 {
- return __arm_vhcaddq_rot90_m_s8 (__inactive, __a, __b, __p);
+ return __arm_vbicq_m_u8 (__inactive, __a, __b, __p);
 }
 
-__extension__ extern __inline int32x4_t
+__extension__ extern __inline uint32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_m (int32x4_t __inactive, int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
+__arm_vbicq_m (uint32x4_t __inactive, uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
 {
- return __arm_vhcaddq_rot90_m_s32 (__inactive, __a, __b, __p);
+ return __arm_vbicq_m_u32 (__inactive, __a, __b, __p);
 }
 
-__extension__ extern __inline int16x8_t
+__extension__ extern __inline uint16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_m (int16x8_t __inactive, int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
+__arm_vbicq_m (uint16x8_t __inactive, uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
 {
- return __arm_vhcaddq_rot90_m_s16 (__inactive, __a, __b, __p);
+ return __arm_vbicq_m_u16 (__inactive, __a, __b, __p);
 }
 
 __extension__ extern __inline int16x8_t
@@ -8725,132 +7927,6 @@  __arm_vmulltq_int_x (uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
  return __arm_vmulltq_int_x_u32 (__a, __b, __p);
 }
 
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x (int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot90_x_s8 (__a, __b, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot90_x_s16 (__a, __b, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot90_x_s32 (__a, __b, __p);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x (uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot90_x_u8 (__a, __b, __p);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x (uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot90_x_u16 (__a, __b, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x (uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot90_x_u32 (__a, __b, __p);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x (int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot270_x_s8 (__a, __b, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot270_x_s16 (__a, __b, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot270_x_s32 (__a, __b, __p);
-}
-
-__extension__ extern __inline uint8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x (uint8x16_t __a, uint8x16_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot270_x_u8 (__a, __b, __p);
-}
-
-__extension__ extern __inline uint16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x (uint16x8_t __a, uint16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot270_x_u16 (__a, __b, __p);
-}
-
-__extension__ extern __inline uint32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x (uint32x4_t __a, uint32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot270_x_u32 (__a, __b, __p);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_x (int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
-{
- return __arm_vhcaddq_rot90_x_s8 (__a, __b, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_x (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vhcaddq_rot90_x_s16 (__a, __b, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot90_x (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vhcaddq_rot90_x_s32 (__a, __b, __p);
-}
-
-__extension__ extern __inline int8x16_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_x (int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
-{
- return __arm_vhcaddq_rot270_x_s8 (__a, __b, __p);
-}
-
-__extension__ extern __inline int16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_x (int16x8_t __a, int16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vhcaddq_rot270_x_s16 (__a, __b, __p);
-}
-
-__extension__ extern __inline int32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vhcaddq_rot270_x (int32x4_t __a, int32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vhcaddq_rot270_x_s32 (__a, __b, __p);
-}
-
 __extension__ extern __inline int8x16_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq_x (int8x16_t __a, int8x16_t __b, mve_pred16_t __p)
@@ -9532,20 +8608,6 @@  __arm_vcmulq (float16x8_t __a, float16x8_t __b)
  return __arm_vcmulq_f16 (__a, __b);
 }
 
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90 (float16x8_t __a, float16x8_t __b)
-{
- return __arm_vcaddq_rot90_f16 (__a, __b);
-}
-
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270 (float16x8_t __a, float16x8_t __b)
-{
- return __arm_vcaddq_rot270_f16 (__a, __b);
-}
-
 __extension__ extern __inline float16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq (float16x8_t __a, float16x8_t __b)
@@ -9588,20 +8650,6 @@  __arm_vcmulq (float32x4_t __a, float32x4_t __b)
  return __arm_vcmulq_f32 (__a, __b);
 }
 
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90 (float32x4_t __a, float32x4_t __b)
-{
- return __arm_vcaddq_rot90_f32 (__a, __b);
-}
-
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270 (float32x4_t __a, float32x4_t __b)
-{
- return __arm_vcaddq_rot270_f32 (__a, __b);
-}
-
 __extension__ extern __inline float32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vbicq (float32x4_t __a, float32x4_t __b)
@@ -9903,34 +8951,6 @@  __arm_vbicq_m (float16x8_t __inactive, float16x8_t __a, float16x8_t __b, mve_pre
  return __arm_vbicq_m_f16 (__inactive, __a, __b, __p);
 }
 
