Index: llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp
===================================================================
--- llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp
+++ llvm/lib/Target/ARM/ARMTargetTransformInfo.cpp
@@ -508,11 +508,22 @@
 int ARMTTIImpl::getCmpSelInstrCost(unsigned Opcode, Type *ValTy, Type *CondTy,
                                    TTI::TargetCostKind CostKind,
                                    const Instruction *I) {
-  // TODO: Handle other cost kinds.
-  if (CostKind != TTI::TCK_RecipThroughput)
+  int ISD = TLI->InstructionOpcodeToISD(Opcode);
+  if (ST->isMClass() &&
+      (CostKind != TTI::TCK_RecipThroughput || !ValTy->isVectorTy())) {
+    // Simplify and assume that more (scalar) instructions reduces throughput,
+    // as well as increasing latency and code size.
+    if (ISD == ISD::SELECT) {
+      // Select costs can vary because they:
+      // - may require one or more conditional mov (including an IT),
+      // - can't operate directly on immediates,
+      // - require live flags, which we can't copy around easily.
+      std::pair<int, MVT> LT = TLI->getTypeLegalizationCost(DL, ValTy);
+      return LT.first + 1;
+    }
     return BaseT::getCmpSelInstrCost(Opcode, ValTy, CondTy, CostKind, I);
+  }
 
-  int ISD = TLI->InstructionOpcodeToISD(Opcode);
   // On NEON a vector select gets lowered to vbsl.
   if (ST->hasNEON() && ValTy->isVectorTy() && ISD == ISD::SELECT) {
     // Lowering of some vector selects is currently far from perfect.
Index: llvm/test/Analysis/CostModel/ARM/arith-ssat.ll
===================================================================
--- llvm/test/Analysis/CostModel/ARM/arith-ssat.ll
+++ llvm/test/Analysis/CostModel/ARM/arith-ssat.ll
@@ -28,22 +28,22 @@
 
 define i32 @add(i32 %arg) {
 ; V8M-RECIP-LABEL: 'add'
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 216 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 144 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 288 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 144 for instruction: %V8I16 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 288 for instruction: %V16I16 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 576 for instruction: %V32I16 = call <32 x i16> @llvm.sadd.sat.v32i16(<32 x i16> undef, <32 x i16> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %I8 = call i8 @llvm.sadd.sat.i8(i8 undef, i8 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 288 for instruction: %V16I8 = call <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8> undef, <16 x i8> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 576 for instruction: %V32I8 = call <32 x i8> @llvm.sadd.sat.v32i8(<32 x i8> undef, <32 x i8> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 1152 for instruction: %V64I8 = call <64 x i8> @llvm.sadd.sat.v64i8(<64 x i8> undef, <64 x i8> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 232 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V8I16 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V16I16 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 640 for instruction: %V32I16 = call <32 x i16> @llvm.sadd.sat.v32i16(<32 x i16> undef, <32 x i16> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %I8 = call i8 @llvm.sadd.sat.i8(i8 undef, i8 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V16I8 = call <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8> undef, <16 x i8> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 640 for instruction: %V32I8 = call <32 x i8> @llvm.sadd.sat.v32i8(<32 x i8> undef, <32 x i8> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 1280 for instruction: %V64I8 = call <64 x i8> @llvm.sadd.sat.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-RECIP-LABEL: 'add'
@@ -66,19 +66,19 @@
 ; NEON-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-RECIP-LABEL: 'add'
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 118 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 302 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 1046 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %I64 = call i64 @llvm.sadd.sat.i64(i64 undef, i64 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 126 for instruction: %V2I64 = call <2 x i64> @llvm.sadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 318 for instruction: %V4I64 = call <4 x i64> @llvm.sadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 1078 for instruction: %V8I64 = call <8 x i64> @llvm.sadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %I32 = call i32 @llvm.sadd.sat.i32(i32 undef, i32 undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.sadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8I32 = call <8 x i32> @llvm.sadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V16I32 = call <16 x i32> @llvm.sadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %I16 = call i16 @llvm.sadd.sat.i16(i16 undef, i16 undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I16 = call <8 x i16> @llvm.sadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.sadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V32I16 = call <32 x i16> @llvm.sadd.sat.v32i16(<32 x i16> undef, <32 x i16> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %I8 = call i8 @llvm.sadd.sat.i8(i8 undef, i8 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %I8 = call i8 @llvm.sadd.sat.i8(i8 undef, i8 undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call <16 x i8> @llvm.sadd.sat.v16i8(<16 x i8> undef, <16 x i8> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I8 = call <32 x i8> @llvm.sadd.sat.v32i8(<32 x i8> undef, <32 x i8> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V64I8 = call <64 x i8> @llvm.sadd.sat.v64i8(<64 x i8> undef, <64 x i8> undef)
@@ -186,22 +186,22 @@
 
 define i32 @sub(i32 %arg) {
 ; V8M-RECIP-LABEL: 'sub'
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 54 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 108 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 216 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 144 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 288 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 144 for instruction: %V8I16 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 288 for instruction: %V16I16 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 576 for instruction: %V32I16 = call <32 x i16> @llvm.ssub.sat.v32i16(<32 x i16> undef, <32 x i16> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %I8 = call i8 @llvm.ssub.sat.i8(i8 undef, i8 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 288 for instruction: %V16I8 = call <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8> undef, <16 x i8> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 576 for instruction: %V32I8 = call <32 x i8> @llvm.ssub.sat.v32i8(<32 x i8> undef, <32 x i8> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 1152 for instruction: %V64I8 = call <64 x i8> @llvm.ssub.sat.v64i8(<64 x i8> undef, <64 x i8> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 58 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 116 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 232 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V8I16 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V16I16 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 640 for instruction: %V32I16 = call <32 x i16> @llvm.ssub.sat.v32i16(<32 x i16> undef, <32 x i16> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %I8 = call i8 @llvm.ssub.sat.i8(i8 undef, i8 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V16I8 = call <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8> undef, <16 x i8> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 640 for instruction: %V32I8 = call <32 x i8> @llvm.ssub.sat.v32i8(<32 x i8> undef, <32 x i8> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 1280 for instruction: %V64I8 = call <64 x i8> @llvm.ssub.sat.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-RECIP-LABEL: 'sub'
@@ -224,19 +224,19 @@
 ; NEON-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-RECIP-LABEL: 'sub'
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 118 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 302 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 1046 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 15 for instruction: %I64 = call i64 @llvm.ssub.sat.i64(i64 undef, i64 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 126 for instruction: %V2I64 = call <2 x i64> @llvm.ssub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 318 for instruction: %V4I64 = call <4 x i64> @llvm.ssub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 1078 for instruction: %V8I64 = call <8 x i64> @llvm.ssub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %I32 = call i32 @llvm.ssub.sat.i32(i32 undef, i32 undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.ssub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V8I32 = call <8 x i32> @llvm.ssub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 62 for instruction: %V16I32 = call <16 x i32> @llvm.ssub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %I16 = call i16 @llvm.ssub.sat.i16(i16 undef, i16 undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V8I16 = call <8 x i16> @llvm.ssub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 34 for instruction: %V16I16 = call <16 x i16> @llvm.ssub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 68 for instruction: %V32I16 = call <32 x i16> @llvm.ssub.sat.v32i16(<32 x i16> undef, <32 x i16> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %I8 = call i8 @llvm.ssub.sat.i8(i8 undef, i8 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %I8 = call i8 @llvm.ssub.sat.i8(i8 undef, i8 undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V16I8 = call <16 x i8> @llvm.ssub.sat.v16i8(<16 x i8> undef, <16 x i8> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V32I8 = call <32 x i8> @llvm.ssub.sat.v32i8(<32 x i8> undef, <32 x i8> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V64I8 = call <64 x i8> @llvm.ssub.sat.v64i8(<64 x i8> undef, <64 x i8> undef)
Index: llvm/test/Analysis/CostModel/ARM/arith-usat.ll
===================================================================
--- llvm/test/Analysis/CostModel/ARM/arith-usat.ll
+++ llvm/test/Analysis/CostModel/ARM/arith-usat.ll
@@ -28,22 +28,22 @@
 
