diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -3244,6 +3244,9 @@
 
     if (auto *VecTy = QT->getAs<VectorType>())
       QT = VecTy->getElementType();
+    else if (QT->isVLSBuiltinType())
+      QT = QT->getVLSEltType(getContext());
+
     if (QT->isIntegerType())
       Result = Builder.CreateBinaryIntrinsic(
           llvm::Intrinsic::abs, EmitScalarExpr(E->getArg(0)),
@@ -3319,6 +3322,9 @@
     QualType Ty = E->getArg(0)->getType();
     if (auto *VecTy = Ty->getAs<VectorType>())
       Ty = VecTy->getElementType();
+    else if (Ty->isVLSBuiltinType())
+      Ty = Ty->getVLSEltType(getContext());
+
     bool IsSigned = Ty->isSignedIntegerType();
     unsigned Opc;
     if (BuiltinIDIfNoAsmLabel == Builtin::BI__builtin_elementwise_add_sat)
@@ -3337,6 +3343,9 @@
       QualType Ty = E->getArg(0)->getType();
       if (auto *VecTy = Ty->getAs<VectorType>())
         Ty = VecTy->getElementType();
+      else if (Ty->isVLSBuiltinType())
+        Ty = Ty->getVLSEltType(getContext());
+
       Result = Builder.CreateBinaryIntrinsic(Ty->isSignedIntegerType()
                                                  ? llvm::Intrinsic::smax
                                                  : llvm::Intrinsic::umax,
diff --git a/clang/lib/Sema/SemaChecking.cpp b/clang/lib/Sema/SemaChecking.cpp
--- a/clang/lib/Sema/SemaChecking.cpp
+++ b/clang/lib/Sema/SemaChecking.cpp
@@ -2041,7 +2041,8 @@
 // false.
 static bool checkMathBuiltinElementType(Sema &S, SourceLocation Loc,
                                         QualType Ty) {
-  if (!Ty->getAs<VectorType>() && !ConstantMatrixType::isValidElementType(Ty)) {
+  if (!Ty->isVLSBuiltinType() && !Ty->getAs<VectorType>() &&
+      !ConstantMatrixType::isValidElementType(Ty)) {
     return S.Diag(Loc, diag::err_builtin_invalid_arg_type)
            << 1 << /* vector, integer or float ty*/ 0 << Ty;
   }
@@ -2054,6 +2055,8 @@
   QualType EltTy = ArgTy;
   if (auto *VecTy = EltTy->getAs<VectorType>())
     EltTy = VecTy->getElementType();
+  else if (EltTy->isVLSBuiltinType())
+    EltTy = EltTy->getVLSEltType(S.Context);
 
   if (!EltTy->isRealFloatingType()) {
     return S.Diag(Loc, diag::err_builtin_invalid_arg_type)
@@ -2618,6 +2621,9 @@
 
     if (auto *VecTy = EltTy->getAs<VectorType>())
       EltTy = VecTy->getElementType();
+    else if (EltTy->isVLSBuiltinType())
+      EltTy = EltTy->getVLSEltType(Context);
+
     if (EltTy->isUnsignedIntegerType()) {
       Diag(TheCall->getArg(0)->getBeginLoc(),
            diag::err_builtin_invalid_arg_type)
@@ -2688,6 +2694,8 @@
 
     if (auto *VecTy = EltTy->getAs<VectorType>())
       EltTy = VecTy->getElementType();
+    else if (EltTy->isVLSBuiltinType())
+      EltTy = EltTy->getVLSEltType(Context);
 
     if (!EltTy->isIntegerType()) {
       Diag(Arg->getBeginLoc(), diag::err_builtin_invalid_arg_type)
@@ -2713,6 +2721,8 @@
 
     if (auto *VecTy = EltTy->getAs<VectorType>())
       EltTy = VecTy->getElementType();
+    else if (EltTy->isVLSBuiltinType())
+      EltTy = EltTy->getVLSEltType(Context);
 
     if (!EltTy->isIntegerType()) {
       Diag(Arg->getBeginLoc(), diag::err_builtin_invalid_arg_type)
diff --git a/clang/test/CodeGen/aarch64-sve-vector-elementwise-ops.c b/clang/test/CodeGen/aarch64-sve-vector-elementwise-ops.c
new file mode 100644
--- /dev/null
+++ b/clang/test/CodeGen/aarch64-sve-vector-elementwise-ops.c
@@ -0,0 +1,528 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve \
+// RUN: -disable-O0-optnone \
+// RUN:  -emit-llvm -o - %s | opt -S -passes=sroa | FileCheck %s
+
+// REQUIRES: aarch64-registered-target
+
+#include <arm_sve.h>
+
+// CHECK-LABEL: @test_builtin_elementwise_abs(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_ABS:%.*]] = call <vscale x 16 x i8> @llvm.abs.nxv16i8(<vscale x 16 x i8> [[VI8:%.*]], i1 false)
+// CHECK-NEXT:    [[ELT_ABS1:%.*]] = call <vscale x 8 x i16> @llvm.abs.nxv8i16(<vscale x 8 x i16> [[VI16:%.*]], i1 false)
+// CHECK-NEXT:    [[ELT_ABS2:%.*]] = call <vscale x 4 x i32> @llvm.abs.nxv4i32(<vscale x 4 x i32> [[VI32:%.*]], i1 false)
+// CHECK-NEXT:    [[ELT_ABS3:%.*]] = call <vscale x 2 x i64> @llvm.abs.nxv2i64(<vscale x 2 x i64> [[VI64:%.*]], i1 false)
+// CHECK-NEXT:    [[ELT_ABS4:%.*]] = call <vscale x 8 x half> @llvm.fabs.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_ABS5:%.*]] = call <vscale x 4 x float> @llvm.fabs.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_ABS6:%.*]] = call <vscale x 2 x double> @llvm.fabs.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_abs(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svint8_t res_vi8 = __builtin_elementwise_abs(vi8);
+  svint16_t res_vi16 = __builtin_elementwise_abs(vi16);
+  svint32_t res_vi32 = __builtin_elementwise_abs(vi32);
+  svint64_t res_vi64 = __builtin_elementwise_abs(vi64);
+  svfloat16_t res_vf16 = __builtin_elementwise_abs(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_abs(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_abs(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_add_sat(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_SAT:%.*]] = call <vscale x 16 x i8> @llvm.sadd.sat.nxv16i8(<vscale x 16 x i8> [[VI8:%.*]], <vscale x 16 x i8> [[VI8]])
+// CHECK-NEXT:    [[ELT_SAT1:%.*]] = call <vscale x 8 x i16> @llvm.sadd.sat.nxv8i16(<vscale x 8 x i16> [[VI16:%.*]], <vscale x 8 x i16> [[VI16]])
+// CHECK-NEXT:    [[ELT_SAT2:%.*]] = call <vscale x 4 x i32> @llvm.sadd.sat.nxv4i32(<vscale x 4 x i32> [[VI32:%.*]], <vscale x 4 x i32> [[VI32]])
+// CHECK-NEXT:    [[ELT_SAT3:%.*]] = call <vscale x 2 x i64> @llvm.sadd.sat.nxv2i64(<vscale x 2 x i64> [[VI64:%.*]], <vscale x 2 x i64> [[VI64]])
+// CHECK-NEXT:    [[ELT_SAT4:%.*]] = call <vscale x 16 x i8> @llvm.uadd.sat.nxv16i8(<vscale x 16 x i8> [[VU8:%.*]], <vscale x 16 x i8> [[VU8]])
+// CHECK-NEXT:    [[ELT_SAT5:%.*]] = call <vscale x 8 x i16> @llvm.uadd.sat.nxv8i16(<vscale x 8 x i16> [[VU16:%.*]], <vscale x 8 x i16> [[VU16]])
+// CHECK-NEXT:    [[ELT_SAT6:%.*]] = call <vscale x 4 x i32> @llvm.uadd.sat.nxv4i32(<vscale x 4 x i32> [[VU32:%.*]], <vscale x 4 x i32> [[VU32]])
+// CHECK-NEXT:    [[ELT_SAT7:%.*]] = call <vscale x 2 x i64> @llvm.uadd.sat.nxv2i64(<vscale x 2 x i64> [[VU64:%.*]], <vscale x 2 x i64> [[VU64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_add_sat(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svint8_t res_vi8 = __builtin_elementwise_add_sat(vi8, vi8);
+  svint16_t res_vi16 = __builtin_elementwise_add_sat(vi16, vi16);
+  svint32_t res_vi32 = __builtin_elementwise_add_sat(vi32, vi32);
+  svint64_t res_vi64 = __builtin_elementwise_add_sat(vi64, vi64);
+  svuint8_t res_vu8 = __builtin_elementwise_add_sat(vu8, vu8);
+  svuint16_t res_vu16 = __builtin_elementwise_add_sat(vu16, vu16);
+  svuint32_t res_vu32 = __builtin_elementwise_add_sat(vu32, vu32);
+  svuint64_t res_vu64 = __builtin_elementwise_add_sat(vu64, vu64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_sub_sat(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_SAT:%.*]] = call <vscale x 16 x i8> @llvm.ssub.sat.nxv16i8(<vscale x 16 x i8> [[VI8:%.*]], <vscale x 16 x i8> [[VI8]])
+// CHECK-NEXT:    [[ELT_SAT1:%.*]] = call <vscale x 8 x i16> @llvm.ssub.sat.nxv8i16(<vscale x 8 x i16> [[VI16:%.*]], <vscale x 8 x i16> [[VI16]])
+// CHECK-NEXT:    [[ELT_SAT2:%.*]] = call <vscale x 4 x i32> @llvm.ssub.sat.nxv4i32(<vscale x 4 x i32> [[VI32:%.*]], <vscale x 4 x i32> [[VI32]])
+// CHECK-NEXT:    [[ELT_SAT3:%.*]] = call <vscale x 2 x i64> @llvm.ssub.sat.nxv2i64(<vscale x 2 x i64> [[VI64:%.*]], <vscale x 2 x i64> [[VI64]])
+// CHECK-NEXT:    [[ELT_SAT4:%.*]] = call <vscale x 16 x i8> @llvm.usub.sat.nxv16i8(<vscale x 16 x i8> [[VU8:%.*]], <vscale x 16 x i8> [[VU8]])
+// CHECK-NEXT:    [[ELT_SAT5:%.*]] = call <vscale x 8 x i16> @llvm.usub.sat.nxv8i16(<vscale x 8 x i16> [[VU16:%.*]], <vscale x 8 x i16> [[VU16]])
+// CHECK-NEXT:    [[ELT_SAT6:%.*]] = call <vscale x 4 x i32> @llvm.usub.sat.nxv4i32(<vscale x 4 x i32> [[VU32:%.*]], <vscale x 4 x i32> [[VU32]])
+// CHECK-NEXT:    [[ELT_SAT7:%.*]] = call <vscale x 2 x i64> @llvm.usub.sat.nxv2i64(<vscale x 2 x i64> [[VU64:%.*]], <vscale x 2 x i64> [[VU64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_sub_sat(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svint8_t res_vi8 = __builtin_elementwise_sub_sat(vi8, vi8);
+  svint16_t res_vi16 = __builtin_elementwise_sub_sat(vi16, vi16);
+  svint32_t res_vi32 = __builtin_elementwise_sub_sat(vi32, vi32);
+  svint64_t res_vi64 = __builtin_elementwise_sub_sat(vi64, vi64);
+  svuint8_t res_vu8 = __builtin_elementwise_sub_sat(vu8, vu8);
+  svuint16_t res_vu16 = __builtin_elementwise_sub_sat(vu16, vu16);
+  svuint32_t res_vu32 = __builtin_elementwise_sub_sat(vu32, vu32);
+  svuint64_t res_vu64 = __builtin_elementwise_sub_sat(vu64, vu64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_max(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_MAX:%.*]] = call <vscale x 16 x i8> @llvm.smax.nxv16i8(<vscale x 16 x i8> [[VI8:%.*]], <vscale x 16 x i8> [[VI8]])
+// CHECK-NEXT:    [[ELT_MAX1:%.*]] = call <vscale x 8 x i16> @llvm.smax.nxv8i16(<vscale x 8 x i16> [[VI16:%.*]], <vscale x 8 x i16> [[VI16]])
+// CHECK-NEXT:    [[ELT_MAX2:%.*]] = call <vscale x 4 x i32> @llvm.smax.nxv4i32(<vscale x 4 x i32> [[VI32:%.*]], <vscale x 4 x i32> [[VI32]])
+// CHECK-NEXT:    [[ELT_MAX3:%.*]] = call <vscale x 2 x i64> @llvm.smax.nxv2i64(<vscale x 2 x i64> [[VI64:%.*]], <vscale x 2 x i64> [[VI64]])
+// CHECK-NEXT:    [[ELT_MAX4:%.*]] = call <vscale x 16 x i8> @llvm.umax.nxv16i8(<vscale x 16 x i8> [[VU8:%.*]], <vscale x 16 x i8> [[VU8]])
+// CHECK-NEXT:    [[ELT_MAX5:%.*]] = call <vscale x 8 x i16> @llvm.umax.nxv8i16(<vscale x 8 x i16> [[VU16:%.*]], <vscale x 8 x i16> [[VU16]])
+// CHECK-NEXT:    [[ELT_MAX6:%.*]] = call <vscale x 4 x i32> @llvm.umax.nxv4i32(<vscale x 4 x i32> [[VU32:%.*]], <vscale x 4 x i32> [[VU32]])
+// CHECK-NEXT:    [[ELT_MAX7:%.*]] = call <vscale x 2 x i64> @llvm.umax.nxv2i64(<vscale x 2 x i64> [[VU64:%.*]], <vscale x 2 x i64> [[VU64]])
+// CHECK-NEXT:    [[ELT_MAX8:%.*]] = call <vscale x 8 x half> @llvm.maxnum.nxv8f16(<vscale x 8 x half> [[VF16:%.*]], <vscale x 8 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_MAX9:%.*]] = call <vscale x 4 x float> @llvm.maxnum.nxv4f32(<vscale x 4 x float> [[VF32:%.*]], <vscale x 4 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_MAX10:%.*]] = call <vscale x 2 x double> @llvm.maxnum.nxv2f64(<vscale x 2 x double> [[VF64:%.*]], <vscale x 2 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_max(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svint8_t res_vi8 = __builtin_elementwise_max(vi8, vi8);
+  svint16_t res_vi16 = __builtin_elementwise_max(vi16, vi16);
