diff --git a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
--- a/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
+++ b/llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
@@ -1569,8 +1569,7 @@
 InstructionCost AArch64TTIImpl::getGatherScatterOpCost(
     unsigned Opcode, Type *DataTy, const Value *Ptr, bool VariableMask,
     Align Alignment, TTI::TargetCostKind CostKind, const Instruction *I) {
-
-  if (!isa<ScalableVectorType>(DataTy))
+  if (useNeonVector(DataTy))
     return BaseT::getGatherScatterOpCost(Opcode, DataTy, Ptr, VariableMask,
                                          Alignment, CostKind, I);
   auto *VT = cast<VectorType>(DataTy);
@@ -1589,7 +1588,11 @@
   ElementCount LegalVF = LT.second.getVectorElementCount();
   InstructionCost MemOpCost =
       getMemoryOpCost(Opcode, VT->getElementType(), Alignment, 0, CostKind, I);
-  return LT.first * MemOpCost * getMaxNumElements(LegalVF, I->getFunction());
+  unsigned int MaxNumElements = getMaxNumElements(LegalVF, I->getFunction());
+  if(isa<FixedVectorType>(DataTy))
+    return (LT.first * MemOpCost * MaxNumElements) / MaxNumElements;
+  else
+    return LT.first * MemOpCost * MaxNumElements;
 }
 
 bool AArch64TTIImpl::useNeonVector(const Type *Ty) const {
diff --git a/llvm/test/Analysis/CostModel/AArch64/sve-gather-vls.ll b/llvm/test/Analysis/CostModel/AArch64/sve-gather-vls.ll
new file mode 100644
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/AArch64/sve-gather-vls.ll
@@ -0,0 +1,51 @@
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=256 | FileCheck %s -D#VBITS=256
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=384 | FileCheck %s -D#VBITS=256
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=512 | FileCheck %s -D#VBITS=512
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=640 | FileCheck %s -D#VBITS=512
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=768 | FileCheck %s -D#VBITS=512
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=896 | FileCheck %s -D#VBITS=512
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1024 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1152 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1280 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1408 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1536 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1664 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1792 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1920 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=2048 | FileCheck %s -D#VBITS=2048
+
+target triple = "aarch64-unknown-linux-gnu"
+
+define void @sve-gather_vls(<512 x i1> %v512i1mask, <256 x i1> %v256i1mask, <128 x i1> %v128i1mask, <16 x i1> %v16i1mask) #0 {
+; CHECK-LABEL: 'sve-gather_vls'
+; CHECK: Cost Model: Found an estimated cost of [[#div(mul(div(2047,VBITS)+1,div(VBITS,8)),div(VBITS,8))]] for instruction: %res.v256i8 = call <256 x i8> @llvm.masked.gather.v256i8.v256p0i8(<256 x i8*> undef, i32 0, <256 x i1> %v256i1mask, <256 x i8> zeroinitializer)
+; CHECK: Cost Model: Found an estimated cost of [[#div(mul(div(4095,VBITS)+1,div(VBITS,16)),div(VBITS,16))]] for instruction: %res.v256i16 = call <256 x i16> @llvm.masked.gather.v256i16.v256p0i16(<256 x i16*> undef, i32 0, <256 x i1> %v256i1mask, <256 x i16> zeroinitializer)
+; CHECK: Cost Model: Found an estimated cost of [[#div(mul(div(511,VBITS)+1,div(VBITS,32)),div(VBITS,32))]] for instruction: %res.v16i32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0i32(<16 x i32*> undef, i32 0, <16 x i1> %v16i1mask, <16 x i32> zeroinitializer)
+; CHECK: Cost Model: Found an estimated cost of [[#div(mul(div(1023,VBITS)+1,div(VBITS,64)),div(VBITS,64))]] for instruction: %res.v16i64 = call <16 x i64> @llvm.masked.gather.v16i64.v16p0i64(<16 x i64*> undef, i32 0, <16 x i1> %v16i1mask, <16 x i64> zeroinitializer)