 define i32 @add(i32 %arg) {
 ; V8M-RECIP-LABEL: 'add'
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <8 x i32> @llvm.uadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I32 = call <16 x i32> @llvm.uadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.uadd.sat.i16(i16 undef, i16 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.uadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I16 = call <16 x i16> @llvm.uadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call <32 x i16> @llvm.uadd.sat.v32i16(<32 x i16> undef, <32 x i16> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.uadd.sat.i8(i8 undef, i8 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.uadd.sat.v16i8(<16 x i8> undef, <16 x i8> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I8 = call <32 x i8> @llvm.uadd.sat.v32i8(<32 x i8> undef, <32 x i8> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V64I8 = call <64 x i8> @llvm.uadd.sat.v64i8(<64 x i8> undef, <64 x i8> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I32 = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I32 = call <8 x i32> @llvm.uadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V16I32 = call <16 x i32> @llvm.uadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.uadd.sat.i16(i16 undef, i16 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I16 = call <8 x i16> @llvm.uadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V16I16 = call <16 x i16> @llvm.uadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V32I16 = call <32 x i16> @llvm.uadd.sat.v32i16(<32 x i16> undef, <32 x i16> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.uadd.sat.i8(i8 undef, i8 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V16I8 = call <16 x i8> @llvm.uadd.sat.v16i8(<16 x i8> undef, <16 x i8> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V32I8 = call <32 x i8> @llvm.uadd.sat.v32i8(<32 x i8> undef, <32 x i8> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 384 for instruction: %V64I8 = call <64 x i8> @llvm.uadd.sat.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-RECIP-LABEL: 'add'
@@ -66,19 +66,19 @@
 ; NEON-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-RECIP-LABEL: 'add'
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 448 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %I64 = call i64 @llvm.uadd.sat.i64(i64 undef, i64 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V2I64 = call <2 x i64> @llvm.uadd.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 136 for instruction: %V4I64 = call <4 x i64> @llvm.uadd.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 464 for instruction: %V8I64 = call <8 x i64> @llvm.uadd.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.uadd.sat.i32(i32 undef, i32 undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call <4 x i32> @llvm.uadd.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I32 = call <8 x i32> @llvm.uadd.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16I32 = call <16 x i32> @llvm.uadd.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.uadd.sat.i16(i16 undef, i16 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.uadd.sat.i16(i16 undef, i16 undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.uadd.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.uadd.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.uadd.sat.v32i16(<32 x i16> undef, <32 x i16> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.uadd.sat.i8(i8 undef, i8 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.uadd.sat.i8(i8 undef, i8 undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16I8 = call <16 x i8> @llvm.uadd.sat.v16i8(<16 x i8> undef, <16 x i8> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V32I8 = call <32 x i8> @llvm.uadd.sat.v32i8(<32 x i8> undef, <32 x i8> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V64I8 = call <64 x i8> @llvm.uadd.sat.v64i8(<64 x i8> undef, <64 x i8> undef)
@@ -186,22 +186,22 @@
 