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m (float32x4_t __inactive, float32x4_t __a, float32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot270_m_f32 (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_m (float16x8_t __inactive, float16x8_t __a, float16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot270_m_f16 (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m (float32x4_t __inactive, float32x4_t __a, float32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot90_m_f32 (__inactive, __a, __b, __p);
-}
-
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_m (float16x8_t __inactive, float16x8_t __a, float16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot90_m_f16 (__inactive, __a, __b, __p);
-}
-
 __extension__ extern __inline float32x4_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vcmlaq_m (float32x4_t __a, float32x4_t __b, float32x4_t __c, mve_pred16_t __p)
@@ -10281,34 +9301,6 @@  __arm_vstrwq_scatter_base_wb_p (uint32x4_t * __addr, const int __offset, float32
  __arm_vstrwq_scatter_base_wb_p_f32 (__addr, __offset, __value, __p);
 }
 
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x (float16x8_t __a, float16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot90_x_f16 (__a, __b, __p);
-}
-
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot90_x (float32x4_t __a, float32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot90_x_f32 (__a, __b, __p);
-}
-
-__extension__ extern __inline float16x8_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x (float16x8_t __a, float16x8_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot270_x_f16 (__a, __b, __p);
-}
-
-__extension__ extern __inline float32x4_t
-__attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
-__arm_vcaddq_rot270_x (float32x4_t __a, float32x4_t __b, mve_pred16_t __p)
-{
- return __arm_vcaddq_rot270_x_f32 (__a, __b, __p);
-}
-
 __extension__ extern __inline float16x8_t
 __attribute__ ((__always_inline__, __gnu_inline__, __artificial__))
 __arm_vcmulq_x (float16x8_t __a, float16x8_t __b, mve_pred16_t __p)
@@ -10920,30 +9912,6 @@  extern void *__ARM_undef;
   int (*)[__ARM_mve_type_float16x8_t][__ARM_mve_type_float16x8_t]: __arm_vornq_f16 (__ARM_mve_coerce(__p0, float16x8_t), __ARM_mve_coerce(__p1, float16x8_t)), \
   int (*)[__ARM_mve_type_float32x4_t][__ARM_mve_type_float32x4_t]: __arm_vornq_f32 (__ARM_mve_coerce(__p0, float32x4_t), __ARM_mve_coerce(__p1, float32x4_t)));})
 
-#define __arm_vcaddq_rot270(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_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vcaddq_rot270_s8 (__ARM_mve_coerce(__p0, int8x16_t), __ARM_mve_coerce(__p1, int8x16_t)), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vcaddq_rot270_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_vcaddq_rot270_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t)), \
-  int (*)[__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t]: __arm_vcaddq_rot270_u8 (__ARM_mve_coerce(__p0, uint8x16_t), __ARM_mve_coerce(__p1, uint8x16_t)), \
-  int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vcaddq_rot270_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_vcaddq_rot270_u32 (__ARM_mve_coerce(__p0, uint32x4_t), __ARM_mve_coerce(__p1, uint32x4_t)), \
-  int (*)[__ARM_mve_type_float16x8_t][__ARM_mve_type_float16x8_t]: __arm_vcaddq_rot270_f16 (__ARM_mve_coerce(__p0, float16x8_t), __ARM_mve_coerce(__p1, float16x8_t)), \
-  int (*)[__ARM_mve_type_float32x4_t][__ARM_mve_type_float32x4_t]: __arm_vcaddq_rot270_f32 (__ARM_mve_coerce(__p0, float32x4_t), __ARM_mve_coerce(__p1, float32x4_t)));})
-
-#define __arm_vcaddq_rot90(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_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vcaddq_rot90_s8 (__ARM_mve_coerce(__p0, int8x16_t), __ARM_mve_coerce(__p1, int8x16_t)), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vcaddq_rot90_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_vcaddq_rot90_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t)), \
-  int (*)[__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t]: __arm_vcaddq_rot90_u8 (__ARM_mve_coerce(__p0, uint8x16_t), __ARM_mve_coerce(__p1, uint8x16_t)), \
-  int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vcaddq_rot90_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_vcaddq_rot90_u32 (__ARM_mve_coerce(__p0, uint32x4_t), __ARM_mve_coerce(__p1, uint32x4_t)), \
-  int (*)[__ARM_mve_type_float16x8_t][__ARM_mve_type_float16x8_t]: __arm_vcaddq_rot90_f16 (__ARM_mve_coerce(__p0, float16x8_t), __ARM_mve_coerce(__p1, float16x8_t)), \
-  int (*)[__ARM_mve_type_float32x4_t][__ARM_mve_type_float32x4_t]: __arm_vcaddq_rot90_f32 (__ARM_mve_coerce(__p0, float32x4_t), __ARM_mve_coerce(__p1, float32x4_t)));})
-
 #define __arm_vcmulq(p0,p1) ({ __typeof(p0) __p0 = (p0); \
   __typeof(p1) __p1 = (p1); \
   _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)])0, \
@@ -10990,20 +9958,6 @@  extern void *__ARM_undef;
   int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vmulltq_int_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_vmulltq_int_u32 (__ARM_mve_coerce(__p0, uint32x4_t), __ARM_mve_coerce(__p1, uint32x4_t)));})
 