+; CHECK: Cost Model: Found an estimated cost of [[#div(mul(div(8191,VBITS)+1,div(VBITS,16)),div(VBITS,16))]] for instruction: %res.v512f16 = call <512 x half> @llvm.masked.gather.v512f16.v512p0f16(<512 x half*> undef, i32 0, <512 x i1> %v512i1mask, <512 x half> zeroinitializer)
+; CHECK: Cost Model: Found an estimated cost of [[#div(mul(div(8191,VBITS)+1,div(VBITS,32)),div(VBITS,32))]] for instruction: %res.v256f32 = call <256 x float> @llvm.masked.gather.v256f32.v256p0f32(<256 x float*> undef, i32 0, <256 x i1> %v256i1mask, <256 x float> zeroinitializer)
+; CHECK: Cost Model: Found an estimated cost of [[#div(mul(div(8191,VBITS)+1,div(VBITS,64)),div(VBITS,64))]] for instruction: %res.v128f64 = call <128 x double> @llvm.masked.gather.v128f64.v128p0f64(<128 x double*> undef, i32 0, <128 x i1> %v128i1mask, <128 x double> zeroinitializer)
+; CHECK: Cost Model: Found an estimated cost of 0 for instruction: ret void
+entry:
+  %res.v256i8 = call <256 x i8> @llvm.masked.gather.v256i8.v256p0i8(<256 x i8*> undef, i32 0, <256 x i1> %v256i1mask, <256 x i8> zeroinitializer)
+  %res.v256i16 = call <256 x i16> @llvm.masked.gather.v256i16.v256p0i16(<256 x i16*> undef, i32 0, <256 x i1> %v256i1mask, <256 x i16> zeroinitializer)
+  %res.v16i32 = call <16 x i32> @llvm.masked.gather.v16i32.v16p0i32(<16 x i32*> undef, i32 0, <16 x i1> %v16i1mask, <16 x i32> zeroinitializer)
+  %res.v16i64 = call <16 x i64> @llvm.masked.gather.v16i64.v16p0i64(<16 x i64*> undef, i32 0, <16 x i1> %v16i1mask, <16 x i64> zeroinitializer)
+
+  %res.v512f16 = call <512 x half> @llvm.masked.gather.v512f16.v512p0f16(<512 x half*> undef, i32 0, <512 x i1> %v512i1mask, <512 x half> zeroinitializer)
+  %res.v256f32 = call <256 x float> @llvm.masked.gather.v256f32.v256p0f32(<256 x float*> undef, i32 0, <256 x i1> %v256i1mask, <256 x float> zeroinitializer)
+  %res.v128f64 = call <128 x double> @llvm.masked.gather.v128f64.v128p0f64(<128 x double*> undef, i32 0, <128 x i1> %v128i1mask, <128 x double> zeroinitializer)
+
+  ret void
+}
+
+declare <256 x i8> @llvm.masked.gather.v256i8.v256p0i8(<256 x i8*>, i32, <256 x i1>, <256 x i8>)
+declare <256 x i16> @llvm.masked.gather.v256i16.v256p0i16(<256 x i16*>, i32, <256 x i1>, <256 x i16>)
+declare <16 x i32> @llvm.masked.gather.v16i32.v16p0i32(<16 x i32*>, i32, <16 x i1>, <16 x i32>)
+declare <16 x i64> @llvm.masked.gather.v16i64.v16p0i64(<16 x i64*>, i32, <16 x i1>, <16 x i64>)
+
+declare <512 x half> @llvm.masked.gather.v512f16.v512p0f16(<512 x half*>, i32, <512 x i1>, <512 x half>)
+declare <256 x float> @llvm.masked.gather.v256f32.v256p0f32(<256 x float*>, i32, <256 x i1>, <256 x float>)
+declare <128 x double> @llvm.masked.gather.v128f64.v128p0f64(<128 x double*>, i32, <128 x i1>, <128 x double>)
+
+attributes #0 = { "target-features"="+sve" }
diff --git a/llvm/test/Analysis/CostModel/AArch64/sve-scatter-vls.ll b/llvm/test/Analysis/CostModel/AArch64/sve-scatter-vls.ll
new file mode 100644
--- /dev/null
+++ b/llvm/test/Analysis/CostModel/AArch64/sve-scatter-vls.ll
@@ -0,0 +1,50 @@
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=256 | FileCheck %s -D#VBITS=256
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=384 | FileCheck %s -D#VBITS=256
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=512 | FileCheck %s -D#VBITS=512
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=640 | FileCheck %s -D#VBITS=512
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=768 | FileCheck %s -D#VBITS=512
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=896 | FileCheck %s -D#VBITS=512
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1024 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1152 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1280 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1408 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1536 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1664 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1792 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=1920 | FileCheck %s -D#VBITS=1024
+; RUN: opt < %s -cost-model -analyze -aarch64-sve-vector-bits-min=2048 | FileCheck %s -D#VBITS=2048
+
+target triple = "aarch64-unknown-linux-gnu"
+
+define void @sve-scatter_vls(<512 x i1> %v512i1mask, <256 x i1> %v256i1mask, <128 x i1> %v128i1mask, <16 x i1> %v16i1mask) #0 {
+; CHECK-LABEL: 'sve-scatter_vls'
+; CHECK: Cost Model: Found an estimated cost of [[#div(mul(div(2047,VBITS)+1,div(VBITS,8)),div(VBITS,8))]] for instruction: call void @llvm.masked.scatter.v256i8.v256p0i8(<256 x i8> undef, <256 x i8*> undef, i32 0, <256 x i1> %v256i1mask)
+; CHECK: Cost Model: Found an estimated cost of [[#div(mul(div(4095,VBITS)+1,div(VBITS,16)),div(VBITS,16))]] for instruction: call void @llvm.masked.scatter.v256i16.v256p0i16(<256 x i16> undef, <256 x i16*> undef, i32 0, <256 x i1> %v256i1mask)
+; CHECK: Cost Model: Found an estimated cost of [[#div(mul(div(511,VBITS)+1,div(VBITS,32)),div(VBITS,32))]] for instruction: call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> undef, <16 x i32*> undef, i32 0, <16 x i1> %v16i1mask)
+; CHECK: Cost Model: Found an estimated cost of [[#div(mul(div(1023,VBITS)+1,div(VBITS,64)),div(VBITS,64))]] for instruction: call void @llvm.masked.scatter.v16i64.v16p0i64(<16 x i64> undef, <16 x i64*> undef, i32 0, <16 x i1> %v16i1mask)
+; CHECK: Cost Model: Found an estimated cost of [[#div(mul(div(8191,VBITS)+1,div(VBITS,16)),div(VBITS,16))]] for instruction: call void @llvm.masked.scatter.v512f16.v512p0f16(<512 x half> undef, <512 x half*> undef, i32 0, <512 x i1> %v512i1mask)
+; CHECK: Cost Model: Found an estimated cost of [[#div(mul(div(8191,VBITS)+1,div(VBITS,32)),div(VBITS,32))]] for instruction: call void @llvm.masked.scatter.v256f32.v256p0f32(<256 x float> undef, <256 x float*> undef, i32 0, <256 x i1> %v256i1mask)
+; CHECK: Cost Model: Found an estimated cost of [[#div(mul(div(8191,VBITS)+1,div(VBITS,64)),div(VBITS,64))]] for instruction: call void @llvm.masked.scatter.v128f64.v128p0f64(<128 x double> undef, <128 x double*> undef, i32 0, <128 x i1> %v128i1mask)
+; CHECK: Cost Model: Found an estimated cost of 0 for instruction: ret void
+entry:
+  call void @llvm.masked.scatter.v256i8.v256p0i8(<256 x i8> undef, <256 x i8*> undef, i32 0, <256 x i1> %v256i1mask)
+  call void @llvm.masked.scatter.v256i16.v256p0i16(<256 x i16> undef, <256 x i16*> undef, i32 0, <256 x i1> %v256i1mask)
+  call void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32> undef, <16 x i32*> undef, i32 0, <16 x i1> %v16i1mask)
+  call void @llvm.masked.scatter.v16i64.v16p0i64(<16 x i64> undef, <16 x i64*> undef, i32 0, <16 x i1> %v16i1mask)
+
+  call void @llvm.masked.scatter.v512f16.v512p0f16(<512 x half> undef, <512 x half*> undef, i32 0, <512 x i1> %v512i1mask)
+  call void @llvm.masked.scatter.v256f32.v256p0f32(<256 x float> undef, <256 x float*> undef, i32 0, <256 x i1> %v256i1mask)
+  call void @llvm.masked.scatter.v128f64.v128p0f64(<128 x double> undef, <128 x double*> undef, i32 0, <128 x i1> %v128i1mask)
+  ret void
+}
+
+declare void @llvm.masked.scatter.v256i8.v256p0i8(<256 x i8>, <256 x i8*>, i32, <256 x i1>)
+declare void @llvm.masked.scatter.v256i16.v256p0i16(<256 x i16>, <256 x i16*>, i32, <256 x i1>)
+declare void @llvm.masked.scatter.v16i32.v16p0i32(<16 x i32>, <16 x i32*>, i32, <16 x i1>)
+declare void @llvm.masked.scatter.v16i64.v16p0i64(<16 x i64>, <16 x i64*>, i32, <16 x i1>)
+
+declare void @llvm.masked.scatter.v512f16.v512p0f16(<512 x half>, <512 x half*>, i32, <512 x i1>)
+declare void @llvm.masked.scatter.v256f32.v256p0f32(<256 x float>, <256 x float*>, i32, <256 x i1>)
+declare void @llvm.masked.scatter.v128f64.v128p0f64(<128 x double>, <128 x double*>, i32, <128 x i1>)
+
+attributes #0 = { "target-features"="+sve" }