+  svint32_t res_vi32 = __builtin_elementwise_max(vi32, vi32);
+  svint64_t res_vi64 = __builtin_elementwise_max(vi64, vi64);
+  svuint8_t res_vu8 = __builtin_elementwise_max(vu8, vu8);
+  svuint16_t res_vu16 = __builtin_elementwise_max(vu16, vu16);
+  svuint32_t res_vu32 = __builtin_elementwise_max(vu32, vu32);
+  svuint64_t res_vu64 = __builtin_elementwise_max(vu64, vu64);
+  svfloat16_t res_vf16 = __builtin_elementwise_max(vf16, vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_max(vf32, vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_max(vf64, vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_min(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_MIN:%.*]] = call <vscale x 16 x i8> @llvm.umin.nxv16i8(<vscale x 16 x i8> [[VI8:%.*]], <vscale x 16 x i8> [[VI8]])
+// CHECK-NEXT:    [[ELT_MIN1:%.*]] = call <vscale x 8 x i16> @llvm.umin.nxv8i16(<vscale x 8 x i16> [[VI16:%.*]], <vscale x 8 x i16> [[VI16]])
+// CHECK-NEXT:    [[ELT_MIN2:%.*]] = call <vscale x 4 x i32> @llvm.umin.nxv4i32(<vscale x 4 x i32> [[VI32:%.*]], <vscale x 4 x i32> [[VI32]])
+// CHECK-NEXT:    [[ELT_MIN3:%.*]] = call <vscale x 2 x i64> @llvm.umin.nxv2i64(<vscale x 2 x i64> [[VI64:%.*]], <vscale x 2 x i64> [[VI64]])
+// CHECK-NEXT:    [[ELT_MIN4:%.*]] = call <vscale x 16 x i8> @llvm.umin.nxv16i8(<vscale x 16 x i8> [[VU8:%.*]], <vscale x 16 x i8> [[VU8]])
+// CHECK-NEXT:    [[ELT_MIN5:%.*]] = call <vscale x 8 x i16> @llvm.umin.nxv8i16(<vscale x 8 x i16> [[VU16:%.*]], <vscale x 8 x i16> [[VU16]])
+// CHECK-NEXT:    [[ELT_MIN6:%.*]] = call <vscale x 4 x i32> @llvm.umin.nxv4i32(<vscale x 4 x i32> [[VU32:%.*]], <vscale x 4 x i32> [[VU32]])
+// CHECK-NEXT:    [[ELT_MIN7:%.*]] = call <vscale x 2 x i64> @llvm.umin.nxv2i64(<vscale x 2 x i64> [[VU64:%.*]], <vscale x 2 x i64> [[VU64]])
+// CHECK-NEXT:    [[ELT_MIN8:%.*]] = call <vscale x 8 x half> @llvm.minnum.nxv8f16(<vscale x 8 x half> [[VF16:%.*]], <vscale x 8 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_MIN9:%.*]] = call <vscale x 4 x float> @llvm.minnum.nxv4f32(<vscale x 4 x float> [[VF32:%.*]], <vscale x 4 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_MIN10:%.*]] = call <vscale x 2 x double> @llvm.minnum.nxv2f64(<vscale x 2 x double> [[VF64:%.*]], <vscale x 2 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_min(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svint8_t res_vi8 = __builtin_elementwise_min(vi8, vi8);
+  svint16_t res_vi16 = __builtin_elementwise_min(vi16, vi16);
+  svint32_t res_vi32 = __builtin_elementwise_min(vi32, vi32);
+  svint64_t res_vi64 = __builtin_elementwise_min(vi64, vi64);
+  svuint8_t res_vu8 = __builtin_elementwise_min(vu8, vu8);
+  svuint16_t res_vu16 = __builtin_elementwise_min(vu16, vu16);
+  svuint32_t res_vu32 = __builtin_elementwise_min(vu32, vu32);
+  svuint64_t res_vu64 = __builtin_elementwise_min(vu64, vu64);
+  svfloat16_t res_vf16 = __builtin_elementwise_min(vf16, vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_min(vf32, vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_min(vf64, vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_bitreverse(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_BITREVERSE:%.*]] = call <vscale x 16 x i8> @llvm.bitreverse.nxv16i8(<vscale x 16 x i8> [[VI8:%.*]])
+// CHECK-NEXT:    [[ELT_BITREVERSE1:%.*]] = call <vscale x 8 x i16> @llvm.bitreverse.nxv8i16(<vscale x 8 x i16> [[VI16:%.*]])
+// CHECK-NEXT:    [[ELT_BITREVERSE2:%.*]] = call <vscale x 4 x i32> @llvm.bitreverse.nxv4i32(<vscale x 4 x i32> [[VI32:%.*]])
+// CHECK-NEXT:    [[ELT_BITREVERSE3:%.*]] = call <vscale x 2 x i64> @llvm.bitreverse.nxv2i64(<vscale x 2 x i64> [[VI64:%.*]])
+// CHECK-NEXT:    [[ELT_BITREVERSE4:%.*]] = call <vscale x 16 x i8> @llvm.bitreverse.nxv16i8(<vscale x 16 x i8> [[VU8:%.*]])
+// CHECK-NEXT:    [[ELT_BITREVERSE5:%.*]] = call <vscale x 8 x i16> @llvm.bitreverse.nxv8i16(<vscale x 8 x i16> [[VU16:%.*]])
+// CHECK-NEXT:    [[ELT_BITREVERSE6:%.*]] = call <vscale x 4 x i32> @llvm.bitreverse.nxv4i32(<vscale x 4 x i32> [[VU32:%.*]])
+// CHECK-NEXT:    [[ELT_BITREVERSE7:%.*]] = call <vscale x 2 x i64> @llvm.bitreverse.nxv2i64(<vscale x 2 x i64> [[VU64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_bitreverse(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svint8_t res_vi8 = __builtin_elementwise_bitreverse(vi8);
+  svint16_t res_vi16 = __builtin_elementwise_bitreverse(vi16);
+  svint32_t res_vi32 = __builtin_elementwise_bitreverse(vi32);
+  svint64_t res_vi64 = __builtin_elementwise_bitreverse(vi64);
+  svuint8_t res_vu8 = __builtin_elementwise_bitreverse(vu8);
+  svuint16_t res_vu16 = __builtin_elementwise_bitreverse(vu16);
+  svuint32_t res_vu32 = __builtin_elementwise_bitreverse(vu32);
+  svuint64_t res_vu64 = __builtin_elementwise_bitreverse(vu64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_ceil(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_CEIL:%.*]] = call <vscale x 8 x half> @llvm.ceil.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_CEIL1:%.*]] = call <vscale x 4 x float> @llvm.ceil.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_CEIL2:%.*]] = call <vscale x 2 x double> @llvm.ceil.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_ceil(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_ceil(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_ceil(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_ceil(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_cos(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_COS:%.*]] = call <vscale x 8 x half> @llvm.cos.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_COS1:%.*]] = call <vscale x 4 x float> @llvm.cos.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_COS2:%.*]] = call <vscale x 2 x double> @llvm.cos.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_cos(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_cos(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_cos(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_cos(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_exp(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_EXP:%.*]] = call <vscale x 8 x half> @llvm.exp.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_EXP1:%.*]] = call <vscale x 4 x float> @llvm.exp.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_EXP2:%.*]] = call <vscale x 2 x double> @llvm.exp.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_exp(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_exp(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_exp(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_exp(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_exp2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_EXP2:%.*]] = call <vscale x 8 x half> @llvm.exp2.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_EXP21:%.*]] = call <vscale x 4 x float> @llvm.exp2.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_EXP22:%.*]] = call <vscale x 2 x double> @llvm.exp2.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_exp2(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_exp2(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_exp2(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_exp2(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_floor(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_FLOOR:%.*]] = call <vscale x 8 x half> @llvm.floor.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_FLOOR1:%.*]] = call <vscale x 4 x float> @llvm.floor.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_FLOOR2:%.*]] = call <vscale x 2 x double> @llvm.floor.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_floor(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_floor(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_floor(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_floor(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_log(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_LOG:%.*]] = call <vscale x 8 x half> @llvm.log.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_LOG1:%.*]] = call <vscale x 4 x float> @llvm.log.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_LOG2:%.*]] = call <vscale x 2 x double> @llvm.log.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_log(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_log(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_log(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_log(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_log10(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_LOG10:%.*]] = call <vscale x 8 x half> @llvm.log10.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_LOG101:%.*]] = call <vscale x 4 x float> @llvm.log10.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_LOG102:%.*]] = call <vscale x 2 x double> @llvm.log10.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_log10(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_log10(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_log10(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_log10(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_log2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_LOG2:%.*]] = call <vscale x 8 x half> @llvm.log2.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_LOG21:%.*]] = call <vscale x 4 x float> @llvm.log2.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_LOG22:%.*]] = call <vscale x 2 x double> @llvm.log2.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_log2(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_log2(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_log2(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_log2(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_pow(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.pow.nxv8f16(<vscale x 8 x half> [[VF16:%.*]], <vscale x 8 x half> [[VF16]])
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.pow.nxv4f32(<vscale x 4 x float> [[VF32:%.*]], <vscale x 4 x float> [[VF32]])
+// CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.pow.nxv2f64(<vscale x 2 x double> [[VF64:%.*]], <vscale x 2 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_pow(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_pow(vf16, vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_pow(vf32, vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_pow(vf64, vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_roundeven(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_ROUNDEVEN:%.*]] = call <vscale x 8 x half> @llvm.roundeven.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_ROUNDEVEN1:%.*]] = call <vscale x 4 x float> @llvm.roundeven.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_ROUNDEVEN2:%.*]] = call <vscale x 2 x double> @llvm.roundeven.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_roundeven(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_roundeven(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_roundeven(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_roundeven(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_round(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_ROUND:%.*]] = call <vscale x 8 x half> @llvm.round.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_ROUND1:%.*]] = call <vscale x 4 x float> @llvm.round.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_ROUND2:%.*]] = call <vscale x 2 x double> @llvm.round.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_round(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_round(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_round(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_round(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_rint(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_RINT:%.*]] = call <vscale x 8 x half> @llvm.rint.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_RINT1:%.*]] = call <vscale x 4 x float> @llvm.rint.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_RINT2:%.*]] = call <vscale x 2 x double> @llvm.rint.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_rint(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_rint(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_rint(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_rint(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_nearbyint(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_NEARBYINT:%.*]] = call <vscale x 8 x half> @llvm.nearbyint.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_NEARBYINT1:%.*]] = call <vscale x 4 x float> @llvm.nearbyint.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_NEARBYINT2:%.*]] = call <vscale x 2 x double> @llvm.nearbyint.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_nearbyint(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_nearbyint(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_nearbyint(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_nearbyint(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_sin(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_SIN:%.*]] = call <vscale x 8 x half> @llvm.sin.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_SIN1:%.*]] = call <vscale x 4 x float> @llvm.sin.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_SIN2:%.*]] = call <vscale x 2 x double> @llvm.sin.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_sin(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_sin(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_sin(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_sin(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_sqrt(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.sqrt.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.sqrt.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.sqrt.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_sqrt(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_sqrt(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_sqrt(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_sqrt(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_trunc(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_TRUNC:%.*]] = call <vscale x 8 x half> @llvm.trunc.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_TRUNC1:%.*]] = call <vscale x 4 x float> @llvm.trunc.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_TRUNC2:%.*]] = call <vscale x 2 x double> @llvm.trunc.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_trunc(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_trunc(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_trunc(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_trunc(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_canonicalize(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_CANONICALIZE:%.*]] = call <vscale x 8 x half> @llvm.canonicalize.nxv8f16(<vscale x 8 x half> [[VF16:%.*]])
+// CHECK-NEXT:    [[ELT_CANONICALIZE1:%.*]] = call <vscale x 4 x float> @llvm.canonicalize.nxv4f32(<vscale x 4 x float> [[VF32:%.*]])
+// CHECK-NEXT:    [[ELT_CANONICALIZE2:%.*]] = call <vscale x 2 x double> @llvm.canonicalize.nxv2f64(<vscale x 2 x double> [[VF64:%.*]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_canonicalize(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_canonicalize(vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_canonicalize(vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_canonicalize(vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_copysign(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.copysign.nxv8f16(<vscale x 8 x half> [[VF16:%.*]], <vscale x 8 x half> [[VF16]])
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.copysign.nxv4f32(<vscale x 4 x float> [[VF32:%.*]], <vscale x 4 x float> [[VF32]])
+// CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.copysign.nxv2f64(<vscale x 2 x double> [[VF64:%.*]], <vscale x 2 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_copysign(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_copysign(vf16, vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_copysign(vf32, vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_copysign(vf64, vf64);
+}
+
+// CHECK-LABEL: @test_builtin_elementwise_fma(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.fma.nxv8f16(<vscale x 8 x half> [[VF16:%.*]], <vscale x 8 x half> [[VF16]], <vscale x 8 x half> [[VF16]])
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.fma.nxv4f32(<vscale x 4 x float> [[VF32:%.*]], <vscale x 4 x float> [[VF32]], <vscale x 4 x float> [[VF32]])
+// CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.fma.nxv2f64(<vscale x 2 x double> [[VF64:%.*]], <vscale x 2 x double> [[VF64]], <vscale x 2 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_fma(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_fma(vf16, vf16, vf16);
+  svfloat32_t res_vf32 = __builtin_elementwise_fma(vf32, vf32, vf32);
+  svfloat64_t res_vf64 = __builtin_elementwise_fma(vf64, vf64, vf64);
+}
diff --git a/clang/test/CodeGen/riscv-rvv-vector-elementwise-ops.c b/clang/test/CodeGen/riscv-rvv-vector-elementwise-ops.c
new file mode 100644
--- /dev/null
+++ b/clang/test/CodeGen/riscv-rvv-vector-elementwise-ops.c
@@ -0,0 +1,553 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 2
+// RUN: %clang_cc1 -triple riscv64-none-linux-gnu -target-feature +zve64d \
+// RUN: -target-feature +f -target-feature +d -target-feature +zvfh -target-feature +zfh -disable-O0-optnone \
+// RUN: -mvscale-min=4 -mvscale-max=4 -emit-llvm -o - %s | \
+// RUN: opt -S -passes=sroa | FileCheck %s
+// REQUIRES: riscv-registered-target
+
+#include <riscv_vector.h>
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_abs
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0:[0-9]+]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_ABS:%.*]] = call <vscale x 8 x i8> @llvm.abs.nxv8i8(<vscale x 8 x i8> [[VI8]], i1 false)
+// CHECK-NEXT:    [[ELT_ABS1:%.*]] = call <vscale x 4 x i16> @llvm.abs.nxv4i16(<vscale x 4 x i16> [[VI16]], i1 false)
+// CHECK-NEXT:    [[ELT_ABS2:%.*]] = call <vscale x 2 x i32> @llvm.abs.nxv2i32(<vscale x 2 x i32> [[VI32]], i1 false)
+// CHECK-NEXT:    [[ELT_ABS3:%.*]] = call <vscale x 1 x i64> @llvm.abs.nxv1i64(<vscale x 1 x i64> [[VI64]], i1 false)
+// CHECK-NEXT:    [[ELT_ABS4:%.*]] = call <vscale x 4 x half> @llvm.fabs.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_ABS5:%.*]] = call <vscale x 2 x float> @llvm.fabs.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_ABS6:%.*]] = call <vscale x 1 x double> @llvm.fabs.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_abs(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vint8m1_t res_vi8 = __builtin_elementwise_abs(vi8);
+  vint16m1_t res_vi16 = __builtin_elementwise_abs(vi16);
+  vint32m1_t res_vi32 = __builtin_elementwise_abs(vi32);
+  vint64m1_t res_vi64 = __builtin_elementwise_abs(vi64);
+  vfloat16m1_t res_vf16 = __builtin_elementwise_abs(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_abs(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_abs(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_add_sat
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_SAT:%.*]] = call <vscale x 8 x i8> @llvm.sadd.sat.nxv8i8(<vscale x 8 x i8> [[VI8]], <vscale x 8 x i8> [[VI8]])
+// CHECK-NEXT:    [[ELT_SAT1:%.*]] = call <vscale x 4 x i16> @llvm.sadd.sat.nxv4i16(<vscale x 4 x i16> [[VI16]], <vscale x 4 x i16> [[VI16]])
+// CHECK-NEXT:    [[ELT_SAT2:%.*]] = call <vscale x 2 x i32> @llvm.sadd.sat.nxv2i32(<vscale x 2 x i32> [[VI32]], <vscale x 2 x i32> [[VI32]])
+// CHECK-NEXT:    [[ELT_SAT3:%.*]] = call <vscale x 1 x i64> @llvm.sadd.sat.nxv1i64(<vscale x 1 x i64> [[VI64]], <vscale x 1 x i64> [[VI64]])
+// CHECK-NEXT:    [[ELT_SAT4:%.*]] = call <vscale x 8 x i8> @llvm.uadd.sat.nxv8i8(<vscale x 8 x i8> [[VU8]], <vscale x 8 x i8> [[VU8]])
+// CHECK-NEXT:    [[ELT_SAT5:%.*]] = call <vscale x 4 x i16> @llvm.uadd.sat.nxv4i16(<vscale x 4 x i16> [[VU16]], <vscale x 4 x i16> [[VU16]])
+// CHECK-NEXT:    [[ELT_SAT6:%.*]] = call <vscale x 2 x i32> @llvm.uadd.sat.nxv2i32(<vscale x 2 x i32> [[VU32]], <vscale x 2 x i32> [[VU32]])
+// CHECK-NEXT:    [[ELT_SAT7:%.*]] = call <vscale x 1 x i64> @llvm.uadd.sat.nxv1i64(<vscale x 1 x i64> [[VU64]], <vscale x 1 x i64> [[VU64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_add_sat(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vint8m1_t res_vi8 = __builtin_elementwise_add_sat(vi8, vi8);
+  vint16m1_t res_vi16 = __builtin_elementwise_add_sat(vi16, vi16);
+  vint32m1_t res_vi32 = __builtin_elementwise_add_sat(vi32, vi32);
+  vint64m1_t res_vi64 = __builtin_elementwise_add_sat(vi64, vi64);
+  vuint8m1_t res_vu8 = __builtin_elementwise_add_sat(vu8, vu8);
+  vuint16m1_t res_vu16 = __builtin_elementwise_add_sat(vu16, vu16);
+  vuint32m1_t res_vu32 = __builtin_elementwise_add_sat(vu32, vu32);
+  vuint64m1_t res_vu64 = __builtin_elementwise_add_sat(vu64, vu64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_sub_sat
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_SAT:%.*]] = call <vscale x 8 x i8> @llvm.ssub.sat.nxv8i8(<vscale x 8 x i8> [[VI8]], <vscale x 8 x i8> [[VI8]])
+// CHECK-NEXT:    [[ELT_SAT1:%.*]] = call <vscale x 4 x i16> @llvm.ssub.sat.nxv4i16(<vscale x 4 x i16> [[VI16]], <vscale x 4 x i16> [[VI16]])
+// CHECK-NEXT:    [[ELT_SAT2:%.*]] = call <vscale x 2 x i32> @llvm.ssub.sat.nxv2i32(<vscale x 2 x i32> [[VI32]], <vscale x 2 x i32> [[VI32]])
