Index: llvm/lib/Target/AArch64/AArch64ISelLowering.h
===================================================================
--- llvm/lib/Target/AArch64/AArch64ISelLowering.h
+++ llvm/lib/Target/AArch64/AArch64ISelLowering.h
@@ -1222,6 +1222,8 @@
 
   bool isConstantUnsignedBitfieldExtractLegal(unsigned Opc, LLT Ty1,
                                               LLT Ty2) const override;
+
+  bool preferScalarizeSplat(unsigned Opc) const override;
 };
 
 namespace AArch64 {
Index: llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
===================================================================
--- llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
+++ llvm/lib/Target/AArch64/AArch64ISelLowering.cpp
@@ -24495,3 +24495,10 @@
 
   return nullptr;
 }
+
+bool AArch64TargetLowering::preferScalarizeSplat(unsigned Opc) const {
+  if (Subtarget->forceStreamingCompatibleSVE() &&
+      (Opc == ISD::ZERO_EXTEND || Opc == ISD::SIGN_EXTEND))
+    return false;
+  return true;
+}
Index: llvm/test/CodeGen/AArch64/aarch64-force-streaming-compatible-sve.ll
===================================================================
--- /dev/null
+++ llvm/test/CodeGen/AArch64/aarch64-force-streaming-compatible-sve.ll
@@ -0,0 +1,48 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple aarch64-none-linux-gnu -mattr=+sve -force-streaming-compatible-sve | FileCheck %s --check-prefix=CHECK
+
+define void @jpeg_add_quant_table(i32 %0, <8 x i64> %1, ptr %2) {
+; CHECK-LABEL: jpeg_add_quant_table:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov z4.s, w0
+; CHECK-NEXT:    // kill: def $q2 killed $q2 def $z2
+; CHECK-NEXT:    ptrue p0.d, vl2
+; CHECK-NEXT:    sunpklo z4.d, z4.s
+; CHECK-NEXT:    // kill: def $q3 killed $q3 def $z3
+; CHECK-NEXT:    // kill: def $q1 killed $q1 def $z1
+; CHECK-NEXT:    // kill: def $q0 killed $q0 def $z0
+; CHECK-NEXT:    mul z2.d, p0/m, z2.d, z4.d
+; CHECK-NEXT:    mul z0.d, p0/m, z0.d, z4.d
+; CHECK-NEXT:    mul z3.d, p0/m, z3.d, z4.d
+; CHECK-NEXT:    cmpgt p2.d, p0/z, z2.d, #0
+; CHECK-NEXT:    mul z1.d, p0/m, z1.d, z4.d
+; CHECK-NEXT:    cmpgt p1.d, p0/z, z3.d, #0
+; CHECK-NEXT:    mov z3.d, p2/z, #-1 // =0xffffffffffffffff
+; CHECK-NEXT:    cmpgt p2.d, p0/z, z1.d, #0
+; CHECK-NEXT:    cmpgt p0.d, p0/z, z0.d, #0
+; CHECK-NEXT:    mov z2.d, p1/z, #-1 // =0xffffffffffffffff
+; CHECK-NEXT:    mov z0.d, p2/z, #-1 // =0xffffffffffffffff
+; CHECK-NEXT:    mov z1.d, p0/z, #-1 // =0xffffffffffffffff
+; CHECK-NEXT:    uzp1 z2.s, z2.s, z2.s
+; CHECK-NEXT:    uzp1 z3.s, z3.s, z3.s
+; CHECK-NEXT:    ptrue p1.s, vl2
+; CHECK-NEXT:    uzp1 z0.s, z0.s, z0.s
+; CHECK-NEXT:    uzp1 z1.s, z1.s, z1.s
+; CHECK-NEXT:    splice z3.s, p1, z3.s, z2.s
+; CHECK-NEXT:    splice z1.s, p1, z1.s, z0.s
+; CHECK-NEXT:    uzp1 z2.h, z3.h, z3.h
+; CHECK-NEXT:    uzp1 z0.h, z1.h, z1.h
+; CHECK-NEXT:    ptrue p0.h, vl4
+; CHECK-NEXT:    splice z0.h, p0, z0.h, z2.h
+; CHECK-NEXT:    and z0.h, z0.h, #0x1
+; CHECK-NEXT:    str q0, [x1]
+; CHECK-NEXT:    ret
+  %4 = sext i32 %0 to i64
+  %5 = insertelement <8 x i64> zeroinitializer, i64 %4, i64 0
+  %6 = shufflevector <8 x i64> %5, <8 x i64> zeroinitializer, <8 x i32> zeroinitializer
+  %7 = mul <8 x i64> %6, %1
+  %.not = icmp sgt <8 x i64> %7, zeroinitializer
+  %8 = zext <8 x i1> %.not to <8 x i16>
+  store <8 x i16> %8, ptr %2, align 2
+  ret void
+}