[LV] Make cost model tests independent of VPValue numbers.

Update tests to not rely on hard-coded VPValue numbers.

Author: fhahn

llvm/test/Transforms/LoopVectorize/AArch64/fully-unrolled-cost.ll

@@ -10,10 +10,10 @@ define i64 @test(ptr %a, ptr %b) #0 {
 ; CHECK: Cost of 1 for VF 8: induction instruction   %i.iv.next = add nuw nsw i64 %i.iv, 1
 ; CHECK-NEXT: Cost of 0 for VF 8: induction instruction   %i.iv = phi i64 [ 0, %entry ], [ %i.iv.next, %for.body ]
 ; CHECK-NEXT: Cost of 1 for VF 8: exit condition instruction   %exitcond.not = icmp eq i64 %i.iv.next, 16
-; CHECK-NEXT: Cost of 0 for VF 8: EMIT vp<%2> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
+; CHECK-NEXT: Cost of 0 for VF 8: EMIT vp<{{.+}}> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
 ; CHECK: Cost for VF 8: 30
 ; CHECK-NEXT: Cost of 0 for VF 16: induction instruction   %i.iv = phi i64 [ 0, %entry ], [ %i.iv.next, %for.body ]
-; CHECK-NEXT: Cost of 0 for VF 16: EMIT vp<%2> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
+; CHECK-NEXT: Cost of 0 for VF 16: EMIT vp<{{.+}}> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
 ; CHECK: Cost for VF 16: 56
 ; CHECK: LV: Selecting VF: 16
 entry:
@@ -44,11 +44,11 @@ define i64 @test_external_iv_user(ptr %a, ptr %b) #0 {
 ; CHECK: Cost of 1 for VF 8: induction instruction   %i.iv.next = add nuw nsw i64 %i.iv, 1
 ; CHECK-NEXT: Cost of 0 for VF 8: induction instruction   %i.iv = phi i64 [ 0, %entry ], [ %i.iv.next, %for.body ]
 ; CHECK-NEXT: Cost of 1 for VF 8: exit condition instruction   %exitcond.not = icmp eq i64 %i.iv.next, 16
-; CHECK-NEXT: Cost of 0 for VF 8: EMIT vp<%2> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
+; CHECK-NEXT: Cost of 0 for VF 8: EMIT vp<{{.+}}> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
 ; CHECK: Cost for VF 8: 30
 ; CHECK-NEXT: Cost of 1 for VF 16: induction instruction   %i.iv.next = add nuw nsw i64 %i.iv, 1
 ; CHECK-NEXT: Cost of 0 for VF 16: induction instruction   %i.iv = phi i64 [ 0, %entry ], [ %i.iv.next, %for.body ]
-; CHECK-NEXT: Cost of 0 for VF 16: EMIT vp<%2> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
+; CHECK-NEXT: Cost of 0 for VF 16: EMIT vp<{{.+}}> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
 ; CHECK: Cost for VF 16: 57
 ; CHECK: LV: Selecting VF: vscale x 2
 entry:
@@ -117,10 +117,10 @@ define i1 @test_extra_cmp_user(ptr nocapture noundef %dst, ptr nocapture noundef
 ; CHECK: Cost of 4 for VF 8: induction instruction   %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
 ; CHECK-NEXT: Cost of 0 for VF 8: induction instruction   %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
 ; CHECK-NEXT: Cost of 4 for VF 8: exit condition instruction   %exitcond.not = icmp eq i64 %indvars.iv.next, 16
-; CHECK-NEXT: Cost of 0 for VF 8: EMIT vp<%3> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
+; CHECK-NEXT: Cost of 0 for VF 8: EMIT vp<{{.+}}> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
 ; CHECK: Cost for VF 8: 12
 ; CHECK-NEXT: Cost of 0 for VF 16: induction instruction   %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
-; CHECK-NEXT: Cost of 0 for VF 16: EMIT vp<%3> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
+; CHECK-NEXT: Cost of 0 for VF 16: EMIT vp<{{.+}}> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
 ; CHECK: Cost for VF 16: 4
 ; CHECK: LV: Selecting VF: 16
 entry:

llvm/test/Transforms/LoopVectorize/AArch64/masked-op-cost.ll

@@ -5,8 +5,8 @@
 target triple = "aarch64-unknown-linux-gnu"
 
 ; CHECK-COST: Checking a loop in 'fixed_width'
-; CHECK-COST: Cost of 10 for VF 2: WIDEN store vp<%6>, ir<2>, vp<%5>
-; CHECK-COST: Cost of 20 for VF 4: WIDEN store vp<%6>, ir<2>, vp<%5>
+; CHECK-COST: Cost of 10 for VF 2: WIDEN store vp<{{.+}}>, ir<2>, vp<{{.+}}>
+; CHECK-COST: Cost of 20 for VF 4: WIDEN store vp<{{.+}}>, ir<2>, vp<{{.+}}>
 ; CHECK-COST: Selecting VF: 1.
 