 define i32 @sub(i32 %arg) {
 ; V8M-RECIP-LABEL: 'sub'
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 18 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 36 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 72 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V8I16 = call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I16 = call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I16 = call <32 x i16> @llvm.usub.sat.v32i16(<32 x i16> undef, <32 x i16> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V16I8 = call <16 x i8> @llvm.usub.sat.v16i8(<16 x i8> undef, <16 x i8> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %V32I8 = call <32 x i8> @llvm.usub.sat.v32i8(<32 x i8> undef, <32 x i8> undef)
-; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 320 for instruction: %V64I8 = call <64 x i8> @llvm.usub.sat.v64i8(<64 x i8> undef, <64 x i8> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %V8I16 = call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V16I16 = call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V32I16 = call <32 x i16> @llvm.usub.sat.v32i16(<32 x i16> undef, <32 x i16> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 96 for instruction: %V16I8 = call <16 x i8> @llvm.usub.sat.v16i8(<16 x i8> undef, <16 x i8> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 192 for instruction: %V32I8 = call <32 x i8> @llvm.usub.sat.v32i8(<32 x i8> undef, <32 x i8> undef)
+; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 384 for instruction: %V64I8 = call <64 x i8> @llvm.usub.sat.v64i8(<64 x i8> undef, <64 x i8> undef)
 ; V8M-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-RECIP-LABEL: 'sub'
@@ -224,19 +224,19 @@
 ; NEON-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-RECIP-LABEL: 'sub'
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 128 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 448 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %I64 = call i64 @llvm.usub.sat.i64(i64 undef, i64 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 44 for instruction: %V2I64 = call <2 x i64> @llvm.usub.sat.v2i64(<2 x i64> undef, <2 x i64> undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 136 for instruction: %V4I64 = call <4 x i64> @llvm.usub.sat.v4i64(<4 x i64> undef, <4 x i64> undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 464 for instruction: %V8I64 = call <8 x i64> @llvm.usub.sat.v8i64(<8 x i64> undef, <8 x i64> undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I32 = call i32 @llvm.usub.sat.i32(i32 undef, i32 undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V4I32 = call <4 x i32> @llvm.usub.sat.v4i32(<4 x i32> undef, <4 x i32> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V8I32 = call <8 x i32> @llvm.usub.sat.v8i32(<8 x i32> undef, <8 x i32> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V16I32 = call <16 x i32> @llvm.usub.sat.v16i32(<16 x i32> undef, <16 x i32> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I16 = call i16 @llvm.usub.sat.i16(i16 undef, i16 undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V8I16 = call <8 x i16> @llvm.usub.sat.v8i16(<8 x i16> undef, <8 x i16> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V16I16 = call <16 x i16> @llvm.usub.sat.v16i16(<16 x i16> undef, <16 x i16> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V32I16 = call <32 x i16> @llvm.usub.sat.v32i16(<32 x i16> undef, <32 x i16> undef)
-; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef)
+; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %I8 = call i8 @llvm.usub.sat.i8(i8 undef, i8 undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %V16I8 = call <16 x i8> @llvm.usub.sat.v16i8(<16 x i8> undef, <16 x i8> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %V32I8 = call <32 x i8> @llvm.usub.sat.v32i8(<32 x i8> undef, <32 x i8> undef)
 ; MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %V64I8 = call <64 x i8> @llvm.usub.sat.v64i8(<64 x i8> undef, <64 x i8> undef)
Index: llvm/test/Analysis/CostModel/ARM/arith.ll
===================================================================
--- llvm/test/Analysis/CostModel/ARM/arith.ll
+++ llvm/test/Analysis/CostModel/ARM/arith.ll
@@ -17,6 +17,7 @@
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %i = and i8 undef, undef
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %j = or i8 undef, undef
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %k = xor i8 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-MAIN-LABEL: 'i8'
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %c = add i8 undef, undef
@@ -28,6 +29,7 @@
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %i = and i8 undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %j = or i8 undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %k = xor i8 undef, undef
+; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-BASE-LABEL: 'i8'
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %c = add i8 undef, undef
@@ -39,6 +41,7 @@
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %i = and i8 undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %j = or i8 undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %k = xor i8 undef, undef
+; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8R-LABEL: 'i8'
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %c = add i8 undef, undef
@@ -50,6 +53,7 @@
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %i = and i8 undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %j = or i8 undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %k = xor i8 undef, undef
+; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %c = add i8 undef, undef
   %d = sub i8 undef, undef
@@ -74,6 +78,7 @@
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %i = and i16 undef, undef
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %j = or i16 undef, undef
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %k = xor i16 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-MAIN-LABEL: 'i16'
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %c = add i16 undef, undef
@@ -85,6 +90,7 @@
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %i = and i16 undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %j = or i16 undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %k = xor i16 undef, undef
+; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-BASE-LABEL: 'i16'
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %c = add i16 undef, undef
@@ -96,6 +102,7 @@
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %i = and i16 undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %j = or i16 undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %k = xor i16 undef, undef
+; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8R-LABEL: 'i16'
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %c = add i16 undef, undef
@@ -107,6 +114,7 @@
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %i = and i16 undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %j = or i16 undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %k = xor i16 undef, undef
+; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %c = add i16 undef, undef
   %d = sub i16 undef, undef
@@ -131,6 +139,7 @@
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %i = and i32 undef, undef
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %j = or i32 undef, undef
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %k = xor i32 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-MAIN-LABEL: 'i32'
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %c = add i32 undef, undef
@@ -142,6 +151,7 @@
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %i = and i32 undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %j = or i32 undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %k = xor i32 undef, undef
+; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-BASE-LABEL: 'i32'
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %c = add i32 undef, undef
@@ -153,6 +163,7 @@
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %i = and i32 undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %j = or i32 undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %k = xor i32 undef, undef
+; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8R-LABEL: 'i32'
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %c = add i32 undef, undef
@@ -164,6 +175,7 @@
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %i = and i32 undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %j = or i32 undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %k = xor i32 undef, undef
+; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %c = add i32 undef, undef
   %d = sub i32 undef, undef
@@ -188,6 +200,7 @@
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %i = and i64 undef, undef
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %j = or i64 undef, undef
 ; CHECK-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %k = xor i64 undef, undef
+; CHECK-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-MAIN-LABEL: 'i64'
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %c = add i64 undef, undef
@@ -199,6 +212,7 @@
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %i = and i64 undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %j = or i64 undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %k = xor i64 undef, undef
+; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-BASE-LABEL: 'i64'
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %c = add i64 undef, undef
@@ -210,6 +224,7 @@
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %i = and i64 undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %j = or i64 undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %k = xor i64 undef, undef
+; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8R-LABEL: 'i64'
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %c = add i64 undef, undef
@@ -221,6 +236,7 @@
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %i = and i64 undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %j = or i64 undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %k = xor i64 undef, undef
+; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %c = add i64 undef, undef
   %d = sub i64 undef, undef
@@ -273,6 +289,7 @@
 ; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %i16 = and <16 x i8> undef, undef
 ; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %j16 = or <16 x i8> undef, undef
 ; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %k16 = xor <16 x i8> undef, undef
+; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-MVE2-LABEL: 'vi8'
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %c2 = add <2 x i8> undef, undef
@@ -311,6 +328,7 @@
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %i16 = and <16 x i8> undef, undef
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %j16 = or <16 x i8> undef, undef
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %k16 = xor <16 x i8> undef, undef
+; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-MVE4-LABEL: 'vi8'
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %c2 = add <2 x i8> undef, undef
@@ -349,6 +367,7 @@
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %i16 = and <16 x i8> undef, undef
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %j16 = or <16 x i8> undef, undef
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %k16 = xor <16 x i8> undef, undef
+; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-MAIN-LABEL: 'vi8'
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %c2 = add <2 x i8> undef, undef
@@ -387,6 +406,7 @@
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %i16 = and <16 x i8> undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %j16 = or <16 x i8> undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %k16 = xor <16 x i8> undef, undef
+; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-BASE-LABEL: 'vi8'
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %c2 = add <2 x i8> undef, undef
@@ -425,6 +445,7 @@
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %i16 = and <16 x i8> undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %j16 = or <16 x i8> undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %k16 = xor <16 x i8> undef, undef
+; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8R-LABEL: 'vi8'
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %c2 = add <2 x i8> undef, undef
@@ -463,6 +484,7 @@
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %i16 = and <16 x i8> undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %j16 = or <16 x i8> undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %k16 = xor <16 x i8> undef, undef
+; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %c2 = add <2 x i8> undef, undef
   %d2 = sub <2 x i8> undef, undef
@@ -541,6 +563,7 @@
 ; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %i16 = and <16 x i16> undef, undef
 ; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %j16 = or <16 x i16> undef, undef
 ; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %k16 = xor <16 x i16> undef, undef
+; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-MVE2-LABEL: 'vi16'
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %c2 = add <2 x i16> undef, undef
@@ -579,6 +602,7 @@
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %i16 = and <16 x i16> undef, undef
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %j16 = or <16 x i16> undef, undef
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %k16 = xor <16 x i16> undef, undef
+; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-MVE4-LABEL: 'vi16'
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %c2 = add <2 x i16> undef, undef
@@ -617,6 +641,7 @@
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %i16 = and <16 x i16> undef, undef
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %j16 = or <16 x i16> undef, undef
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %k16 = xor <16 x i16> undef, undef
+; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-MAIN-LABEL: 'vi16'
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %c2 = add <2 x i16> undef, undef
@@ -655,6 +680,7 @@
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %i16 = and <16 x i16> undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %j16 = or <16 x i16> undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %k16 = xor <16 x i16> undef, undef
+; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-BASE-LABEL: 'vi16'
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %c2 = add <2 x i16> undef, undef
@@ -693,6 +719,7 @@
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %i16 = and <16 x i16> undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %j16 = or <16 x i16> undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %k16 = xor <16 x i16> undef, undef
+; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8R-LABEL: 'vi16'
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %c2 = add <2 x i16> undef, undef
@@ -731,6 +758,7 @@
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %i16 = and <16 x i16> undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %j16 = or <16 x i16> undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %k16 = xor <16 x i16> undef, undef
+; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %c2 = add <2 x i16> undef, undef
   %d2 = sub <2 x i16> undef, undef
@@ -809,6 +837,7 @@
 ; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %i16 = and <16 x i32> undef, undef
 ; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %j16 = or <16 x i32> undef, undef
 ; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %k16 = xor <16 x i32> undef, undef
+; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-MVE2-LABEL: 'vi32'
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %c2 = add <2 x i32> undef, undef
@@ -847,6 +876,7 @@
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %i16 = and <16 x i32> undef, undef
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %j16 = or <16 x i32> undef, undef
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %k16 = xor <16 x i32> undef, undef
+; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-MVE4-LABEL: 'vi32'
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %c2 = add <2 x i32> undef, undef
@@ -885,6 +915,7 @@
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %i16 = and <16 x i32> undef, undef
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %j16 = or <16 x i32> undef, undef
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %k16 = xor <16 x i32> undef, undef
+; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-MAIN-LABEL: 'vi32'
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %c2 = add <2 x i32> undef, undef
@@ -923,6 +954,7 @@
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %i16 = and <16 x i32> undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %j16 = or <16 x i32> undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %k16 = xor <16 x i32> undef, undef
+; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-BASE-LABEL: 'vi32'
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %c2 = add <2 x i32> undef, undef
@@ -961,6 +993,7 @@
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %i16 = and <16 x i32> undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %j16 = or <16 x i32> undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %k16 = xor <16 x i32> undef, undef
+; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8R-LABEL: 'vi32'
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %c2 = add <2 x i32> undef, undef
@@ -999,6 +1032,7 @@
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %i16 = and <16 x i32> undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %j16 = or <16 x i32> undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %k16 = xor <16 x i32> undef, undef
+; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %c2 = add <2 x i32> undef, undef
   %d2 = sub <2 x i32> undef, undef
@@ -1077,6 +1111,7 @@
 ; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %i16 = and <16 x i64> undef, undef
 ; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %j16 = or <16 x i64> undef, undef
 ; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %k16 = xor <16 x i64> undef, undef
+; CHECK-MVE1-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-MVE2-LABEL: 'vi64'
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %c2 = add <2 x i64> undef, undef
@@ -1115,6 +1150,7 @@
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %i16 = and <16 x i64> undef, undef
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %j16 = or <16 x i64> undef, undef
 ; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %k16 = xor <16 x i64> undef, undef
+; CHECK-MVE2-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-MVE4-LABEL: 'vi64'
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %c2 = add <2 x i64> undef, undef
@@ -1153,6 +1189,7 @@
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %i16 = and <16 x i64> undef, undef
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %j16 = or <16 x i64> undef, undef
 ; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %k16 = xor <16 x i64> undef, undef
+; CHECK-MVE4-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-MAIN-LABEL: 'vi64'
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %c2 = add <2 x i64> undef, undef
@@ -1191,6 +1228,7 @@
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %i16 = and <16 x i64> undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %j16 = or <16 x i64> undef, undef
 ; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %k16 = xor <16 x i64> undef, undef
+; CHECK-V8M-MAIN-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8M-BASE-LABEL: 'vi64'
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %c2 = add <2 x i64> undef, undef
@@ -1229,6 +1267,7 @@
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %i16 = and <16 x i64> undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %j16 = or <16 x i64> undef, undef
 ; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %k16 = xor <16 x i64> undef, undef
+; CHECK-V8M-BASE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-V8R-LABEL: 'vi64'
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %c2 = add <2 x i64> undef, undef
@@ -1267,6 +1306,7 @@
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %i16 = and <16 x i64> undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %j16 = or <16 x i64> undef, undef
 ; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %k16 = xor <16 x i64> undef, undef
+; CHECK-V8R-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %c2 = add <2 x i64> undef, undef
   %d2 = sub <2 x i64> undef, undef
Index: llvm/test/Analysis/CostModel/ARM/immediates.ll
===================================================================
--- llvm/test/Analysis/CostModel/ARM/immediates.ll
+++ llvm/test/Analysis/CostModel/ARM/immediates.ll
@@ -25,8 +25,8 @@
 ; CHECK-T1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cmp_244 = icmp ne i32 undef, 244
 ; CHECK-T1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cmp_256 = icmp uge i32 undef, 256
 ; CHECK-T1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cmp_1024 = icmp ult i32 undef, 1024
-; CHECK-T1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %select_1_0 = select i1 undef, i32 1, i32 0
-; CHECK-T1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %select_7_255 = select i1 undef, i32 7, i32 255
+; CHECK-T1-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %select_1_0 = select i1 undef, i32 1, i32 0
+; CHECK-T1-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %select_7_255 = select i1 undef, i32 7, i32 255
 ; CHECK-T1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 1
 ;
 ; CHECK-T2-SIZE-LABEL: 'const_costs'
@@ -47,8 +47,8 @@
 ; CHECK-T2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cmp_244 = icmp ne i32 undef, 244
 ; CHECK-T2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cmp_256 = icmp uge i32 undef, 256
 ; CHECK-T2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cmp_1024 = icmp ult i32 undef, 1024
-; CHECK-T2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %select_1_0 = select i1 undef, i32 1, i32 0
-; CHECK-T2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %select_7_255 = select i1 undef, i32 7, i32 255
+; CHECK-T2-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %select_1_0 = select i1 undef, i32 1, i32 0
+; CHECK-T2-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %select_7_255 = select i1 undef, i32 7, i32 255
 ; CHECK-T2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 1
 ;
 ; CHECK-T1-LATENCY-LABEL: 'const_costs'
@@ -113,8 +113,8 @@
 ; CHECK-T1-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cmp_244 = icmp ne i32 undef, 244
 ; CHECK-T1-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cmp_256 = icmp uge i32 undef, 256
 ; CHECK-T1-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cmp_1024 = icmp ult i32 undef, 1024
-; CHECK-T1-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %select_1_0 = select i1 undef, i32 1, i32 0
-; CHECK-T1-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %select_7_255 = select i1 undef, i32 7, i32 255
+; CHECK-T1-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %select_1_0 = select i1 undef, i32 1, i32 0
+; CHECK-T1-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %select_7_255 = select i1 undef, i32 7, i32 255
 ; CHECK-T1-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 1
 ;
 ; CHECK-T2-THROUGHPUT-LABEL: 'const_costs'
@@ -135,8 +135,8 @@
 ; CHECK-T2-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cmp_244 = icmp ne i32 undef, 244
 ; CHECK-T2-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cmp_256 = icmp uge i32 undef, 256
 ; CHECK-T2-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %cmp_1024 = icmp ult i32 undef, 1024
-; CHECK-T2-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %select_1_0 = select i1 undef, i32 1, i32 0
-; CHECK-T2-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %select_7_255 = select i1 undef, i32 7, i32 255
+; CHECK-T2-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %select_1_0 = select i1 undef, i32 1, i32 0
+; CHECK-T2-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %select_7_255 = select i1 undef, i32 7, i32 255
 ; CHECK-T2-THROUGHPUT-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 1
 ;
   %add_1 = add i32 undef, 1
Index: llvm/test/Analysis/CostModel/ARM/reduce-smax.ll
===================================================================
--- llvm/test/Analysis/CostModel/ARM/reduce-smax.ll
+++ llvm/test/Analysis/CostModel/ARM/reduce-smax.ll
@@ -6,10 +6,10 @@
 define i32 @reduce_i64(i32 %arg) {
 ; V8M-LABEL: 'reduce_i64'
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.smax.v1i64(<1 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smax.v2i64(<2 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smax.v4i64(<4 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smax.v8i64(<8 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 167 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smax.v16i64(<16 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smax.v2i64(<2 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smax.v4i64(<4 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smax.v8i64(<8 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 182 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smax.v16i64(<16 x i64> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i64'
@@ -22,10 +22,10 @@
 ;
 ; MVE-LABEL: 'reduce_i64'
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.smax.v1i64(<1 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smax.v2i64(<2 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smax.v4i64(<4 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 282 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smax.v8i64(<8 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 970 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smax.v16i64(<16 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smax.v2i64(<2 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smax.v4i64(<4 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 298 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smax.v8i64(<8 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 1002 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smax.v16i64(<16 x i64> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
   %V1  = call i64 @llvm.experimental.vector.reduce.smax.v1i64(<1 x i64> undef)
@@ -38,11 +38,11 @@
 