+// CHECK-NEXT:    [[ELT_SAT3:%.*]] = call <vscale x 1 x i64> @llvm.ssub.sat.nxv1i64(<vscale x 1 x i64> [[VI64]], <vscale x 1 x i64> [[VI64]])
+// CHECK-NEXT:    [[ELT_SAT4:%.*]] = call <vscale x 8 x i8> @llvm.usub.sat.nxv8i8(<vscale x 8 x i8> [[VU8]], <vscale x 8 x i8> [[VU8]])
+// CHECK-NEXT:    [[ELT_SAT5:%.*]] = call <vscale x 4 x i16> @llvm.usub.sat.nxv4i16(<vscale x 4 x i16> [[VU16]], <vscale x 4 x i16> [[VU16]])
+// CHECK-NEXT:    [[ELT_SAT6:%.*]] = call <vscale x 2 x i32> @llvm.usub.sat.nxv2i32(<vscale x 2 x i32> [[VU32]], <vscale x 2 x i32> [[VU32]])
+// CHECK-NEXT:    [[ELT_SAT7:%.*]] = call <vscale x 1 x i64> @llvm.usub.sat.nxv1i64(<vscale x 1 x i64> [[VU64]], <vscale x 1 x i64> [[VU64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_sub_sat(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vint8m1_t res_vi8 = __builtin_elementwise_sub_sat(vi8, vi8);
+  vint16m1_t res_vi16 = __builtin_elementwise_sub_sat(vi16, vi16);
+  vint32m1_t res_vi32 = __builtin_elementwise_sub_sat(vi32, vi32);
+  vint64m1_t res_vi64 = __builtin_elementwise_sub_sat(vi64, vi64);
+  vuint8m1_t res_vu8 = __builtin_elementwise_sub_sat(vu8, vu8);
+  vuint16m1_t res_vu16 = __builtin_elementwise_sub_sat(vu16, vu16);
+  vuint32m1_t res_vu32 = __builtin_elementwise_sub_sat(vu32, vu32);
+  vuint64m1_t res_vu64 = __builtin_elementwise_sub_sat(vu64, vu64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_max
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_MAX:%.*]] = call <vscale x 8 x i8> @llvm.smax.nxv8i8(<vscale x 8 x i8> [[VI8]], <vscale x 8 x i8> [[VI8]])
+// CHECK-NEXT:    [[ELT_MAX1:%.*]] = call <vscale x 4 x i16> @llvm.smax.nxv4i16(<vscale x 4 x i16> [[VI16]], <vscale x 4 x i16> [[VI16]])
+// CHECK-NEXT:    [[ELT_MAX2:%.*]] = call <vscale x 2 x i32> @llvm.smax.nxv2i32(<vscale x 2 x i32> [[VI32]], <vscale x 2 x i32> [[VI32]])
+// CHECK-NEXT:    [[ELT_MAX3:%.*]] = call <vscale x 1 x i64> @llvm.smax.nxv1i64(<vscale x 1 x i64> [[VI64]], <vscale x 1 x i64> [[VI64]])
+// CHECK-NEXT:    [[ELT_MAX4:%.*]] = call <vscale x 8 x i8> @llvm.umax.nxv8i8(<vscale x 8 x i8> [[VU8]], <vscale x 8 x i8> [[VU8]])
+// CHECK-NEXT:    [[ELT_MAX5:%.*]] = call <vscale x 4 x i16> @llvm.umax.nxv4i16(<vscale x 4 x i16> [[VU16]], <vscale x 4 x i16> [[VU16]])
+// CHECK-NEXT:    [[ELT_MAX6:%.*]] = call <vscale x 2 x i32> @llvm.umax.nxv2i32(<vscale x 2 x i32> [[VU32]], <vscale x 2 x i32> [[VU32]])
+// CHECK-NEXT:    [[ELT_MAX7:%.*]] = call <vscale x 1 x i64> @llvm.umax.nxv1i64(<vscale x 1 x i64> [[VU64]], <vscale x 1 x i64> [[VU64]])
+// CHECK-NEXT:    [[ELT_MAX8:%.*]] = call <vscale x 4 x half> @llvm.maxnum.nxv4f16(<vscale x 4 x half> [[VF16]], <vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_MAX9:%.*]] = call <vscale x 2 x float> @llvm.maxnum.nxv2f32(<vscale x 2 x float> [[VF32]], <vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_MAX10:%.*]] = call <vscale x 1 x double> @llvm.maxnum.nxv1f64(<vscale x 1 x double> [[VF64]], <vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_max(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vint8m1_t res_vi8 = __builtin_elementwise_max(vi8, vi8);
+  vint16m1_t res_vi16 = __builtin_elementwise_max(vi16, vi16);
+  vint32m1_t res_vi32 = __builtin_elementwise_max(vi32, vi32);
+  vint64m1_t res_vi64 = __builtin_elementwise_max(vi64, vi64);
+  vuint8m1_t res_vu8 = __builtin_elementwise_max(vu8, vu8);
+  vuint16m1_t res_vu16 = __builtin_elementwise_max(vu16, vu16);
+  vuint32m1_t res_vu32 = __builtin_elementwise_max(vu32, vu32);
+  vuint64m1_t res_vu64 = __builtin_elementwise_max(vu64, vu64);
+  vfloat16m1_t res_vf16 = __builtin_elementwise_max(vf16, vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_max(vf32, vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_max(vf64, vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_min
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_MIN:%.*]] = call <vscale x 8 x i8> @llvm.umin.nxv8i8(<vscale x 8 x i8> [[VI8]], <vscale x 8 x i8> [[VI8]])
+// CHECK-NEXT:    [[ELT_MIN1:%.*]] = call <vscale x 4 x i16> @llvm.umin.nxv4i16(<vscale x 4 x i16> [[VI16]], <vscale x 4 x i16> [[VI16]])
+// CHECK-NEXT:    [[ELT_MIN2:%.*]] = call <vscale x 2 x i32> @llvm.umin.nxv2i32(<vscale x 2 x i32> [[VI32]], <vscale x 2 x i32> [[VI32]])
+// CHECK-NEXT:    [[ELT_MIN3:%.*]] = call <vscale x 1 x i64> @llvm.umin.nxv1i64(<vscale x 1 x i64> [[VI64]], <vscale x 1 x i64> [[VI64]])
+// CHECK-NEXT:    [[ELT_MIN4:%.*]] = call <vscale x 8 x i8> @llvm.umin.nxv8i8(<vscale x 8 x i8> [[VU8]], <vscale x 8 x i8> [[VU8]])
+// CHECK-NEXT:    [[ELT_MIN5:%.*]] = call <vscale x 4 x i16> @llvm.umin.nxv4i16(<vscale x 4 x i16> [[VU16]], <vscale x 4 x i16> [[VU16]])
+// CHECK-NEXT:    [[ELT_MIN6:%.*]] = call <vscale x 2 x i32> @llvm.umin.nxv2i32(<vscale x 2 x i32> [[VU32]], <vscale x 2 x i32> [[VU32]])
+// CHECK-NEXT:    [[ELT_MIN7:%.*]] = call <vscale x 1 x i64> @llvm.umin.nxv1i64(<vscale x 1 x i64> [[VU64]], <vscale x 1 x i64> [[VU64]])
+// CHECK-NEXT:    [[ELT_MIN8:%.*]] = call <vscale x 4 x half> @llvm.minnum.nxv4f16(<vscale x 4 x half> [[VF16]], <vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_MIN9:%.*]] = call <vscale x 2 x float> @llvm.minnum.nxv2f32(<vscale x 2 x float> [[VF32]], <vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_MIN10:%.*]] = call <vscale x 1 x double> @llvm.minnum.nxv1f64(<vscale x 1 x double> [[VF64]], <vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_min(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vint8m1_t res_vi8 = __builtin_elementwise_min(vi8, vi8);
+  vint16m1_t res_vi16 = __builtin_elementwise_min(vi16, vi16);
+  vint32m1_t res_vi32 = __builtin_elementwise_min(vi32, vi32);
+  vint64m1_t res_vi64 = __builtin_elementwise_min(vi64, vi64);
+  vuint8m1_t res_vu8 = __builtin_elementwise_min(vu8, vu8);
+  vuint16m1_t res_vu16 = __builtin_elementwise_min(vu16, vu16);
+  vuint32m1_t res_vu32 = __builtin_elementwise_min(vu32, vu32);
+  vuint64m1_t res_vu64 = __builtin_elementwise_min(vu64, vu64);
+  vfloat16m1_t res_vf16 = __builtin_elementwise_min(vf16, vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_min(vf32, vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_min(vf64, vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_bitreverse
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_BITREVERSE:%.*]] = call <vscale x 8 x i8> @llvm.bitreverse.nxv8i8(<vscale x 8 x i8> [[VI8]])
+// CHECK-NEXT:    [[ELT_BITREVERSE1:%.*]] = call <vscale x 4 x i16> @llvm.bitreverse.nxv4i16(<vscale x 4 x i16> [[VI16]])
+// CHECK-NEXT:    [[ELT_BITREVERSE2:%.*]] = call <vscale x 2 x i32> @llvm.bitreverse.nxv2i32(<vscale x 2 x i32> [[VI32]])
+// CHECK-NEXT:    [[ELT_BITREVERSE3:%.*]] = call <vscale x 1 x i64> @llvm.bitreverse.nxv1i64(<vscale x 1 x i64> [[VI64]])
+// CHECK-NEXT:    [[ELT_BITREVERSE4:%.*]] = call <vscale x 8 x i8> @llvm.bitreverse.nxv8i8(<vscale x 8 x i8> [[VU8]])
+// CHECK-NEXT:    [[ELT_BITREVERSE5:%.*]] = call <vscale x 4 x i16> @llvm.bitreverse.nxv4i16(<vscale x 4 x i16> [[VU16]])
+// CHECK-NEXT:    [[ELT_BITREVERSE6:%.*]] = call <vscale x 2 x i32> @llvm.bitreverse.nxv2i32(<vscale x 2 x i32> [[VU32]])
+// CHECK-NEXT:    [[ELT_BITREVERSE7:%.*]] = call <vscale x 1 x i64> @llvm.bitreverse.nxv1i64(<vscale x 1 x i64> [[VU64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_bitreverse(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vint8m1_t res_vi8 = __builtin_elementwise_bitreverse(vi8);
+  vint16m1_t res_vi16 = __builtin_elementwise_bitreverse(vi16);
+  vint32m1_t res_vi32 = __builtin_elementwise_bitreverse(vi32);
+  vint64m1_t res_vi64 = __builtin_elementwise_bitreverse(vi64);
+  vuint8m1_t res_vu8 = __builtin_elementwise_bitreverse(vu8);
+  vuint16m1_t res_vu16 = __builtin_elementwise_bitreverse(vu16);
+  vuint32m1_t res_vu32 = __builtin_elementwise_bitreverse(vu32);
+  vuint64m1_t res_vu64 = __builtin_elementwise_bitreverse(vu64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_ceil
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_CEIL:%.*]] = call <vscale x 4 x half> @llvm.ceil.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_CEIL1:%.*]] = call <vscale x 2 x float> @llvm.ceil.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_CEIL2:%.*]] = call <vscale x 1 x double> @llvm.ceil.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_ceil(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_ceil(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_ceil(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_ceil(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_cos
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_COS:%.*]] = call <vscale x 4 x half> @llvm.cos.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_COS1:%.*]] = call <vscale x 2 x float> @llvm.cos.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_COS2:%.*]] = call <vscale x 1 x double> @llvm.cos.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_cos(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_cos(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_cos(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_cos(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_exp
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_EXP:%.*]] = call <vscale x 4 x half> @llvm.exp.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_EXP1:%.*]] = call <vscale x 2 x float> @llvm.exp.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_EXP2:%.*]] = call <vscale x 1 x double> @llvm.exp.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_exp(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_exp(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_exp(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_exp(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_exp2
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_EXP2:%.*]] = call <vscale x 4 x half> @llvm.exp2.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_EXP21:%.*]] = call <vscale x 2 x float> @llvm.exp2.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_EXP22:%.*]] = call <vscale x 1 x double> @llvm.exp2.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_exp2(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_exp2(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_exp2(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_exp2(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_floor
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_FLOOR:%.*]] = call <vscale x 4 x half> @llvm.floor.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_FLOOR1:%.*]] = call <vscale x 2 x float> @llvm.floor.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_FLOOR2:%.*]] = call <vscale x 1 x double> @llvm.floor.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_floor(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_floor(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_floor(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_floor(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_log