 ; We should decide this loop is not worth vectorising using fixed width vectors

llvm/test/Transforms/LoopVectorize/ARM/mve-icmpcost.ll

@@ -22,53 +22,53 @@ target triple = "thumbv8.1m.main-arm-none-eabi"
 ; CHECK: Cost of 1 for VF 2: induction instruction   %inc = add nuw nsw i32 %i.016, 1
 ; CHECK: Cost of 0 for VF 2: induction instruction   %i.016 = phi i32 [ 0, %for.body.lr.ph ], [ %inc, %for.inc ]
 ; CHECK: Cost of 1 for VF 2: exit condition instruction   %exitcond.not = icmp eq i32 %inc, %n
-; CHECK: Cost of 0 for VF 2: EMIT vp<%2> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
-; CHECK: Cost of 0 for VF 2: vp<%3> = SCALAR-STEPS vp<%2>, ir<1>
-; CHECK: Cost of 0 for VF 2: CLONE ir<%arrayidx> = getelementptr inbounds ir<%s>, vp<%3>
-; CHECK: Cost of 0 for VF 2: vp<%4> = vector-pointer ir<%arrayidx>
-; CHECK: Cost of 18 for VF 2: WIDEN ir<%1> = load vp<%4>
+; CHECK: Cost of 0 for VF 2: EMIT vp<{{.+}}> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
+; CHECK: Cost of 0 for VF 2: vp<{{.+}}> = SCALAR-STEPS vp<{{.+}}>, ir<1>
+; CHECK: Cost of 0 for VF 2: CLONE ir<%arrayidx> = getelementptr inbounds ir<%s>, vp<{{.+}}>
+; CHECK: Cost of 0 for VF 2: vp<{{.+}}> = vector-pointer ir<%arrayidx>
+; CHECK: Cost of 18 for VF 2: WIDEN ir<%1> = load vp<{{.+}}>
 ; CHECK: Cost of 4 for VF 2: WIDEN-CAST ir<%conv> = sext ir<%1> to i32
 ; CHECK: Cost of 20 for VF 2: WIDEN ir<%cmp2> = icmp sgt ir<%conv>, ir<%conv1>
 ; CHECK: Cost of 26 for VF 2: WIDEN ir<%conv6> = add ir<%1>, ir<%0>
-; CHECK: Cost of 0 for VF 2: CLONE ir<%arrayidx7> = getelementptr ir<%d>, vp<%3>
-; CHECK: Cost of 0 for VF 2: vp<%5> = vector-pointer ir<%arrayidx7>
-; CHECK: Cost of 16 for VF 2: WIDEN store vp<%5>, ir<%conv6>, ir<%cmp2>
-; CHECK: Cost of 0 for VF 2: EMIT vp<%index.next> = add nuw vp<%2>, vp<%0>
-; CHECK: Cost of 0 for VF 2: EMIT branch-on-count vp<%index.next>, vp<%1>
+; CHECK: Cost of 0 for VF 2: CLONE ir<%arrayidx7> = getelementptr ir<%d>, vp<{{.+}}>
+; CHECK: Cost of 0 for VF 2: vp<{{.+}}> = vector-pointer ir<%arrayidx7>
+; CHECK: Cost of 16 for VF 2: WIDEN store vp<{{.+}}>, ir<%conv6>, ir<%cmp2>
+; CHECK: Cost of 0 for VF 2: EMIT vp<%index.next> = add nuw vp<{{.+}}>, vp<{{.+}}>
+; CHECK: Cost of 0 for VF 2: EMIT branch-on-count vp<%index.next>, vp<{{.+}}>
 ; CHECK: Cost for VF 2: 86 (Estimated cost per lane: 43.
 ; CHECK: Cost of 1 for VF 4: induction instruction   %inc = add nuw nsw i32 %i.016, 1
 ; CHECK: Cost of 0 for VF 4: induction instruction   %i.016 = phi i32 [ 0, %for.body.lr.ph ], [ %inc, %for.inc ]
 ; CHECK: Cost of 1 for VF 4: exit condition instruction   %exitcond.not = icmp eq i32 %inc, %n
-; CHECK: Cost of 0 for VF 4: EMIT vp<%2> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