 define i32 @reduce_i32(i32 %arg) {
 ; V8M-LABEL: 'reduce_i32'
-; V8M-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smax.v2i32(<2 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.smax.v4i32(<4 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.smax.v8i32(<8 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.smax.v16i32(<16 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 187 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.smax.v32i32(<32 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smax.v2i32(<2 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.smax.v4i32(<4 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.smax.v8i32(<8 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.smax.v16i32(<16 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 218 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.smax.v32i32(<32 x i32> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i32'
@@ -54,7 +54,7 @@
 ; NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-LABEL: 'reduce_i32'
-; MVE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smax.v2i32(<2 x i32> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smax.v2i32(<2 x i32> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.smax.v4i32(<4 x i32> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 240 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.smax.v8i32(<8 x i32> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 632 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.smax.v16i32(<16 x i32> undef)
@@ -71,12 +71,12 @@
 
 define i32 @reduce_i16(i32 %arg) {
 ; V8M-LABEL: 'reduce_i16'
-; V8M-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.smax.v2i16(<2 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smax.v4i16(<4 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smax.v8i16(<8 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smax.v16i16(<16 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 187 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smax.v32i16(<32 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 379 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smax.v64i16(<64 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.smax.v2i16(<2 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smax.v4i16(<4 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smax.v8i16(<8 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smax.v16i16(<16 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 218 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smax.v32i16(<32 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 442 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smax.v64i16(<64 x i16> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i16'
@@ -89,7 +89,7 @@
 ; NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-LABEL: 'reduce_i16'
-; MVE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.smax.v2i16(<2 x i16> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.smax.v2i16(<2 x i16> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smax.v4i16(<4 x i16> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 788 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smax.v8i16(<8 x i16> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 1176 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smax.v16i16(<16 x i16> undef)
@@ -108,13 +108,13 @@
 
 define i32 @reduce_i8(i32 %arg) {
 ; V8M-LABEL: 'reduce_i8'
-; V8M-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.smax.v2i8(<2 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i8 @llvm.experimental.vector.reduce.smax.v4i8(<4 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smax.v8i8(<8 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.v16i8(<16 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 187 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smax.v32i8(<32 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 379 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.v64i8(<64 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 763 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.v128i8(<128 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.smax.v2i8(<2 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i8 @llvm.experimental.vector.reduce.smax.v4i8(<4 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smax.v8i8(<8 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.v16i8(<16 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 218 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smax.v32i8(<32 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 442 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smax.v64i8(<64 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 890 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smax.v128i8(<128 x i8> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i8'
@@ -128,7 +128,7 @@
 ; NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-LABEL: 'reduce_i8'
-; MVE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.smax.v2i8(<2 x i8> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.smax.v2i8(<2 x i8> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4 = call i8 @llvm.experimental.vector.reduce.smax.v4i8(<4 x i8> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 788 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smax.v8i8(<8 x i8> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 4128 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smax.v16i8(<16 x i8> undef)
Index: llvm/test/Analysis/CostModel/ARM/reduce-smin.ll
===================================================================
--- llvm/test/Analysis/CostModel/ARM/reduce-smin.ll
+++ llvm/test/Analysis/CostModel/ARM/reduce-smin.ll
@@ -6,10 +6,10 @@
 define i32 @reduce_i64(i32 %arg) {
 ; V8M-LABEL: 'reduce_i64'
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.smin.v1i64(<1 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smin.v2i64(<2 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smin.v4i64(<4 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smin.v8i64(<8 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 167 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smin.v16i64(<16 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smin.v2i64(<2 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smin.v4i64(<4 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smin.v8i64(<8 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 182 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smin.v16i64(<16 x i64> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i64'
@@ -22,10 +22,10 @@
 ;
 ; MVE-LABEL: 'reduce_i64'
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.smin.v1i64(<1 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smin.v2i64(<2 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smin.v4i64(<4 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 282 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smin.v8i64(<8 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 970 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smin.v16i64(<16 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.smin.v2i64(<2 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.smin.v4i64(<4 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 298 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.smin.v8i64(<8 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 1002 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.smin.v16i64(<16 x i64> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
   %V1  = call i64 @llvm.experimental.vector.reduce.smin.v1i64(<1 x i64> undef)
@@ -38,11 +38,11 @@
 