-#define __arm_vhcaddq_rot270(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_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vhcaddq_rot270_s8 (__ARM_mve_coerce(__p0, int8x16_t), __ARM_mve_coerce(__p1, int8x16_t)), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vhcaddq_rot270_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_vhcaddq_rot270_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t)));})
-
-#define __arm_vhcaddq_rot90(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_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vhcaddq_rot90_s8 (__ARM_mve_coerce(__p0, int8x16_t), __ARM_mve_coerce(__p1, int8x16_t)), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vhcaddq_rot90_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_vhcaddq_rot90_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t)));})
-
 #define __arm_vmullbq_int(p0,p1) ({ __typeof(p0) __p0 = (p0); \
   __typeof(p1) __p1 = (p1); \
   _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)])0, \
@@ -11139,32 +10093,6 @@  extern void *__ARM_undef;
   int (*)[__ARM_mve_type_float16x8_t][__ARM_mve_type_float16x8_t][__ARM_mve_type_float16x8_t]: __arm_vbicq_m_f16 (__ARM_mve_coerce(__p0, float16x8_t), __ARM_mve_coerce(__p1, float16x8_t), __ARM_mve_coerce(__p2, float16x8_t), p3), \
   int (*)[__ARM_mve_type_float32x4_t][__ARM_mve_type_float32x4_t][__ARM_mve_type_float32x4_t]: __arm_vbicq_m_f32 (__ARM_mve_coerce(__p0, float32x4_t), __ARM_mve_coerce(__p1, float32x4_t), __ARM_mve_coerce(__p2, float32x4_t), p3));})
 