-; CHECK: Cost of 0 for VF 4: vp<%3> = SCALAR-STEPS vp<%2>, ir<1>
-; CHECK: Cost of 0 for VF 4: CLONE ir<%arrayidx> = getelementptr inbounds ir<%s>, vp<%3>
-; CHECK: Cost of 0 for VF 4: vp<%4> = vector-pointer ir<%arrayidx>
-; CHECK: Cost of 2 for VF 4: WIDEN ir<%1> = load vp<%4>
+; CHECK: Cost of 0 for VF 4: EMIT vp<{{.+}}> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
+; CHECK: Cost of 0 for VF 4: vp<{{.+}}> = SCALAR-STEPS vp<{{.+}}>, ir<1>
+; CHECK: Cost of 0 for VF 4: CLONE ir<%arrayidx> = getelementptr inbounds ir<%s>, vp<{{.+}}>
+; CHECK: Cost of 0 for VF 4: vp<{{.+}}> = vector-pointer ir<%arrayidx>
+; CHECK: Cost of 2 for VF 4: WIDEN ir<%1> = load vp<{{.+}}>
 ; CHECK: Cost of 0 for VF 4: WIDEN-CAST ir<%conv> = sext ir<%1> to i32
 ; CHECK: Cost of 2 for VF 4: WIDEN ir<%cmp2> = icmp sgt ir<%conv>, ir<%conv1>
 ; CHECK: Cost of 2 for VF 4: WIDEN ir<%conv6> = add ir<%1>, ir<%0>
-; CHECK: Cost of 0 for VF 4: CLONE ir<%arrayidx7> = getelementptr ir<%d>, vp<%3>
-; CHECK: Cost of 0 for VF 4: vp<%5> = vector-pointer ir<%arrayidx7>
-; CHECK: Cost of 2 for VF 4: WIDEN store vp<%5>, ir<%conv6>, ir<%cmp2>
-; CHECK: Cost of 0 for VF 4: EMIT vp<%index.next> = add nuw vp<%2>, vp<%0>
-; CHECK: Cost of 0 for VF 4: EMIT branch-on-count vp<%index.next>, vp<%1>
+; CHECK: Cost of 0 for VF 4: CLONE ir<%arrayidx7> = getelementptr ir<%d>, vp<{{.+}}>
+; CHECK: Cost of 0 for VF 4: vp<{{.+}}> = vector-pointer ir<%arrayidx7>
+; CHECK: Cost of 2 for VF 4: WIDEN store vp<{{.+}}>, ir<%conv6>, ir<%cmp2>
+; CHECK: Cost of 0 for VF 4: EMIT vp<%index.next> = add nuw vp<{{.+}}>, vp<{{.+}}>
+; CHECK: Cost of 0 for VF 4: EMIT branch-on-count vp<%index.next>, vp<{{.+}}>
 ; CHECK: Cost for VF 4: 10 (Estimated cost per lane: 2.
 ; CHECK: Cost of 1 for VF 8: induction instruction   %inc = add nuw nsw i32 %i.016, 1
 ; CHECK: Cost of 0 for VF 8: induction instruction   %i.016 = phi i32 [ 0, %for.body.lr.ph ], [ %inc, %for.inc ]
 ; CHECK: Cost of 1 for VF 8: exit condition instruction   %exitcond.not = icmp eq i32 %inc, %n
-; CHECK: Cost of 0 for VF 8: EMIT vp<%2> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
-; CHECK: Cost of 0 for VF 8: vp<%3> = SCALAR-STEPS vp<%2>, ir<1>
-; CHECK: Cost of 0 for VF 8: CLONE ir<%arrayidx> = getelementptr inbounds ir<%s>, vp<%3>
-; CHECK: Cost of 0 for VF 8: vp<%4> = vector-pointer ir<%arrayidx>
-; CHECK: Cost of 2 for VF 8: WIDEN ir<%1> = load vp<%4>
+; CHECK: Cost of 0 for VF 8: EMIT vp<{{.+}}> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
+; CHECK: Cost of 0 for VF 8: vp<{{.+}}> = SCALAR-STEPS vp<{{.+}}>, ir<1>
+; CHECK: Cost of 0 for VF 8: CLONE ir<%arrayidx> = getelementptr inbounds ir<%s>, vp<{{.+}}>
+; CHECK: Cost of 0 for VF 8: vp<{{.+}}> = vector-pointer ir<%arrayidx>