 define i32 @reduce_i32(i32 %arg) {
 ; V8M-LABEL: 'reduce_i32'
-; V8M-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smin.v2i32(<2 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.smin.v4i32(<4 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.smin.v8i32(<8 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.smin.v16i32(<16 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 187 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.smin.v32i32(<32 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smin.v2i32(<2 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.smin.v4i32(<4 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.smin.v8i32(<8 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.smin.v16i32(<16 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 218 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.smin.v32i32(<32 x i32> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i32'
@@ -54,7 +54,7 @@
 ; NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-LABEL: 'reduce_i32'
-; MVE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smin.v2i32(<2 x i32> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.smin.v2i32(<2 x i32> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.smin.v4i32(<4 x i32> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 240 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.smin.v8i32(<8 x i32> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 632 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.smin.v16i32(<16 x i32> undef)
@@ -71,12 +71,12 @@
 
 define i32 @reduce_i16(i32 %arg) {
 ; V8M-LABEL: 'reduce_i16'
-; V8M-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.smin.v2i16(<2 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smin.v4i16(<4 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smin.v8i16(<8 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smin.v16i16(<16 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 187 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smin.v32i16(<32 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 379 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smin.v64i16(<64 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.smin.v2i16(<2 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smin.v4i16(<4 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smin.v8i16(<8 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smin.v16i16(<16 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 218 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.smin.v32i16(<32 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 442 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.smin.v64i16(<64 x i16> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i16'
@@ -89,7 +89,7 @@
 ; NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-LABEL: 'reduce_i16'
-; MVE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.smin.v2i16(<2 x i16> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.smin.v2i16(<2 x i16> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.smin.v4i16(<4 x i16> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 788 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.smin.v8i16(<8 x i16> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 1176 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.smin.v16i16(<16 x i16> undef)
@@ -108,13 +108,13 @@
 
 define i32 @reduce_i8(i32 %arg) {
 ; V8M-LABEL: 'reduce_i8'
-; V8M-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.smin.v2i8(<2 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i8 @llvm.experimental.vector.reduce.smin.v4i8(<4 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smin.v8i8(<8 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.v16i8(<16 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 187 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smin.v32i8(<32 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 379 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.v64i8(<64 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 763 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.v128i8(<128 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.smin.v2i8(<2 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i8 @llvm.experimental.vector.reduce.smin.v4i8(<4 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smin.v8i8(<8 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.v16i8(<16 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 218 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.smin.v32i8(<32 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 442 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.smin.v64i8(<64 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 890 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.smin.v128i8(<128 x i8> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i8'
@@ -128,7 +128,7 @@
 ; NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-LABEL: 'reduce_i8'
-; MVE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.smin.v2i8(<2 x i8> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.smin.v2i8(<2 x i8> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4 = call i8 @llvm.experimental.vector.reduce.smin.v4i8(<4 x i8> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 788 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.smin.v8i8(<8 x i8> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 4128 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.smin.v16i8(<16 x i8> undef)
Index: llvm/test/Analysis/CostModel/ARM/reduce-umax.ll
===================================================================
--- llvm/test/Analysis/CostModel/ARM/reduce-umax.ll
+++ llvm/test/Analysis/CostModel/ARM/reduce-umax.ll
@@ -6,10 +6,10 @@
 define i32 @reduce_i64(i32 %arg) {
 ; V8M-LABEL: 'reduce_i64'
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.umax.v1i64(<1 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umax.v2i64(<2 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umax.v4i64(<4 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umax.v8i64(<8 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 167 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umax.v16i64(<16 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umax.v2i64(<2 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umax.v4i64(<4 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umax.v8i64(<8 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 182 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umax.v16i64(<16 x i64> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i64'
@@ -22,10 +22,10 @@
 ;
 ; MVE-LABEL: 'reduce_i64'
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.umax.v1i64(<1 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umax.v2i64(<2 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umax.v4i64(<4 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 282 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umax.v8i64(<8 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 970 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umax.v16i64(<16 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umax.v2i64(<2 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umax.v4i64(<4 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 298 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umax.v8i64(<8 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 1002 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umax.v16i64(<16 x i64> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
   %V1  = call i64 @llvm.experimental.vector.reduce.umax.v1i64(<1 x i64> undef)
@@ -38,11 +38,11 @@
 
 define i32 @reduce_i32(i32 %arg) {
 ; V8M-LABEL: 'reduce_i32'
-; V8M-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umax.v2i32(<2 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.umax.v4i32(<4 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.umax.v8i32(<8 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.umax.v16i32(<16 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 187 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.umax.v32i32(<32 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umax.v2i32(<2 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.umax.v4i32(<4 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.umax.v8i32(<8 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.umax.v16i32(<16 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 218 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.umax.v32i32(<32 x i32> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i32'
@@ -54,7 +54,7 @@
 ; NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-LABEL: 'reduce_i32'
-; MVE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umax.v2i32(<2 x i32> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umax.v2i32(<2 x i32> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.umax.v4i32(<4 x i32> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 240 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.umax.v8i32(<8 x i32> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 632 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.umax.v16i32(<16 x i32> undef)
@@ -71,12 +71,12 @@
 
 define i32 @reduce_i16(i32 %arg) {
 ; V8M-LABEL: 'reduce_i16'
-; V8M-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.umax.v2i16(<2 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umax.v4i16(<4 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umax.v8i16(<8 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umax.v16i16(<16 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 187 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umax.v32i16(<32 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 379 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umax.v64i16(<64 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.umax.v2i16(<2 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umax.v4i16(<4 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umax.v8i16(<8 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umax.v16i16(<16 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 218 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umax.v32i16(<32 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 442 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umax.v64i16(<64 x i16> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i16'
@@ -89,7 +89,7 @@
 ; NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-LABEL: 'reduce_i16'
-; MVE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.umax.v2i16(<2 x i16> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.umax.v2i16(<2 x i16> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umax.v4i16(<4 x i16> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 788 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umax.v8i16(<8 x i16> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 1176 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umax.v16i16(<16 x i16> undef)
@@ -108,13 +108,13 @@
 
 define i32 @reduce_i8(i32 %arg) {
 ; V8M-LABEL: 'reduce_i8'
-; V8M-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.umax.v2i8(<2 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i8 @llvm.experimental.vector.reduce.umax.v4i8(<4 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umax.v8i8(<8 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.v16i8(<16 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 187 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umax.v32i8(<32 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 379 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.v64i8(<64 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 763 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.v128i8(<128 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.umax.v2i8(<2 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i8 @llvm.experimental.vector.reduce.umax.v4i8(<4 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umax.v8i8(<8 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.v16i8(<16 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 218 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umax.v32i8(<32 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 442 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umax.v64i8(<64 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 890 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umax.v128i8(<128 x i8> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i8'
@@ -128,7 +128,7 @@
 ; NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-LABEL: 'reduce_i8'
-; MVE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.umax.v2i8(<2 x i8> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.umax.v2i8(<2 x i8> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4 = call i8 @llvm.experimental.vector.reduce.umax.v4i8(<4 x i8> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 788 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umax.v8i8(<8 x i8> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 4128 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umax.v16i8(<16 x i8> undef)
Index: llvm/test/Analysis/CostModel/ARM/reduce-umin.ll
===================================================================
--- llvm/test/Analysis/CostModel/ARM/reduce-umin.ll
+++ llvm/test/Analysis/CostModel/ARM/reduce-umin.ll
@@ -6,10 +6,10 @@
 define i32 @reduce_i64(i32 %arg) {
 ; V8M-LABEL: 'reduce_i64'
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.umin.v1i64(<1 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 13 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umin.v2i64(<2 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 35 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umin.v4i64(<4 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 79 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umin.v8i64(<8 x i64> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 167 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umin.v16i64(<16 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 14 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umin.v2i64(<2 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 38 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umin.v4i64(<4 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 86 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umin.v8i64(<8 x i64> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 182 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umin.v16i64(<16 x i64> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i64'
@@ -22,10 +22,10 @@
 ;
 ; MVE-LABEL: 'reduce_i64'
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %V1 = call i64 @llvm.experimental.vector.reduce.umin.v1i64(<1 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umin.v2i64(<2 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 98 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umin.v4i64(<4 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 282 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umin.v8i64(<8 x i64> undef)
-; MVE-NEXT:  Cost Model: Found an estimated cost of 970 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umin.v16i64(<16 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V2 = call i64 @llvm.experimental.vector.reduce.umin.v2i64(<2 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V4 = call i64 @llvm.experimental.vector.reduce.umin.v4i64(<4 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 298 for instruction: %V8 = call i64 @llvm.experimental.vector.reduce.umin.v8i64(<8 x i64> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 1002 for instruction: %V16 = call i64 @llvm.experimental.vector.reduce.umin.v16i64(<16 x i64> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
   %V1  = call i64 @llvm.experimental.vector.reduce.umin.v1i64(<1 x i64> undef)
@@ -38,11 +38,11 @@
 
 define i32 @reduce_i32(i32 %arg) {
 ; V8M-LABEL: 'reduce_i32'
-; V8M-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umin.v2i32(<2 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.umin.v4i32(<4 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.umin.v8i32(<8 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.umin.v16i32(<16 x i32> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 187 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.umin.v32i32(<32 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umin.v2i32(<2 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.umin.v4i32(<4 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.umin.v8i32(<8 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.umin.v16i32(<16 x i32> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 218 for instruction: %V32 = call i32 @llvm.experimental.vector.reduce.umin.v32i32(<32 x i32> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i32'
@@ -54,7 +54,7 @@
 ; NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-LABEL: 'reduce_i32'
-; MVE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umin.v2i32(<2 x i32> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i32 @llvm.experimental.vector.reduce.umin.v2i32(<2 x i32> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4 = call i32 @llvm.experimental.vector.reduce.umin.v4i32(<4 x i32> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 240 for instruction: %V8 = call i32 @llvm.experimental.vector.reduce.umin.v8i32(<8 x i32> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 632 for instruction: %V16 = call i32 @llvm.experimental.vector.reduce.umin.v16i32(<16 x i32> undef)
@@ -71,12 +71,12 @@
 
 define i32 @reduce_i16(i32 %arg) {
 ; V8M-LABEL: 'reduce_i16'
-; V8M-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.umin.v2i16(<2 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umin.v4i16(<4 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umin.v8i16(<8 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umin.v16i16(<16 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 187 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umin.v32i16(<32 x i16> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 379 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umin.v64i16(<64 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.umin.v2i16(<2 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umin.v4i16(<4 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umin.v8i16(<8 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umin.v16i16(<16 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 218 for instruction: %V32 = call i16 @llvm.experimental.vector.reduce.umin.v32i16(<32 x i16> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 442 for instruction: %V64 = call i16 @llvm.experimental.vector.reduce.umin.v64i16(<64 x i16> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i16'
@@ -89,7 +89,7 @@
 ; NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-LABEL: 'reduce_i16'
-; MVE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.umin.v2i16(<2 x i16> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i16 @llvm.experimental.vector.reduce.umin.v2i16(<2 x i16> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4 = call i16 @llvm.experimental.vector.reduce.umin.v4i16(<4 x i16> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 788 for instruction: %V8 = call i16 @llvm.experimental.vector.reduce.umin.v8i16(<8 x i16> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 1176 for instruction: %V16 = call i16 @llvm.experimental.vector.reduce.umin.v16i16(<16 x i16> undef)
@@ -108,13 +108,13 @@
 