-#define __arm_vcaddq_rot270_m(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_int8x16_t][__ARM_mve_type_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vcaddq_rot270_m_s8 (__ARM_mve_coerce(__p0, int8x16_t), __ARM_mve_coerce(__p1, int8x16_t), __ARM_mve_coerce(__p2, int8x16_t), p3), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vcaddq_rot270_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_vcaddq_rot270_m_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3), \
-  int (*)[__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t]: __arm_vcaddq_rot270_m_u8 (__ARM_mve_coerce(__p0, uint8x16_t), __ARM_mve_coerce(__p1, uint8x16_t), __ARM_mve_coerce(__p2, uint8x16_t), p3), \
-  int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vcaddq_rot270_m_u16 (__ARM_mve_coerce(__p0, uint16x8_t), __ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, uint16x8_t), p3), \
-  int (*)[__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vcaddq_rot270_m_u32 (__ARM_mve_coerce(__p0, uint32x4_t), __ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t), p3), \
-  int (*)[__ARM_mve_type_float16x8_t][__ARM_mve_type_float16x8_t][__ARM_mve_type_float16x8_t]: __arm_vcaddq_rot270_m_f16 (__ARM_mve_coerce(__p0, float16x8_t), __ARM_mve_coerce(__p1, float16x8_t), __ARM_mve_coerce(__p2, float16x8_t), p3), \
-  int (*)[__ARM_mve_type_float32x4_t][__ARM_mve_type_float32x4_t][__ARM_mve_type_float32x4_t]: __arm_vcaddq_rot270_m_f32 (__ARM_mve_coerce(__p0, float32x4_t), __ARM_mve_coerce(__p1, float32x4_t), __ARM_mve_coerce(__p2, float32x4_t), p3));})
-
-#define __arm_vcaddq_rot90_m(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_int8x16_t][__ARM_mve_type_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vcaddq_rot90_m_s8 (__ARM_mve_coerce(__p0, int8x16_t), __ARM_mve_coerce(__p1, int8x16_t), __ARM_mve_coerce(__p2, int8x16_t), p3), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vcaddq_rot90_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_vcaddq_rot90_m_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3), \
-  int (*)[__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t]: __arm_vcaddq_rot90_m_u8 (__ARM_mve_coerce(__p0, uint8x16_t), __ARM_mve_coerce(__p1, uint8x16_t), __ARM_mve_coerce(__p2, uint8x16_t), p3), \
-  int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vcaddq_rot90_m_u16 (__ARM_mve_coerce(__p0, uint16x8_t), __ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, uint16x8_t), p3), \
-  int (*)[__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vcaddq_rot90_m_u32 (__ARM_mve_coerce(__p0, uint32x4_t), __ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t), p3), \
-  int (*)[__ARM_mve_type_float16x8_t][__ARM_mve_type_float16x8_t][__ARM_mve_type_float16x8_t]: __arm_vcaddq_rot90_m_f16 (__ARM_mve_coerce(__p0, float16x8_t), __ARM_mve_coerce(__p1, float16x8_t), __ARM_mve_coerce(__p2, float16x8_t), p3), \
-  int (*)[__ARM_mve_type_float32x4_t][__ARM_mve_type_float32x4_t][__ARM_mve_type_float32x4_t]: __arm_vcaddq_rot90_m_f32 (__ARM_mve_coerce(__p0, float32x4_t), __ARM_mve_coerce(__p1, float32x4_t), __ARM_mve_coerce(__p2, float32x4_t), p3));})
-
 #define __arm_vcmlaq_m(p0,p1,p2,p3) ({ __typeof(p0) __p0 = (p0); \
   __typeof(p1) __p1 = (p1); \
   __typeof(p2) __p2 = (p2); \
@@ -11558,30 +10486,6 @@  extern void *__ARM_undef;
   int (*)[__ARM_mve_type_float16x8_t][__ARM_mve_type_float16x8_t]: __arm_vbicq_x_f16 (__ARM_mve_coerce(__p1, float16x8_t), __ARM_mve_coerce(__p2, float16x8_t), p3), \
   int (*)[__ARM_mve_type_float32x4_t][__ARM_mve_type_float32x4_t]: __arm_vbicq_x_f32 (__ARM_mve_coerce(__p1, float32x4_t), __ARM_mve_coerce(__p2, float32x4_t), p3));})
 