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_LOG:%.*]] = call <vscale x 4 x half> @llvm.log.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_LOG1:%.*]] = call <vscale x 2 x float> @llvm.log.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_LOG2:%.*]] = call <vscale x 1 x double> @llvm.log.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_log(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_log(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_log(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_log(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_log10
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_LOG10:%.*]] = call <vscale x 4 x half> @llvm.log10.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_LOG101:%.*]] = call <vscale x 2 x float> @llvm.log10.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_LOG102:%.*]] = call <vscale x 1 x double> @llvm.log10.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_log10(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_log10(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_log10(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_log10(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_log2
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_LOG2:%.*]] = call <vscale x 4 x half> @llvm.log2.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_LOG21:%.*]] = call <vscale x 2 x float> @llvm.log2.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_LOG22:%.*]] = call <vscale x 1 x double> @llvm.log2.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_log2(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_log2(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_log2(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_log2(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_pow
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.pow.nxv4f16(<vscale x 4 x half> [[VF16]], <vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x float> @llvm.pow.nxv2f32(<vscale x 2 x float> [[VF32]], <vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 1 x double> @llvm.pow.nxv1f64(<vscale x 1 x double> [[VF64]], <vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_pow(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_pow(vf16, vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_pow(vf32, vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_pow(vf64, vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_roundeven
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_ROUNDEVEN:%.*]] = call <vscale x 4 x half> @llvm.roundeven.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_ROUNDEVEN1:%.*]] = call <vscale x 2 x float> @llvm.roundeven.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_ROUNDEVEN2:%.*]] = call <vscale x 1 x double> @llvm.roundeven.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_roundeven(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_roundeven(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_roundeven(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_roundeven(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_round
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_ROUND:%.*]] = call <vscale x 4 x half> @llvm.round.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_ROUND1:%.*]] = call <vscale x 2 x float> @llvm.round.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_ROUND2:%.*]] = call <vscale x 1 x double> @llvm.round.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_round(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_round(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_round(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_round(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_rint
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_RINT:%.*]] = call <vscale x 4 x half> @llvm.rint.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_RINT1:%.*]] = call <vscale x 2 x float> @llvm.rint.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_RINT2:%.*]] = call <vscale x 1 x double> @llvm.rint.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_rint(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_rint(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_rint(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_rint(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_nearbyint
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_NEARBYINT:%.*]] = call <vscale x 4 x half> @llvm.nearbyint.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_NEARBYINT1:%.*]] = call <vscale x 2 x float> @llvm.nearbyint.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_NEARBYINT2:%.*]] = call <vscale x 1 x double> @llvm.nearbyint.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_nearbyint(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_nearbyint(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_nearbyint(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_nearbyint(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_sin
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_SIN:%.*]] = call <vscale x 4 x half> @llvm.sin.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_SIN1:%.*]] = call <vscale x 2 x float> @llvm.sin.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_SIN2:%.*]] = call <vscale x 1 x double> @llvm.sin.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_sin(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_sin(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_sin(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_sin(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_sqrt
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.sqrt.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x float> @llvm.sqrt.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 1 x double> @llvm.sqrt.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_sqrt(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_sqrt(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_sqrt(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_sqrt(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_trunc
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_TRUNC:%.*]] = call <vscale x 4 x half> @llvm.trunc.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_TRUNC1:%.*]] = call <vscale x 2 x float> @llvm.trunc.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_TRUNC2:%.*]] = call <vscale x 1 x double> @llvm.trunc.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_trunc(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_trunc(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_trunc(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_trunc(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_canonicalize
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[ELT_CANONICALIZE:%.*]] = call <vscale x 4 x half> @llvm.canonicalize.nxv4f16(<vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[ELT_CANONICALIZE1:%.*]] = call <vscale x 2 x float> @llvm.canonicalize.nxv2f32(<vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[ELT_CANONICALIZE2:%.*]] = call <vscale x 1 x double> @llvm.canonicalize.nxv1f64(<vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_canonicalize(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_canonicalize(vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_canonicalize(vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_canonicalize(vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_copysign
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.copysign.nxv4f16(<vscale x 4 x half> [[VF16]], <vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x float> @llvm.copysign.nxv2f32(<vscale x 2 x float> [[VF32]], <vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 1 x double> @llvm.copysign.nxv1f64(<vscale x 1 x double> [[VF64]], <vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_copysign(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_copysign(vf16, vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_copysign(vf32, vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_copysign(vf64, vf64);
+}
+
+// CHECK-LABEL: define dso_local void @test_builtin_elementwise_fma
+// CHECK-SAME: (<vscale x 8 x i8> [[VI8:%.*]], <vscale x 4 x i16> [[VI16:%.*]], <vscale x 2 x i32> [[VI32:%.*]], <vscale x 1 x i64> [[VI64:%.*]], <vscale x 8 x i8> [[VU8:%.*]], <vscale x 4 x i16> [[VU16:%.*]], <vscale x 2 x i32> [[VU32:%.*]], <vscale x 1 x i64> [[VU64:%.*]], <vscale x 4 x half> [[VF16:%.*]], <vscale x 2 x float> [[VF32:%.*]], <vscale x 1 x double> [[VF64:%.*]]) #[[ATTR0]] {
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x half> @llvm.fma.nxv4f16(<vscale x 4 x half> [[VF16]], <vscale x 4 x half> [[VF16]], <vscale x 4 x half> [[VF16]])
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x float> @llvm.fma.nxv2f32(<vscale x 2 x float> [[VF32]], <vscale x 2 x float> [[VF32]], <vscale x 2 x float> [[VF32]])
+// CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 1 x double> @llvm.fma.nxv1f64(<vscale x 1 x double> [[VF64]], <vscale x 1 x double> [[VF64]], <vscale x 1 x double> [[VF64]])
+// CHECK-NEXT:    ret void
+//
+void test_builtin_elementwise_fma(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_fma(vf16, vf16, vf16);
+  vfloat32m1_t res_vf32 = __builtin_elementwise_fma(vf32, vf32, vf32);
+  vfloat64m1_t res_vf64 = __builtin_elementwise_fma(vf64, vf64, vf64);
+}
diff --git a/clang/test/Sema/aarch64-sve-vector-elementwise.c b/clang/test/Sema/aarch64-sve-vector-elementwise.c
new file mode 100644
--- /dev/null
+++ b/clang/test/Sema/aarch64-sve-vector-elementwise.c
@@ -0,0 +1,519 @@
+// RUN: %clang_cc1 -verify -triple aarch64-none-linux-gnu -target-feature +sve -fsyntax-only %s
+
+// REQUIRES: aarch64-registered-target
+
+#include <arm_sve.h>
+
+void test_builtin_elementwise_abs(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_abs(vu8);
+  // expected-error@-1 {{1st argument must be a signed integer or floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_abs(vu16);
+  // expected-error@-1 {{1st argument must be a signed integer or floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_abs(vu32);
+  // expected-error@-1 {{1st argument must be a signed integer or floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_abs(vu64);
+  // expected-error@-1 {{1st argument must be a signed integer or floating point type}}
+}
+
+void test_builtin_elementwise_add_sat(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_add_sat(vf16, vf16);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+  svfloat32_t res_vf32 = __builtin_elementwise_add_sat(vf32, vf32);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+  svfloat64_t res_vf64 = __builtin_elementwise_add_sat(vf64, vf64);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+}
+
+void test_builtin_elementwise_sub_sat(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_sub_sat(vf16, vf16);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+  svfloat32_t res_vf32 = __builtin_elementwise_sub_sat(vf32, vf32);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+  svfloat64_t res_vf64 = __builtin_elementwise_sub_sat(vf64, vf64);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+}
+
+void test_builtin_elementwise_bitreverse(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svfloat16_t res_vf16 = __builtin_elementwise_bitreverse(vf16);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+  svfloat32_t res_vf32 = __builtin_elementwise_bitreverse(vf32);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+  svfloat64_t res_vf64 = __builtin_elementwise_bitreverse(vf64);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+}
+
+void test_builtin_elementwise_ceil(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_ceil(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_ceil(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_ceil(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_ceil(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_ceil(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_ceil(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_ceil(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_ceil(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_cos(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_cos(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_cos(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_cos(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_cos(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_cos(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_cos(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_cos(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_cos(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_exp(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_exp(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_exp(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_exp(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_exp(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_exp(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_exp(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_exp(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_exp(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_exp2(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_exp2(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_exp2(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_exp2(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_exp2(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_exp2(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_exp2(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_exp2(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_exp2(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_floor(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_floor(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_floor(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_floor(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_floor(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_floor(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_floor(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_floor(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_floor(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_log(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_log(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_log(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_log(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_log(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_log(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_log(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_log(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_log(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_log10(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_log10(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_log10(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_log10(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_log10(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_log10(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_log10(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_log10(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_log10(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_log2(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_log2(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_log2(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_log2(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_log2(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_log2(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_log2(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_log2(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_log2(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_pow(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_pow(vu8, vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_pow(vu16, vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_pow(vu32, vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_pow(vu64, vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_pow(vi8, vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_pow(vi16, vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_pow(vi32, vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_pow(vi64, vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_roundeven(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_roundeven(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_roundeven(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_roundeven(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_roundeven(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_roundeven(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_roundeven(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_roundeven(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_roundeven(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_round(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_round(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_round(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_round(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_round(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_round(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_round(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_round(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_round(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_rint(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_rint(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_rint(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_rint(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_rint(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_rint(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_rint(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_rint(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_rint(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_nearbyint(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_nearbyint(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_nearbyint(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_nearbyint(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_nearbyint(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_nearbyint(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_nearbyint(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_nearbyint(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_nearbyint(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_sin(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_sin(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_sin(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_sin(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_sin(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_sin(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_sin(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_sin(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_sin(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_sqrt(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_sqrt(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_sqrt(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_sqrt(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_sqrt(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_sqrt(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_sqrt(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_sqrt(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_sqrt(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_trunc(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_trunc(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_trunc(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_trunc(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_trunc(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_trunc(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_trunc(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_trunc(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_trunc(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_canonicalize(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_canonicalize(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_canonicalize(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_canonicalize(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_canonicalize(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_canonicalize(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_canonicalize(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_canonicalize(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_canonicalize(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_copysign(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_copysign(vu8, vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_copysign(vu16, vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_copysign(vu32, vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_copysign(vu64, vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_copysign(vi8, vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_copysign(vi16, vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_copysign(vi32, vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_copysign(vi64, vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_fma(svint8_t vi8, svint16_t vi16,
+                                  svint32_t vi32, svint64_t vi64,
+                                  svuint8_t vu8, svuint16_t vu16,
+                                  svuint32_t vu32, svuint64_t vu64,
+                                  svfloat16_t vf16, svfloat32_t vf32,
+                                  svfloat64_t vf64) {
+  svuint8_t res_vu8 = __builtin_elementwise_fma(vu8, vu8, vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vu16 = __builtin_elementwise_fma(vu16, vu16, vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vu32 = __builtin_elementwise_fma(vu32, vu32, vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vu64 = __builtin_elementwise_fma(vu64, vu64, vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint8_t res_vi8 = __builtin_elementwise_fma(vi8, vi8, vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint16_t res_vi16 = __builtin_elementwise_fma(vi16, vi16, vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint32_t res_vi32 = __builtin_elementwise_fma(vi32, vi32, vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  svuint64_t res_vi64 = __builtin_elementwise_fma(vi64, vi64, vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
diff --git a/clang/test/Sema/aarch64-sve-vector-exp-ops.c b/clang/test/Sema/aarch64-sve-vector-exp-ops.c
deleted file mode 100644
--- a/clang/test/Sema/aarch64-sve-vector-exp-ops.c
+++ /dev/null
@@ -1,17 +0,0 @@
-// RUN: %clang_cc1 -triple aarch64 -target-feature +sve \
-// RUN:   -disable-O0-optnone -o - -fsyntax-only %s -verify
-// REQUIRES: aarch64-registered-target
-
-#include <arm_sve.h>
-
-svfloat32_t test_exp_vv_i8mf8(svfloat32_t v) {
-
-  return __builtin_elementwise_exp(v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
-
-svfloat32_t test_exp2_vv_i8mf8(svfloat32_t v) {
-
-  return __builtin_elementwise_exp2(v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
diff --git a/clang/test/Sema/aarch64-sve-vector-log-ops.c b/clang/test/Sema/aarch64-sve-vector-log-ops.c
deleted file mode 100644
--- a/clang/test/Sema/aarch64-sve-vector-log-ops.c
+++ /dev/null
@@ -1,23 +0,0 @@
-// RUN: %clang_cc1 -triple aarch64 -target-feature +sve \
-// RUN:   -disable-O0-optnone -o - -fsyntax-only %s -verify
-// REQUIRES: aarch64-registered-target
-
-#include <arm_sve.h>
-
-svfloat32_t test_log_vv_i8mf8(svfloat32_t v) {
-
-  return __builtin_elementwise_log(v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
-
-svfloat32_t test_log10_vv_i8mf8(svfloat32_t v) {
-
-  return __builtin_elementwise_log10(v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
-
-svfloat32_t test_log2_vv_i8mf8(svfloat32_t v) {
-
-  return __builtin_elementwise_log2(v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
diff --git a/clang/test/Sema/aarch64-sve-vector-pow-ops.c b/clang/test/Sema/aarch64-sve-vector-pow-ops.c
deleted file mode 100644
--- a/clang/test/Sema/aarch64-sve-vector-pow-ops.c
+++ /dev/null
@@ -1,11 +0,0 @@
-// RUN: %clang_cc1 -triple aarch64 -target-feature +sve \
-// RUN:   -disable-O0-optnone -o - -fsyntax-only %s -verify
-// REQUIRES: aarch64-registered-target
-
-#include <arm_sve.h>
-
-svfloat32_t test_pow_vv_i8mf8(svfloat32_t v) {
-
-  return __builtin_elementwise_pow(v, v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
diff --git a/clang/test/Sema/aarch64-sve-vector-trig-ops.c b/clang/test/Sema/aarch64-sve-vector-trig-ops.c
deleted file mode 100644
--- a/clang/test/Sema/aarch64-sve-vector-trig-ops.c
+++ /dev/null
@@ -1,18 +0,0 @@
-// RUN: %clang_cc1 -triple aarch64 -target-feature +sve \
-// RUN:   -disable-O0-optnone -o - -fsyntax-only %s -verify
-// REQUIRES: aarch64-registered-target
-
-#include <arm_sve.h>
-
-
-svfloat32_t test_sin_vv_i8mf8(svfloat32_t v) {
-  
-  return __builtin_elementwise_sin(v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
-
-svfloat32_t test_cos_vv_i8mf8(svfloat32_t v) {
-
-  return __builtin_elementwise_cos(v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
diff --git a/clang/test/Sema/riscv-rvv-vector-elementwise.c b/clang/test/Sema/riscv-rvv-vector-elementwise.c
new file mode 100644
--- /dev/null
+++ b/clang/test/Sema/riscv-rvv-vector-elementwise.c
@@ -0,0 +1,520 @@
+// RUN: %clang_cc1 -triple riscv64 -target-feature +f -target-feature +d \
+// RUN:   -target-feature +v -target-feature +zfh -target-feature +zvfh \
+// RUN:   -disable-O0-optnone -o - -fsyntax-only %s -verify 
+// REQUIRES: riscv-registered-target
+
+#include <riscv_vector.h>
+
+void test_builtin_elementwise_abs(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                 vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_abs(vu8);
+  // expected-error@-1 {{1st argument must be a signed integer or floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_abs(vu16);
+  // expected-error@-1 {{1st argument must be a signed integer or floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_abs(vu32);
+  // expected-error@-1 {{1st argument must be a signed integer or floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_abs(vu64);
+  // expected-error@-1 {{1st argument must be a signed integer or floating point type}}
+}
+
+void test_builtin_elementwise_add_sat(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                 vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_add_sat(vf16, vf16);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+  vfloat32m1_t res_vf32 = __builtin_elementwise_add_sat(vf32, vf32);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+  vfloat64m1_t res_vf64 = __builtin_elementwise_add_sat(vf64, vf64);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+}
+
+void test_builtin_elementwise_sub_sat(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                 vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_sub_sat(vf16, vf16);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+  vfloat32m1_t res_vf32 = __builtin_elementwise_sub_sat(vf32, vf32);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+  vfloat64m1_t res_vf64 = __builtin_elementwise_sub_sat(vf64, vf64);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+}
+
+void test_builtin_elementwise_bitreverse(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vfloat16m1_t res_vf16 = __builtin_elementwise_bitreverse(vf16);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+  vfloat32m1_t res_vf32 = __builtin_elementwise_bitreverse(vf32);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+  vfloat64m1_t res_vf64 = __builtin_elementwise_bitreverse(vf64);
+  // expected-error@-1 {{1st argument must be a vector of integers}}
+}
+
+void test_builtin_elementwise_ceil(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_ceil(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_ceil(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_ceil(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_ceil(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_ceil(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_ceil(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_ceil(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_ceil(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_cos(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_cos(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_cos(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_cos(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_cos(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_cos(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_cos(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_cos(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_cos(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_exp(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_exp(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_exp(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_exp(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_exp(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_exp(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_exp(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_exp(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_exp(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_exp2(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_exp2(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_exp2(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_exp2(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_exp2(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_exp2(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_exp2(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_exp2(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_exp2(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_floor(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_floor(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_floor(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_floor(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_floor(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_floor(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_floor(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_floor(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_floor(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_log(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_log(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_log(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_log(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_log(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_log(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_log(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_log(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_log(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_log10(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_log10(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_log10(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_log10(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_log10(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_log10(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_log10(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_log10(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_log10(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_log2(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_log2(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_log2(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_log2(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_log2(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_log2(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_log2(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_log2(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_log2(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_pow(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_pow(vu8, vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_pow(vu16, vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_pow(vu32, vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_pow(vu64, vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_pow(vi8, vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_pow(vi16, vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_pow(vi32, vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_pow(vi64, vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_roundeven(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_roundeven(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_roundeven(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_roundeven(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_roundeven(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_roundeven(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_roundeven(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_roundeven(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_roundeven(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_round(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_round(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_round(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_round(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_round(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_round(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_round(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_round(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_round(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_rint(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_rint(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_rint(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_rint(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_rint(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_rint(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_rint(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_rint(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_rint(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_nearbyint(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_nearbyint(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_nearbyint(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_nearbyint(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_nearbyint(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_nearbyint(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_nearbyint(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_nearbyint(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_nearbyint(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_sin(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_sin(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_sin(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_sin(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_sin(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_sin(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_sin(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_sin(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_sin(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_sqrt(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_sqrt(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_sqrt(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_sqrt(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_sqrt(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_sqrt(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_sqrt(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_sqrt(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_sqrt(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_trunc(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_trunc(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_trunc(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_trunc(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_trunc(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_trunc(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_trunc(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_trunc(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_trunc(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_canonicalize(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_canonicalize(vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_canonicalize(vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_canonicalize(vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_canonicalize(vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_canonicalize(vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_canonicalize(vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_canonicalize(vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_canonicalize(vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_copysign(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_copysign(vu8, vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_copysign(vu16, vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_copysign(vu32, vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_copysign(vu64, vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_copysign(vi8, vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_copysign(vi16, vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_copysign(vi32, vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_copysign(vi64, vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
+
+void test_builtin_elementwise_fma(vint8m1_t vi8, vint16m1_t vi16,
+                                  vint32m1_t vi32, vint64m1_t vi64,
+                                  vuint8m1_t vu8, vuint16m1_t vu16,
+                                  vuint32m1_t vu32, vuint64m1_t vu64,
+                                  vfloat16m1_t vf16, vfloat32m1_t vf32,
+                                  vfloat64m1_t vf64) {
+  vuint8m1_t res_vu8 = __builtin_elementwise_fma(vu8, vu8, vu8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vu16 = __builtin_elementwise_fma(vu16, vu16, vu16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vu32 = __builtin_elementwise_fma(vu32, vu32, vu32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vu64 = __builtin_elementwise_fma(vu64, vu64, vu64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint8m1_t res_vi8 = __builtin_elementwise_fma(vi8, vi8, vi8);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint16m1_t res_vi16 = __builtin_elementwise_fma(vi16, vi16, vi16);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint32m1_t res_vi32 = __builtin_elementwise_fma(vi32, vi32, vi32);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+  vuint64m1_t res_vi64 = __builtin_elementwise_fma(vi64, vi64, vi64);
+  // expected-error@-1 {{1st argument must be a floating point type}}
+}
diff --git a/clang/test/Sema/riscv-rvv-vector-exp-ops.c b/clang/test/Sema/riscv-rvv-vector-exp-ops.c
deleted file mode 100644
--- a/clang/test/Sema/riscv-rvv-vector-exp-ops.c
+++ /dev/null
@@ -1,19 +0,0 @@
-// RUN: %clang_cc1 -triple riscv64 -target-feature +f -target-feature +d \
-// RUN:   -target-feature +v -target-feature +zfh -target-feature +zvfh \
-// RUN:   -disable-O0-optnone -o - -fsyntax-only %s -verify 
-// REQUIRES: riscv-registered-target
-
-#include <riscv_vector.h>
-
-
-vfloat32mf2_t test_exp_vv_i8mf8(vfloat32mf2_t v) {
-
-  return __builtin_elementwise_exp(v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
-
-vfloat32mf2_t test_exp2_vv_i8mf8(vfloat32mf2_t v) {
-
-  return __builtin_elementwise_exp2(v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
diff --git a/clang/test/Sema/riscv-rvv-vector-log-ops.c b/clang/test/Sema/riscv-rvv-vector-log-ops.c
deleted file mode 100644
--- a/clang/test/Sema/riscv-rvv-vector-log-ops.c
+++ /dev/null
@@ -1,25 +0,0 @@
-// RUN: %clang_cc1 -triple riscv64 -target-feature +f -target-feature +d \
-// RUN:   -target-feature +v -target-feature +zfh -target-feature +zvfh \
-// RUN:   -disable-O0-optnone -o - -fsyntax-only %s -verify 
-// REQUIRES: riscv-registered-target
-
-#include <riscv_vector.h>
-
-
-vfloat32mf2_t test_log_vv_i8mf8(vfloat32mf2_t v) {
-
-  return __builtin_elementwise_log(v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
-
-vfloat32mf2_t test_log10_vv_i8mf8(vfloat32mf2_t v) {
-
-  return __builtin_elementwise_log10(v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
-
-vfloat32mf2_t test_log2_vv_i8mf8(vfloat32mf2_t v) {
-
-  return __builtin_elementwise_log2(v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
diff --git a/clang/test/Sema/riscv-rvv-vector-trig-ops.c b/clang/test/Sema/riscv-rvv-vector-trig-ops.c
deleted file mode 100644
--- a/clang/test/Sema/riscv-rvv-vector-trig-ops.c
+++ /dev/null
@@ -1,19 +0,0 @@
-// RUN: %clang_cc1 -triple riscv64 -target-feature +f -target-feature +d \
-// RUN:   -target-feature +v -target-feature +zfh -target-feature +zvfh \
-// RUN:   -disable-O0-optnone -o - -fsyntax-only %s -verify 
-// REQUIRES: riscv-registered-target
-
-#include <riscv_vector.h>
-
-
-vfloat32mf2_t test_sin_vv_i8mf8(vfloat32mf2_t v) {
-
-  return __builtin_elementwise_sin(v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
-
-vfloat32mf2_t test_cos_vv_i8mf8(vfloat32mf2_t v) {
-
-  return __builtin_elementwise_cos(v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}
diff --git a/clang/test/Sema/riscv-sve-vector-pow-ops.c b/clang/test/Sema/riscv-sve-vector-pow-ops.c
deleted file mode 100644
--- a/clang/test/Sema/riscv-sve-vector-pow-ops.c
+++ /dev/null
@@ -1,13 +0,0 @@
-// RUN: %clang_cc1 -triple riscv64 -target-feature +f -target-feature +d \
-// RUN:   -target-feature +v -target-feature +zfh -target-feature +experimental-zvfh \
-// RUN:   -disable-O0-optnone -o - -fsyntax-only %s -verify 
-// REQUIRES: riscv-registered-target
-
-#include <riscv_vector.h>
-
-
-vfloat32mf2_t test_pow_vv_i8mf8(vfloat32mf2_t v) {
-
-  return __builtin_elementwise_pow(v, v);
-  // expected-error@-1 {{1st argument must be a vector, integer or floating point type}}
-}