 define i32 @reduce_i8(i32 %arg) {
 ; V8M-LABEL: 'reduce_i8'
-; V8M-NEXT:  Cost Model: Found an estimated cost of 7 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.umin.v2i8(<2 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %V4 = call i8 @llvm.experimental.vector.reduce.umin.v4i8(<4 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 43 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umin.v8i8(<8 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 91 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.v16i8(<16 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 187 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umin.v32i8(<32 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 379 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.v64i8(<64 x i8> undef)
-; V8M-NEXT:  Cost Model: Found an estimated cost of 763 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.v128i8(<128 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.umin.v2i8(<2 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 22 for instruction: %V4 = call i8 @llvm.experimental.vector.reduce.umin.v4i8(<4 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umin.v8i8(<8 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 106 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.v16i8(<16 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 218 for instruction: %V32 = call i8 @llvm.experimental.vector.reduce.umin.v32i8(<32 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 442 for instruction: %V64 = call i8 @llvm.experimental.vector.reduce.umin.v64i8(<64 x i8> undef)
+; V8M-NEXT:  Cost Model: Found an estimated cost of 890 for instruction: %V128 = call i8 @llvm.experimental.vector.reduce.umin.v128i8(<128 x i8> undef)
 ; V8M-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; NEON-LABEL: 'reduce_i8'
@@ -128,7 +128,7 @@
 ; NEON-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret i32 undef
 ;
 ; MVE-LABEL: 'reduce_i8'
-; MVE-NEXT:  Cost Model: Found an estimated cost of 42 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.umin.v2i8(<2 x i8> undef)
+; MVE-NEXT:  Cost Model: Found an estimated cost of 46 for instruction: %V2 = call i8 @llvm.experimental.vector.reduce.umin.v2i8(<2 x i8> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 140 for instruction: %V4 = call i8 @llvm.experimental.vector.reduce.umin.v4i8(<4 x i8> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 788 for instruction: %V8 = call i8 @llvm.experimental.vector.reduce.umin.v8i8(<8 x i8> undef)
 ; MVE-NEXT:  Cost Model: Found an estimated cost of 4128 for instruction: %V16 = call i8 @llvm.experimental.vector.reduce.umin.v16i8(<16 x i8> undef)
Index: llvm/test/Analysis/CostModel/ARM/select.ll
===================================================================
--- llvm/test/Analysis/CostModel/ARM/select.ll
+++ llvm/test/Analysis/CostModel/ARM/select.ll
@@ -11,34 +11,35 @@
 define void @selects() {
   ; Scalar values
 ; CHECK-MVE-RECIP-LABEL: 'selects'
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v1 = select i1 undef, i8 undef, i8 undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2 = select i1 undef, i16 undef, i16 undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v3 = select i1 undef, i32 undef, i32 undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4 = select i1 undef, i64 undef, i64 undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v5 = select i1 undef, float undef, float undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v6 = select i1 undef, double undef, double undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v7 = select <2 x i1> undef, <2 x i8> undef, <2 x i8> undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v1 = select i1 undef, i8 undef, i8 undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2 = select i1 undef, i16 undef, i16 undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v3 = select i1 undef, i32 undef, i32 undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4 = select i1 undef, i64 undef, i64 undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v5 = select i1 undef, float undef, float undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v6 = select i1 undef, double undef, double undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v7 = select <2 x i1> undef, <2 x i8> undef, <2 x i8> undef
 ; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8 = select <4 x i1> undef, <4 x i8> undef, <4 x i8> undef
 ; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v9 = select <8 x i1> undef, <8 x i8> undef, <8 x i8> undef
 ; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v10 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v11 = select <2 x i1> undef, <2 x i16> undef, <2 x i16> undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v11 = select <2 x i1> undef, <2 x i16> undef, <2 x i16> undef
 ; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v12 = select <4 x i1> undef, <4 x i16> undef, <4 x i16> undef
 ; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v13 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
 ; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v13b = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v14 = select <2 x i1> undef, <2 x i32> undef, <2 x i32> undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v14 = select <2 x i1> undef, <2 x i32> undef, <2 x i32> undef
 ; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v15 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
 ; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v15b = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
 ; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v15c = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v16 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v16a = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 160 for instruction: %v16b = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 576 for instruction: %v16c = select <16 x i1> undef, <16 x i64> undef, <16 x i64> undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v16 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 56 for instruction: %v16a = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 176 for instruction: %v16b = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 608 for instruction: %v16c = select <16 x i1> undef, <16 x i64> undef, <16 x i64> undef
 ; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v17 = select <2 x i1> undef, <2 x float> undef, <2 x float> undef
 ; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v18 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v19 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v20 = select <1 x i1> undef, <1 x i32> undef, <1 x i32> undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v19 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v20 = select <1 x i1> undef, <1 x i32> undef, <1 x i32> undef
 ; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v21 = select <3 x i1> undef, <3 x float> undef, <3 x float> undef
-; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 60 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 70 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-MVE-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-NEON-RECIP-LABEL: 'selects'
 ; CHECK-NEON-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v1 = select i1 undef, i8 undef, i8 undef
@@ -69,96 +70,100 @@
 ; CHECK-NEON-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v20 = select <1 x i1> undef, <1 x i32> undef, <1 x i32> undef
 ; CHECK-NEON-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v21 = select <3 x i1> undef, <3 x float> undef, <3 x float> undef
 ; CHECK-NEON-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-NEON-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-THUMB1-RECIP-LABEL: 'selects'
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v1 = select i1 undef, i8 undef, i8 undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2 = select i1 undef, i16 undef, i16 undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v3 = select i1 undef, i32 undef, i32 undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4 = select i1 undef, i64 undef, i64 undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v5 = select i1 undef, float undef, float undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v6 = select i1 undef, double undef, double undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v7 = select <2 x i1> undef, <2 x i8> undef, <2 x i8> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8 = select <4 x i1> undef, <4 x i8> undef, <4 x i8> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v9 = select <8 x i1> undef, <8 x i8> undef, <8 x i8> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %v10 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v11 = select <2 x i1> undef, <2 x i16> undef, <2 x i16> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v12 = select <4 x i1> undef, <4 x i16> undef, <4 x i16> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v13 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %v13b = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v14 = select <2 x i1> undef, <2 x i32> undef, <2 x i32> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v15 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v15b = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %v15c = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v16 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v16a = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %v16b = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %v16c = select <16 x i1> undef, <16 x i64> undef, <16 x i64> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v17 = select <2 x i1> undef, <2 x float> undef, <2 x float> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v18 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v19 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v20 = select <1 x i1> undef, <1 x i32> undef, <1 x i32> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v21 = select <3 x i1> undef, <3 x float> undef, <3 x float> undef
-; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v1 = select i1 undef, i8 undef, i8 undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2 = select i1 undef, i16 undef, i16 undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v3 = select i1 undef, i32 undef, i32 undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4 = select i1 undef, i64 undef, i64 undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v5 = select i1 undef, float undef, float undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v6 = select i1 undef, double undef, double undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v7 = select <2 x i1> undef, <2 x i8> undef, <2 x i8> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v8 = select <4 x i1> undef, <4 x i8> undef, <4 x i8> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v9 = select <8 x i1> undef, <8 x i8> undef, <8 x i8> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v10 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v11 = select <2 x i1> undef, <2 x i16> undef, <2 x i16> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v12 = select <4 x i1> undef, <4 x i16> undef, <4 x i16> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v13 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v13b = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v14 = select <2 x i1> undef, <2 x i32> undef, <2 x i32> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v15 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v15b = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v15c = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v16 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v16a = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %v16b = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %v16c = select <16 x i1> undef, <16 x i64> undef, <16 x i64> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v17 = select <2 x i1> undef, <2 x float> undef, <2 x float> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v18 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v19 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v20 = select <1 x i1> undef, <1 x i32> undef, <1 x i32> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v21 = select <3 x i1> undef, <3 x float> undef, <3 x float> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-THUMB1-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-THUMB2-RECIP-LABEL: 'selects'
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v1 = select i1 undef, i8 undef, i8 undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2 = select i1 undef, i16 undef, i16 undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v3 = select i1 undef, i32 undef, i32 undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v4 = select i1 undef, i64 undef, i64 undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v5 = select i1 undef, float undef, float undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v6 = select i1 undef, double undef, double undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v7 = select <2 x i1> undef, <2 x i8> undef, <2 x i8> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v8 = select <4 x i1> undef, <4 x i8> undef, <4 x i8> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v9 = select <8 x i1> undef, <8 x i8> undef, <8 x i8> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %v10 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v11 = select <2 x i1> undef, <2 x i16> undef, <2 x i16> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v12 = select <4 x i1> undef, <4 x i16> undef, <4 x i16> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v13 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %v13b = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v14 = select <2 x i1> undef, <2 x i32> undef, <2 x i32> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v15 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v15b = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %v15c = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v16 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 16 for instruction: %v16a = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 32 for instruction: %v16b = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 64 for instruction: %v16c = select <16 x i1> undef, <16 x i64> undef, <16 x i64> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v17 = select <2 x i1> undef, <2 x float> undef, <2 x float> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v18 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 8 for instruction: %v19 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v20 = select <1 x i1> undef, <1 x i32> undef, <1 x i32> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v21 = select <3 x i1> undef, <3 x float> undef, <3 x float> undef
-; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v1 = select i1 undef, i8 undef, i8 undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2 = select i1 undef, i16 undef, i16 undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v3 = select i1 undef, i32 undef, i32 undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4 = select i1 undef, i64 undef, i64 undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v5 = select i1 undef, float undef, float undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v6 = select i1 undef, double undef, double undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v7 = select <2 x i1> undef, <2 x i8> undef, <2 x i8> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v8 = select <4 x i1> undef, <4 x i8> undef, <4 x i8> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v9 = select <8 x i1> undef, <8 x i8> undef, <8 x i8> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v10 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v11 = select <2 x i1> undef, <2 x i16> undef, <2 x i16> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v12 = select <4 x i1> undef, <4 x i16> undef, <4 x i16> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v13 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v13b = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v14 = select <2 x i1> undef, <2 x i32> undef, <2 x i32> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v15 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 24 for instruction: %v15b = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 48 for instruction: %v15c = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v16 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 20 for instruction: %v16a = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 40 for instruction: %v16b = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 80 for instruction: %v16c = select <16 x i1> undef, <16 x i64> undef, <16 x i64> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 6 for instruction: %v17 = select <2 x i1> undef, <2 x float> undef, <2 x float> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 12 for instruction: %v18 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 10 for instruction: %v19 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v20 = select <1 x i1> undef, <1 x i32> undef, <1 x i32> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v21 = select <3 x i1> undef, <3 x float> undef, <3 x float> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 25 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-THUMB2-RECIP-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-MVE-SIZE-LABEL: 'selects'
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v1 = select i1 undef, i8 undef, i8 undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2 = select i1 undef, i16 undef, i16 undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v3 = select i1 undef, i32 undef, i32 undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4 = select i1 undef, i64 undef, i64 undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v5 = select i1 undef, float undef, float undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v6 = select i1 undef, double undef, double undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v7 = select <2 x i1> undef, <2 x i8> undef, <2 x i8> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8 = select <4 x i1> undef, <4 x i8> undef, <4 x i8> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v9 = select <8 x i1> undef, <8 x i8> undef, <8 x i8> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v10 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v11 = select <2 x i1> undef, <2 x i16> undef, <2 x i16> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v12 = select <4 x i1> undef, <4 x i16> undef, <4 x i16> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v13 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v13b = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v14 = select <2 x i1> undef, <2 x i32> undef, <2 x i32> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v15 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v15b = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v15c = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16a = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16b = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16c = select <16 x i1> undef, <16 x i64> undef, <16 x i64> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v17 = select <2 x i1> undef, <2 x float> undef, <2 x float> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v18 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v19 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v20 = select <1 x i1> undef, <1 x i32> undef, <1 x i32> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v21 = select <3 x i1> undef, <3 x float> undef, <3 x float> undef
-; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v1 = select i1 undef, i8 undef, i8 undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2 = select i1 undef, i16 undef, i16 undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v3 = select i1 undef, i32 undef, i32 undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4 = select i1 undef, i64 undef, i64 undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v5 = select i1 undef, float undef, float undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v6 = select i1 undef, double undef, double undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v7 = select <2 x i1> undef, <2 x i8> undef, <2 x i8> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v8 = select <4 x i1> undef, <4 x i8> undef, <4 x i8> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v9 = select <8 x i1> undef, <8 x i8> undef, <8 x i8> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v10 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v11 = select <2 x i1> undef, <2 x i16> undef, <2 x i16> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v12 = select <4 x i1> undef, <4 x i16> undef, <4 x i16> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v13 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v13b = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v14 = select <2 x i1> undef, <2 x i32> undef, <2 x i32> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v15 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v15b = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v15c = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v16 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v16a = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v16b = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16c = select <16 x i1> undef, <16 x i64> undef, <16 x i64> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v17 = select <2 x i1> undef, <2 x float> undef, <2 x float> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v18 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v19 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v20 = select <1 x i1> undef, <1 x i32> undef, <1 x i32> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v21 = select <3 x i1> undef, <3 x float> undef, <3 x float> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-MVE-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-NEON-SIZE-LABEL: 'selects'
 ; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v1 = select i1 undef, i8 undef, i8 undef
@@ -174,81 +179,84 @@
 ; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v11 = select <2 x i1> undef, <2 x i16> undef, <2 x i16> undef
 ; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v12 = select <4 x i1> undef, <4 x i16> undef, <4 x i16> undef
 ; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v13 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
-; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v13b = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
+; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v13b = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
 ; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v14 = select <2 x i1> undef, <2 x i32> undef, <2 x i32> undef
 ; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v15 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
-; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v15b = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
-; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v15c = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
+; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v15b = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
+; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v15c = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
 ; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
-; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16a = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
-; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16b = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
-; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16c = select <16 x i1> undef, <16 x i64> undef, <16 x i64> undef
+; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 19 for instruction: %v16a = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
+; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 50 for instruction: %v16b = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
+; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 100 for instruction: %v16c = select <16 x i1> undef, <16 x i64> undef, <16 x i64> undef
 ; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v17 = select <2 x i1> undef, <2 x float> undef, <2 x float> undef
 ; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v18 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
 ; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v19 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
 ; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v20 = select <1 x i1> undef, <1 x i32> undef, <1 x i32> undef
 ; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v21 = select <3 x i1> undef, <3 x float> undef, <3 x float> undef
-; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 4 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-NEON-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-THUMB1-SIZE-LABEL: 'selects'
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v1 = select i1 undef, i8 undef, i8 undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2 = select i1 undef, i16 undef, i16 undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v3 = select i1 undef, i32 undef, i32 undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4 = select i1 undef, i64 undef, i64 undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v5 = select i1 undef, float undef, float undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v6 = select i1 undef, double undef, double undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v7 = select <2 x i1> undef, <2 x i8> undef, <2 x i8> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8 = select <4 x i1> undef, <4 x i8> undef, <4 x i8> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v9 = select <8 x i1> undef, <8 x i8> undef, <8 x i8> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v10 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v11 = select <2 x i1> undef, <2 x i16> undef, <2 x i16> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v12 = select <4 x i1> undef, <4 x i16> undef, <4 x i16> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v13 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v13b = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v14 = select <2 x i1> undef, <2 x i32> undef, <2 x i32> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v15 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v15b = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v15c = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16a = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16b = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16c = select <16 x i1> undef, <16 x i64> undef, <16 x i64> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v17 = select <2 x i1> undef, <2 x float> undef, <2 x float> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v18 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v19 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v20 = select <1 x i1> undef, <1 x i32> undef, <1 x i32> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v21 = select <3 x i1> undef, <3 x float> undef, <3 x float> undef
-; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v1 = select i1 undef, i8 undef, i8 undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2 = select i1 undef, i16 undef, i16 undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v3 = select i1 undef, i32 undef, i32 undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4 = select i1 undef, i64 undef, i64 undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v5 = select i1 undef, float undef, float undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v6 = select i1 undef, double undef, double undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v7 = select <2 x i1> undef, <2 x i8> undef, <2 x i8> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8 = select <4 x i1> undef, <4 x i8> undef, <4 x i8> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v9 = select <8 x i1> undef, <8 x i8> undef, <8 x i8> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v10 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v11 = select <2 x i1> undef, <2 x i16> undef, <2 x i16> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v12 = select <4 x i1> undef, <4 x i16> undef, <4 x i16> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v13 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v13b = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v14 = select <2 x i1> undef, <2 x i32> undef, <2 x i32> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v15 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v15b = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v15c = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v16 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16a = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v16b = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %v16c = select <16 x i1> undef, <16 x i64> undef, <16 x i64> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v17 = select <2 x i1> undef, <2 x float> undef, <2 x float> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v18 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v19 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v20 = select <1 x i1> undef, <1 x i32> undef, <1 x i32> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v21 = select <3 x i1> undef, <3 x float> undef, <3 x float> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-THUMB1-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
 ; CHECK-THUMB2-SIZE-LABEL: 'selects'
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v1 = select i1 undef, i8 undef, i8 undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v2 = select i1 undef, i16 undef, i16 undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v3 = select i1 undef, i32 undef, i32 undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v4 = select i1 undef, i64 undef, i64 undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v5 = select i1 undef, float undef, float undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v6 = select i1 undef, double undef, double undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v7 = select <2 x i1> undef, <2 x i8> undef, <2 x i8> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v8 = select <4 x i1> undef, <4 x i8> undef, <4 x i8> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v9 = select <8 x i1> undef, <8 x i8> undef, <8 x i8> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v10 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v11 = select <2 x i1> undef, <2 x i16> undef, <2 x i16> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v12 = select <4 x i1> undef, <4 x i16> undef, <4 x i16> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v13 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v13b = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v14 = select <2 x i1> undef, <2 x i32> undef, <2 x i32> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v15 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v15b = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v15c = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16a = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16b = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v16c = select <16 x i1> undef, <16 x i64> undef, <16 x i64> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v17 = select <2 x i1> undef, <2 x float> undef, <2 x float> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v18 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v19 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v20 = select <1 x i1> undef, <1 x i32> undef, <1 x i32> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v21 = select <3 x i1> undef, <3 x float> undef, <3 x float> undef
-; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v1 = select i1 undef, i8 undef, i8 undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v2 = select i1 undef, i16 undef, i16 undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v3 = select i1 undef, i32 undef, i32 undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v4 = select i1 undef, i64 undef, i64 undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v5 = select i1 undef, float undef, float undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v6 = select i1 undef, double undef, double undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v7 = select <2 x i1> undef, <2 x i8> undef, <2 x i8> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v8 = select <4 x i1> undef, <4 x i8> undef, <4 x i8> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v9 = select <8 x i1> undef, <8 x i8> undef, <8 x i8> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v10 = select <16 x i1> undef, <16 x i8> undef, <16 x i8> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v11 = select <2 x i1> undef, <2 x i16> undef, <2 x i16> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v12 = select <4 x i1> undef, <4 x i16> undef, <4 x i16> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v13 = select <8 x i1> undef, <8 x i16> undef, <8 x i16> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v13b = select <16 x i1> undef, <16 x i16> undef, <16 x i16> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v14 = select <2 x i1> undef, <2 x i32> undef, <2 x i32> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v15 = select <4 x i1> undef, <4 x i32> undef, <4 x i32> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v15b = select <8 x i1> undef, <8 x i32> undef, <8 x i32> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v15c = select <16 x i1> undef, <16 x i32> undef, <16 x i32> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v16 = select <2 x i1> undef, <2 x i64> undef, <2 x i64> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 9 for instruction: %v16a = select <4 x i1> undef, <4 x i64> undef, <4 x i64> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v16b = select <8 x i1> undef, <8 x i64> undef, <8 x i64> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 33 for instruction: %v16c = select <16 x i1> undef, <16 x i64> undef, <16 x i64> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 3 for instruction: %v17 = select <2 x i1> undef, <2 x float> undef, <2 x float> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v18 = select <4 x i1> undef, <4 x float> undef, <4 x float> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v19 = select <2 x i1> undef, <2 x double> undef, <2 x double> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 2 for instruction: %v20 = select <1 x i1> undef, <1 x i32> undef, <1 x i32> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 5 for instruction: %v21 = select <3 x i1> undef, <3 x float> undef, <3 x float> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 17 for instruction: %v22 = select <5 x i1> undef, <5 x double> undef, <5 x double> undef
+; CHECK-THUMB2-SIZE-NEXT:  Cost Model: Found an estimated cost of 1 for instruction: ret void
 ;
   %v1 = select i1 undef, i8 undef, i8 undef
   %v2 = select i1 undef, i16 undef, i16 undef