-#define __arm_vcaddq_rot270_x(p1,p2,p3) ({ __typeof(p1) __p1 = (p1); \
-  __typeof(p2) __p2 = (p2); \
-  _Generic( (int (*)[__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
-  int (*)[__ARM_mve_type_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vcaddq_rot270_x_s8 (__ARM_mve_coerce(__p1, int8x16_t), __ARM_mve_coerce(__p2, int8x16_t), p3), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vcaddq_rot270_x_s16 (__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_vcaddq_rot270_x_s32 (__ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3), \
-  int (*)[__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t]: __arm_vcaddq_rot270_x_u8 (__ARM_mve_coerce(__p1, uint8x16_t), __ARM_mve_coerce(__p2, uint8x16_t), p3), \
-  int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vcaddq_rot270_x_u16 (__ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, uint16x8_t), p3), \
-  int (*)[__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vcaddq_rot270_x_u32 (__ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t), p3), \
-  int (*)[__ARM_mve_type_float16x8_t][__ARM_mve_type_float16x8_t]: __arm_vcaddq_rot270_x_f16 (__ARM_mve_coerce(__p1, float16x8_t), __ARM_mve_coerce(__p2, float16x8_t), p3), \
-  int (*)[__ARM_mve_type_float32x4_t][__ARM_mve_type_float32x4_t]: __arm_vcaddq_rot270_x_f32 (__ARM_mve_coerce(__p1, float32x4_t), __ARM_mve_coerce(__p2, float32x4_t), p3));})
-
-#define __arm_vcaddq_rot90_x(p1,p2,p3) ({ __typeof(p1) __p1 = (p1); \
-  __typeof(p2) __p2 = (p2); \
-  _Generic( (int (*)[__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
-  int (*)[__ARM_mve_type_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vcaddq_rot90_x_s8 (__ARM_mve_coerce(__p1, int8x16_t), __ARM_mve_coerce(__p2, int8x16_t), p3), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vcaddq_rot90_x_s16 (__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_vcaddq_rot90_x_s32 (__ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3), \
-  int (*)[__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t]: __arm_vcaddq_rot90_x_u8 (__ARM_mve_coerce(__p1, uint8x16_t), __ARM_mve_coerce(__p2, uint8x16_t), p3), \
-  int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vcaddq_rot90_x_u16 (__ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, uint16x8_t), p3), \
-  int (*)[__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vcaddq_rot90_x_u32 (__ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t), p3), \
-  int (*)[__ARM_mve_type_float16x8_t][__ARM_mve_type_float16x8_t]: __arm_vcaddq_rot90_x_f16 (__ARM_mve_coerce(__p1, float16x8_t), __ARM_mve_coerce(__p2, float16x8_t), p3), \
-  int (*)[__ARM_mve_type_float32x4_t][__ARM_mve_type_float32x4_t]: __arm_vcaddq_rot90_x_f32 (__ARM_mve_coerce(__p1, float32x4_t), __ARM_mve_coerce(__p2, float32x4_t), p3));})
-
 #define __arm_vcmulq_rot180_x(p1,p2,p3) ({ __typeof(p1) __p1 = (p1); \
   __typeof(p2) __p2 = (p2); \
   _Generic( (int (*)[__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
@@ -11710,40 +10614,6 @@  extern void *__ARM_undef;
   int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vmullbq_int_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_vmullbq_int_u32 (__ARM_mve_coerce(__p0, uint32x4_t), __ARM_mve_coerce(__p1, uint32x4_t)));})
 
-#define __arm_vhcaddq_rot90(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_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vhcaddq_rot90_s8 (__ARM_mve_coerce(__p0, int8x16_t), __ARM_mve_coerce(__p1, int8x16_t)), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vhcaddq_rot90_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_vhcaddq_rot90_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t)));})
-
-#define __arm_vhcaddq_rot270(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_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vhcaddq_rot270_s8 (__ARM_mve_coerce(__p0, int8x16_t), __ARM_mve_coerce(__p1, int8x16_t)), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vhcaddq_rot270_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_vhcaddq_rot270_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t)));})
-
-#define __arm_vcaddq_rot90(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_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vcaddq_rot90_s8 (__ARM_mve_coerce(__p0, int8x16_t), __ARM_mve_coerce(__p1, int8x16_t)), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vcaddq_rot90_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_vcaddq_rot90_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t)), \
-  int (*)[__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t]: __arm_vcaddq_rot90_u8 (__ARM_mve_coerce(__p0, uint8x16_t), __ARM_mve_coerce(__p1, uint8x16_t)), \
-  int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vcaddq_rot90_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_vcaddq_rot90_u32 (__ARM_mve_coerce(__p0, uint32x4_t), __ARM_mve_coerce(__p1, uint32x4_t)));})
-
-#define __arm_vcaddq_rot270(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_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vcaddq_rot270_s8 (__ARM_mve_coerce(__p0, int8x16_t), __ARM_mve_coerce(__p1, int8x16_t)), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vcaddq_rot270_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_vcaddq_rot270_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t)), \
-  int (*)[__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t]: __arm_vcaddq_rot270_u8 (__ARM_mve_coerce(__p0, uint8x16_t), __ARM_mve_coerce(__p1, uint8x16_t)), \
-  int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vcaddq_rot270_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_vcaddq_rot270_u32 (__ARM_mve_coerce(__p0, uint32x4_t), __ARM_mve_coerce(__p1, uint32x4_t)));})
-
 #define __arm_vbicq(p0,p1) ({ __typeof(p0) __p0 = (p0); \
   __typeof(p1) __p1 = (p1); \
   _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)])0, \
@@ -11797,28 +10667,6 @@  extern void *__ARM_undef;
   int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vbicq_m_u16 (__ARM_mve_coerce(__p0, uint16x8_t), __ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, uint16x8_t), p3), \
   int (*)[__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vbicq_m_u32 (__ARM_mve_coerce(__p0, uint32x4_t), __ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t), p3));})
 
-#define __arm_vcaddq_rot270_m(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_int8x16_t][__ARM_mve_type_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vcaddq_rot270_m_s8 (__ARM_mve_coerce(__p0, int8x16_t), __ARM_mve_coerce(__p1, int8x16_t), __ARM_mve_coerce(__p2, int8x16_t), p3), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vcaddq_rot270_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_vcaddq_rot270_m_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3), \
-  int (*)[__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t]: __arm_vcaddq_rot270_m_u8 (__ARM_mve_coerce(__p0, uint8x16_t), __ARM_mve_coerce(__p1, uint8x16_t), __ARM_mve_coerce(__p2, uint8x16_t), p3), \
-  int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vcaddq_rot270_m_u16 (__ARM_mve_coerce(__p0, uint16x8_t), __ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, uint16x8_t), p3), \
-  int (*)[__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vcaddq_rot270_m_u32 (__ARM_mve_coerce(__p0, uint32x4_t), __ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t), p3));})
-
-#define __arm_vcaddq_rot90_m(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_int8x16_t][__ARM_mve_type_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vcaddq_rot90_m_s8 (__ARM_mve_coerce(__p0, int8x16_t), __ARM_mve_coerce(__p1, int8x16_t), __ARM_mve_coerce(__p2, int8x16_t), p3), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vcaddq_rot90_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_vcaddq_rot90_m_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3), \
-  int (*)[__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t]: __arm_vcaddq_rot90_m_u8 (__ARM_mve_coerce(__p0, uint8x16_t), __ARM_mve_coerce(__p1, uint8x16_t), __ARM_mve_coerce(__p2, uint8x16_t), p3), \
-  int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vcaddq_rot90_m_u16 (__ARM_mve_coerce(__p0, uint16x8_t), __ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, uint16x8_t), p3), \
-  int (*)[__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vcaddq_rot90_m_u32 (__ARM_mve_coerce(__p0, uint32x4_t), __ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t), p3));})
-
 #define __arm_vornq_m(p0,p1,p2,p3) ({ __typeof(p0) __p0 = (p0); \
   __typeof(p1) __p1 = (p1); \
   __typeof(p2) __p2 = (p2); \
@@ -12067,26 +10915,6 @@  extern void *__ARM_undef;
   int (*)[__ARM_mve_type_uint32x4_t]: __arm_vuninitializedq_u32 (), \
   int (*)[__ARM_mve_type_uint64x2_t]: __arm_vuninitializedq_u64 ());})
 
-#define __arm_vcaddq_rot270_x(p1,p2,p3) ({ __typeof(p1) __p1 = (p1); \
-  __typeof(p2) __p2 = (p2); \
-  _Generic( (int (*)[__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
-  int (*)[__ARM_mve_type_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vcaddq_rot270_x_s8 (__ARM_mve_coerce(__p1, int8x16_t), __ARM_mve_coerce(__p2, int8x16_t), p3), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vcaddq_rot270_x_s16 (__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_vcaddq_rot270_x_s32 (__ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3), \
-  int (*)[__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t]: __arm_vcaddq_rot270_x_u8 (__ARM_mve_coerce(__p1, uint8x16_t), __ARM_mve_coerce(__p2, uint8x16_t), p3), \
-  int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vcaddq_rot270_x_u16 (__ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, uint16x8_t), p3), \
-  int (*)[__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vcaddq_rot270_x_u32 (__ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t), p3));})
-
-#define __arm_vcaddq_rot90_x(p1,p2,p3) ({ __typeof(p1) __p1 = (p1); \
-  __typeof(p2) __p2 = (p2); \
-  _Generic( (int (*)[__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
-  int (*)[__ARM_mve_type_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vcaddq_rot90_x_s8 (__ARM_mve_coerce(__p1, int8x16_t), __ARM_mve_coerce(__p2, int8x16_t), p3), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vcaddq_rot90_x_s16 (__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_vcaddq_rot90_x_s32 (__ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3), \
-  int (*)[__ARM_mve_type_uint8x16_t][__ARM_mve_type_uint8x16_t]: __arm_vcaddq_rot90_x_u8 (__ARM_mve_coerce(__p1, uint8x16_t), __ARM_mve_coerce(__p2, uint8x16_t), p3), \
-  int (*)[__ARM_mve_type_uint16x8_t][__ARM_mve_type_uint16x8_t]: __arm_vcaddq_rot90_x_u16 (__ARM_mve_coerce(__p1, uint16x8_t), __ARM_mve_coerce(__p2, uint16x8_t), p3), \
-  int (*)[__ARM_mve_type_uint32x4_t][__ARM_mve_type_uint32x4_t]: __arm_vcaddq_rot90_x_u32 (__ARM_mve_coerce(__p1, uint32x4_t), __ARM_mve_coerce(__p2, uint32x4_t), p3));})
-
 #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, \
@@ -12251,20 +11079,6 @@  extern void *__ARM_undef;
   int (*)[__ARM_mve_type_int_n]: __arm_vddupq_x_n_u32 ((uint32_t) __p1, p2, p3), \
   int (*)[__ARM_mve_type_uint32_t_ptr]: __arm_vddupq_x_wb_u32 (__ARM_mve_coerce_u32_ptr(__p1, uint32_t *), p2, p3));})
 
-#define __arm_vhcaddq_rot270_x(p1,p2,p3) ({ __typeof(p1) __p1 = (p1); \
-  __typeof(p2) __p2 = (p2); \
-  _Generic( (int (*)[__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
-  int (*)[__ARM_mve_type_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vhcaddq_rot270_x_s8 (__ARM_mve_coerce(__p1, int8x16_t), __ARM_mve_coerce(__p2, int8x16_t), p3), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vhcaddq_rot270_x_s16 (__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_vhcaddq_rot270_x_s32 (__ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3));})
-
-#define __arm_vhcaddq_rot90_x(p1,p2,p3) ({ __typeof(p1) __p1 = (p1); \
-  __typeof(p2) __p2 = (p2); \
-  _Generic( (int (*)[__ARM_mve_typeid(__p1)][__ARM_mve_typeid(__p2)])0, \
-  int (*)[__ARM_mve_type_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vhcaddq_rot90_x_s8 (__ARM_mve_coerce(__p1, int8x16_t), __ARM_mve_coerce(__p2, int8x16_t), p3), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vhcaddq_rot90_x_s16 (__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_vhcaddq_rot90_x_s32 (__ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3));})
-
 #define __arm_vadciq(p0,p1,p2) ({ __typeof(p0) __p0 = (p0); \
   __typeof(p1) __p1 = (p1); \
   _Generic( (int (*)[__ARM_mve_typeid(__p0)][__ARM_mve_typeid(__p1)])0, \
@@ -12358,22 +11172,6 @@  extern void *__ARM_undef;
   int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint16x8_t]: __arm_vldrbq_gather_offset_z_u16 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint16x8_t), p2), \
   int (*)[__ARM_mve_type_uint8_t_ptr][__ARM_mve_type_uint32x4_t]: __arm_vldrbq_gather_offset_z_u32 (__ARM_mve_coerce_u8_ptr(p0, uint8_t *), __ARM_mve_coerce(__p1, uint32x4_t), p2));})
 
-#define __arm_vhcaddq_rot270_m(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_int8x16_t][__ARM_mve_type_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vhcaddq_rot270_m_s8 (__ARM_mve_coerce(__p0, int8x16_t), __ARM_mve_coerce(__p1, int8x16_t), __ARM_mve_coerce(__p2, int8x16_t), p3), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vhcaddq_rot270_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_vhcaddq_rot270_m_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3));})
-
-#define __arm_vhcaddq_rot90_m(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_int8x16_t][__ARM_mve_type_int8x16_t][__ARM_mve_type_int8x16_t]: __arm_vhcaddq_rot90_m_s8 (__ARM_mve_coerce(__p0, int8x16_t), __ARM_mve_coerce(__p1, int8x16_t), __ARM_mve_coerce(__p2, int8x16_t), p3), \
-  int (*)[__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t][__ARM_mve_type_int16x8_t]: __arm_vhcaddq_rot90_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_vhcaddq_rot90_m_s32 (__ARM_mve_coerce(__p0, int32x4_t), __ARM_mve_coerce(__p1, int32x4_t), __ARM_mve_coerce(__p2, int32x4_t), p3));})
-
 #define __arm_vmullbq_int_m(p0,p1,p2,p3) ({ __typeof(p0) __p0 = (p0); \
   __typeof(p1) __p1 = (p1); \
   __typeof(p2) __p2 = (p2); \