+; CHECK: Cost of 2 for VF 8: WIDEN ir<%1> = load vp<{{.+}}>
 ; CHECK: Cost of 2 for VF 8: WIDEN-CAST ir<%conv> = sext ir<%1> to i32
 ; CHECK: Cost of 36 for VF 8: WIDEN ir<%cmp2> = icmp sgt ir<%conv>, ir<%conv1>
 ; CHECK: Cost of 2 for VF 8: WIDEN ir<%conv6> = add ir<%1>, ir<%0>
-; CHECK: Cost of 0 for VF 8: CLONE ir<%arrayidx7> = getelementptr ir<%d>, vp<%3>
-; CHECK: Cost of 0 for VF 8: vp<%5> = vector-pointer ir<%arrayidx7>
-; CHECK: Cost of 2 for VF 8: WIDEN store vp<%5>, ir<%conv6>, ir<%cmp2>
-; CHECK: Cost of 0 for VF 8: EMIT vp<%index.next> = add nuw vp<%2>, vp<%0>
-; CHECK: Cost of 0 for VF 8: EMIT branch-on-count vp<%index.next>, vp<%1>
+; CHECK: Cost of 0 for VF 8: CLONE ir<%arrayidx7> = getelementptr ir<%d>, vp<{{.+}}>
+; CHECK: Cost of 0 for VF 8: vp<{{.+}}> = vector-pointer ir<%arrayidx7>
+; CHECK: Cost of 2 for VF 8: WIDEN store vp<{{.+}}>, ir<%conv6>, ir<%cmp2>
+; CHECK: Cost of 0 for VF 8: EMIT vp<%index.next> = add nuw vp<{{.+}}>, vp<{{.+}}>
+; CHECK: Cost of 0 for VF 8: EMIT branch-on-count vp<%index.next>, vp<{{.+}}>
 ; CHECK: Cost for VF 8: 46 (Estimated cost per lane: 5.
 ; CHECK: LV: Selecting VF: 4.
 define void @expensive_icmp(ptr noalias nocapture %d, ptr nocapture readonly %s, i32 %n, i16 zeroext %m) #0 {
@@ -155,7 +155,7 @@ for.inc:                                          ; preds = %for.body, %if.then
 ; CHECK: Cost of 0 for VF 2: WIDEN-CAST ir<%conv4> = trunc ir<%spec.select.i> to i8
 ; CHECK: Cost of 0 for VF 2: vp<[[VEC_PTR3:%.+]]> = vector-pointer vp<%next.gep>.1
 ; CHECK: Cost of 18 for VF 2: WIDEN store vp<[[VEC_PTR3]]>, ir<%conv4>
-; CHECK: Cost of 0 for VF 2: EMIT vp<%index.next> = add nuw vp<[[CAN_IV]]>, vp<%0>
+; CHECK: Cost of 0 for VF 2: EMIT vp<%index.next> = add nuw vp<[[CAN_IV]]>, vp<{{.+}}>
 ; CHECK: Cost of 0 for VF 2: EMIT branch-on-count vp<%index.next>, vp<{{.+}}>
 ; CHECK: Cost for VF 2: 130 (Estimated cost per lane: 65.
 ; CHECK: Cost of 1 for VF 4: induction instruction   %dec = add i32 %blkCnt.012, -1
@@ -187,7 +187,7 @@ for.inc:                                          ; preds = %for.body, %if.then
 ; CHECK: Cost of 0 for VF 4: WIDEN-CAST ir<%conv4> = trunc ir<%spec.select.i> to i8
 ; CHECK: Cost of 0 for VF 4: vp<[[VEC_PTR2:%.+]]> = vector-pointer vp<%next.gep>.1
 ; CHECK: Cost of 2 for VF 4: WIDEN store vp<[[VEC_PTR2]]>, ir<%conv4>
-; CHECK: Cost of 0 for VF 4: EMIT vp<%index.next> = add nuw vp<[[CAN_IV]]>, vp<%0>
+; CHECK: Cost of 0 for VF 4: EMIT vp<%index.next> = add nuw vp<[[CAN_IV]]>, vp<{{.+}}>
 ; CHECK: Cost of 0 for VF 4: EMIT branch-on-count vp<%index.next>, vp<{{.+}}>
 ; CHECK: Cost for VF 4: 14 (Estimated cost per lane: 3.
 ; CHECK: Cost of 1 for VF 8: induction instruction   %dec = add i32 %blkCnt.012, -1

llvm/test/Transforms/LoopVectorize/X86/reduction-small-size.ll

@@ -1,5 +1,5 @@
 ; REQUIRES: asserts
-; RUN: opt < %s -passes=loop-vectorize,dce,instcombine -mcpu=core-axv2 -force-vector-interleave=1 -debug-only=loop-vectorize -S < %s 2>&1  | FileCheck %s
+; RUN: opt -passes=loop-vectorize,dce,instcombine -mcpu=core-axv2 -force-vector-interleave=1 -debug-only=loop-vectorize -S %s 2>&1  | FileCheck %s
 
 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
 
@@ -32,24 +32,24 @@ target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
 ; CHECK: Cost of 1 for VF 2: induction instruction   %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %for.body.preheader ]
 ; CHECK: Cost of 1 for VF 2: exit condition instruction   %exitcond = icmp eq i32 %lftr.wideiv, %n
 ; CHECK: Cost of 0 for VF 2: exit condition instruction   %lftr.wideiv = trunc i64 %indvars.iv.next to i32
-; CHECK: Cost of 0 for VF 2: EMIT vp<%3> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
-; CHECK: Cost of 1 for VF 2: WIDEN-REDUCTION-PHI ir<%sum.013> = phi ir<0>, vp<%8>
-; CHECK: Cost of 0 for VF 2: vp<%4> = SCALAR-STEPS vp<%3>, ir<1>
-; CHECK: Cost of 0 for VF 2: CLONE ir<%arrayidx> = getelementptr inbounds ir<%a>, vp<%4>
-; CHECK: Cost of 0 for VF 2: vp<%5> = vector-pointer ir<%arrayidx>
-; CHECK: Cost of 1 for VF 2: WIDEN ir<%0> = load vp<%5>
+; CHECK: Cost of 0 for VF 2: EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<%index.next>
+; CHECK: Cost of 1 for VF 2: WIDEN-REDUCTION-PHI ir<%sum.013> = phi ir<0>, vp<[[EXT:%.+]]>
+; CHECK: Cost of 0 for VF 2: vp<[[STEPS:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1>
+; CHECK: Cost of 0 for VF 2: CLONE ir<%arrayidx> = getelementptr inbounds ir<%a>, vp<[[STEPS]]>
+; CHECK: Cost of 0 for VF 2: vp<[[VECP1:%.+]]> = vector-pointer ir<%arrayidx>
+; CHECK: Cost of 1 for VF 2: WIDEN ir<%0> = load vp<[[VECP1]]>
 ; CHECK: Cost of 0 for VF 2: WIDEN-CAST ir<%conv> = zext ir<%0> to i32
-; CHECK: Cost of 0 for VF 2: CLONE ir<%arrayidx2> = getelementptr inbounds ir<%b>, vp<%4>
-; CHECK: Cost of 0 for VF 2: vp<%6> = vector-pointer ir<%arrayidx2>
-; CHECK: Cost of 1 for VF 2: WIDEN ir<%1> = load vp<%6>
+; CHECK: Cost of 0 for VF 2: CLONE ir<%arrayidx2> = getelementptr inbounds ir<%b>, vp<[[STEPS]]>
+; CHECK: Cost of 0 for VF 2: vp<[[VECP2:%.+]]> = vector-pointer ir<%arrayidx2>
+; CHECK: Cost of 1 for VF 2: WIDEN ir<%1> = load vp<[[VECP2]]>
 ; CHECK: Cost of 0 for VF 2: WIDEN-CAST ir<%conv3> = zext ir<%1> to i32
 ; CHECK: Cost of 0 for VF 2: WIDEN ir<%conv4> = and ir<%sum.013>, ir<255>
 ; CHECK: Cost of 1 for VF 2: WIDEN ir<%add> = add ir<%conv>, ir<%conv4>
 ; CHECK: Cost of 1 for VF 2: WIDEN ir<%add5> = add ir<%add>, ir<%conv3>
-; CHECK: Cost of 0 for VF 2: WIDEN-CAST vp<%7> = trunc ir<%add5> to i8
-; CHECK: Cost of 0 for VF 2: WIDEN-CAST vp<%8> = zext vp<%7> to i32
-; CHECK: Cost of 0 for VF 2: EMIT vp<%index.next> = add nuw vp<%3>, vp<%0>
-; CHECK: Cost of 0 for VF 2: EMIT branch-on-count vp<%index.next>, vp<%1>
+; CHECK: Cost of 0 for VF 2: WIDEN-CAST vp<[[TRUNC:%.+]]> = trunc ir<%add5> to i8
+; CHECK: Cost of 0 for VF 2: WIDEN-CAST vp<[[EXT]]> = zext vp<[[TRUNC]]> to i32
+; CHECK: Cost of 0 for VF 2: EMIT vp<%index.next> = add nuw vp<[[CAN_IV]]>, vp<{{.+}}>
+; CHECK: Cost of 0 for VF 2: EMIT branch-on-count vp<%index.next>, vp<{{.+}}>
 ;
 define i8 @reduction_i8(ptr nocapture readonly %a, ptr nocapture readonly %b, i32 %n) {
 entry: