diff options
| author | Michael Kruse <llvm-project@meinersbur.de> | 2025-01-03 10:22:51 +0100 |
|---|---|---|
| committer | Michael Kruse <llvm-project@meinersbur.de> | 2025-01-03 10:22:51 +0100 |
| commit | 38500d63e14ce340236840f60d356cdefb56a52c (patch) | |
| tree | 17edbec446ce9b50d2f215a483b83afb293a635d /llvm/test/Transforms/LoopVectorize/ARM | |
| parent | 1a3d5daaef7a6a63448a497da3eff7fc9e23df26 (diff) | |
| parent | 27f30029741ecf023baece7b3dde1ff9011ffefc (diff) | |
Merge branch 'main' into users/meinersbur/flang_runtime_split-headersusers/meinersbur/flang_runtime_split-headers
Diffstat (limited to 'llvm/test/Transforms/LoopVectorize/ARM')
9 files changed, 298 insertions, 298 deletions
diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-gather-scatter-tailpred.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-gather-scatter-tailpred.ll index fda9ef2cf6c2..b629dfd4fce9 100644 --- a/llvm/test/Transforms/LoopVectorize/ARM/mve-gather-scatter-tailpred.ll +++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-gather-scatter-tailpred.ll @@ -353,7 +353,7 @@ define void @test_stride_loopinvar_4i32(ptr readonly %data, ptr noalias nocaptur ; CHECK: middle.block: ; CHECK-NEXT: br i1 true, label [[END:%.*]], label [[SCALAR_PH]] ; CHECK: scalar.ph: -; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ], [ 0, [[VECTOR_SCEVCHECK]] ] +; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[VECTOR_SCEVCHECK]] ], [ 0, [[ENTRY:%.*]] ] ; CHECK-NEXT: br label [[FOR_BODY:%.*]] ; CHECK: for.body: ; CHECK-NEXT: [[I_023:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-hoist-runtime-checks.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-hoist-runtime-checks.ll index c18592eec9f8..845cd08cbae1 100644 --- a/llvm/test/Transforms/LoopVectorize/ARM/mve-hoist-runtime-checks.ll +++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-hoist-runtime-checks.ll @@ -82,7 +82,7 @@ define void @diff_checks(ptr nocapture noundef writeonly %dst, ptr nocapture nou ; CHECK: middle.block: ; CHECK-NEXT: br i1 true, label [[INNER_LOOP_EXIT]], label [[SCALAR_PH]] ; CHECK: scalar.ph: -; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[OUTER_LOOP]] ], [ 0, [[VECTOR_MEMCHECK]] ] +; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[VECTOR_MEMCHECK]] ], [ 0, [[OUTER_LOOP]] ] ; CHECK-NEXT: br label [[INNER_LOOP:%.*]] ; CHECK: inner.loop: ; CHECK-NEXT: [[J_021_US:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[INC_US:%.*]], [[INNER_LOOP]] ] diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-icmpcost.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-icmpcost.ll index 57bc21093ca9..b22910316d7c 100644 --- a/llvm/test/Transforms/LoopVectorize/ARM/mve-icmpcost.ll +++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-icmpcost.ll @@ -27,7 +27,7 @@ target triple = "thumbv8.1m.main-arm-none-eabi" ; 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 4 for VF 2: WIDEN-CAST ir<%conv> = sext ir<%1> to i32 +; 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> @@ -44,7 +44,7 @@ target triple = "thumbv8.1m.main-arm-none-eabi" ; 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: WIDEN-CAST ir<%conv> = sext ir<%1> to i32 +; 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> @@ -61,7 +61,7 @@ target triple = "thumbv8.1m.main-arm-none-eabi" ; 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 2 for VF 8: WIDEN-CAST ir<%conv> = sext ir<%1> to i32 +; 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> @@ -135,28 +135,28 @@ for.inc: ; preds = %for.body, %if.then ; CHECK: Cost of 0 for VF 2: induction instruction %incdec.ptr2 = getelementptr inbounds i8, ptr %pSrcB.addr.09, i32 1 ; CHECK: Cost of 0 for VF 2: induction instruction %pSrcB.addr.09 = phi ptr [ %incdec.ptr2, %while.body ], [ %pSrcB, %while.body.preheader ] ; CHECK: Cost of 1 for VF 2: exit condition instruction %cmp.not = icmp eq i32 %dec, 0 -; 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: EMIT vp<%next.gep> = ptradd ir<%pSrcA>, vp<%3> -; CHECK: Cost of 0 for VF 2: vp<%4> = SCALAR-STEPS vp<%2>, ir<1> -; CHECK: Cost of 0 for VF 2: EMIT vp<%next.gep>.1 = ptradd ir<%pDst>, vp<%4> -; CHECK: Cost of 0 for VF 2: vp<%5> = SCALAR-STEPS vp<%2>, ir<1> -; CHECK: Cost of 0 for VF 2: EMIT vp<%next.gep>.2 = ptradd ir<%pSrcB>, vp<%5> -; CHECK: Cost of 0 for VF 2: vp<%6> = vector-pointer vp<%next.gep> -; CHECK: Cost of 18 for VF 2: WIDEN ir<%0> = load vp<%6> -; CHECK: Cost of 4 for VF 2: WIDEN-CAST ir<%conv1> = sext ir<%0> to i32 -; CHECK: Cost of 0 for VF 2: vp<%7> = vector-pointer vp<%next.gep>.2 -; CHECK: Cost of 18 for VF 2: WIDEN ir<%1> = load vp<%7> -; CHECK: Cost of 4 for VF 2: WIDEN-CAST ir<%conv3> = sext ir<%1> to i32 +; CHECK: Cost of 0 for VF 2: EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<%index.next> +; CHECK: Cost of 0 for VF 2: vp<[[STEPS1:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1> +; CHECK: Cost of 0 for VF 2: EMIT vp<%next.gep> = ptradd ir<%pSrcA>, vp<[[STEPS1]]> +; CHECK: Cost of 0 for VF 2: vp<[[STEPS2:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1> +; CHECK: Cost of 0 for VF 2: EMIT vp<%next.gep>.1 = ptradd ir<%pDst>, vp<[[STEPS2]]> +; CHECK: Cost of 0 for VF 2: vp<[[STEPS3:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1> +; CHECK: Cost of 0 for VF 2: EMIT vp<%next.gep>.2 = ptradd ir<%pSrcB>, vp<[[STEPS3]]> +; CHECK: Cost of 0 for VF 2: vp<[[VEC_PTR:%.+]]> = vector-pointer vp<%next.gep> +; CHECK: Cost of 18 for VF 2: WIDEN ir<%0> = load vp<[[VEC_PTR]]> +; CHECK: Cost of 4 for VF 2: WIDEN-CAST ir<%conv1> = sext ir<%0> to i32 +; CHECK: Cost of 0 for VF 2: vp<[[VEC_PTR2:%.+]]> = vector-pointer vp<%next.gep>.2 +; CHECK: Cost of 18 for VF 2: WIDEN ir<%1> = load vp<[[VEC_PTR2]]> +; CHECK: Cost of 4 for VF 2: WIDEN-CAST ir<%conv3> = sext ir<%1> to i32 ; CHECK: Cost of 26 for VF 2: WIDEN ir<%mul> = mul nsw ir<%conv3>, ir<%conv1> ; CHECK: Cost of 18 for VF 2: WIDEN ir<%shr> = ashr ir<%mul>, ir<7> ; CHECK: Cost of 0 for VF 2: WIDEN ir<%2> = icmp slt ir<%shr>, ir<127> ; CHECK: Cost of 22 for VF 2: WIDEN-SELECT ir<%spec.select.i> = select ir<%2>, ir<%shr>, ir<127> -; 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<%8> = vector-pointer vp<%next.gep>.1 -; CHECK: Cost of 18 for VF 2: WIDEN store vp<%8>, ir<%conv4> -; 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: 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 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 ; CHECK: Cost of 0 for VF 4: induction instruction %blkCnt.012 = phi i32 [ %dec, %while.body ], [ %blockSize, %while.body.preheader ] @@ -167,28 +167,28 @@ for.inc: ; preds = %for.body, %if.then ; CHECK: Cost of 0 for VF 4: induction instruction %incdec.ptr2 = getelementptr inbounds i8, ptr %pSrcB.addr.09, i32 1 ; CHECK: Cost of 0 for VF 4: induction instruction %pSrcB.addr.09 = phi ptr [ %incdec.ptr2, %while.body ], [ %pSrcB, %while.body.preheader ] ; CHECK: Cost of 1 for VF 4: exit condition instruction %cmp.not = icmp eq i32 %dec, 0 -; 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: EMIT vp<%next.gep> = ptradd ir<%pSrcA>, vp<%3> -; CHECK: Cost of 0 for VF 4: vp<%4> = SCALAR-STEPS vp<%2>, ir<1> -; CHECK: Cost of 0 for VF 4: EMIT vp<%next.gep>.1 = ptradd ir<%pDst>, vp<%4> -; CHECK: Cost of 0 for VF 4: vp<%5> = SCALAR-STEPS vp<%2>, ir<1> -; CHECK: Cost of 0 for VF 4: EMIT vp<%next.gep>.2 = ptradd ir<%pSrcB>, vp<%5> -; CHECK: Cost of 0 for VF 4: vp<%6> = vector-pointer vp<%next.gep> -; CHECK: Cost of 2 for VF 4: WIDEN ir<%0> = load vp<%6> -; CHECK: Cost of 0 for VF 4: WIDEN-CAST ir<%conv1> = sext ir<%0> to i32 -; CHECK: Cost of 0 for VF 4: vp<%7> = vector-pointer vp<%next.gep>.2 -; CHECK: Cost of 2 for VF 4: WIDEN ir<%1> = load vp<%7> -; CHECK: Cost of 0 for VF 4: WIDEN-CAST ir<%conv3> = sext ir<%1> to i32 +; CHECK: Cost of 0 for VF 4: EMIT vp<[[CAN_IV:%.]]> = CANONICAL-INDUCTION ir<0>, vp<%index.next> +; CHECK: Cost of 0 for VF 4: vp<[[STEPS1:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1> +; CHECK: Cost of 0 for VF 4: EMIT vp<%next.gep> = ptradd ir<%pSrcA>, vp<[[STEPS1]]> +; CHECK: Cost of 0 for VF 4: vp<[[STEPS2:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1> +; CHECK: Cost of 0 for VF 4: EMIT vp<%next.gep>.1 = ptradd ir<%pDst>, vp<[[STEPS2]]> +; CHECK: Cost of 0 for VF 4: vp<[[STEPS3:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1> +; CHECK: Cost of 0 for VF 4: EMIT vp<%next.gep>.2 = ptradd ir<%pSrcB>, vp<[[STEPS3]]> +; CHECK: Cost of 0 for VF 4: vp<[[VEC_PTR1:%.+]]> = vector-pointer vp<%next.gep> +; CHECK: Cost of 2 for VF 4: WIDEN ir<%0> = load vp<[[VEC_PTR1]]> +; CHECK: Cost of 0 for VF 4: WIDEN-CAST ir<%conv1> = sext ir<%0> to i32 +; CHECK: Cost of 0 for VF 4: vp<[[VEC_PTR2:%.+]]> = vector-pointer vp<%next.gep>.2 +; CHECK: Cost of 2 for VF 4: WIDEN ir<%1> = load vp<[[VEC_PTR2]]> +; CHECK: Cost of 0 for VF 4: WIDEN-CAST ir<%conv3> = sext ir<%1> to i32 ; CHECK: Cost of 2 for VF 4: WIDEN ir<%mul> = mul nsw ir<%conv3>, ir<%conv1> ; CHECK: Cost of 2 for VF 4: WIDEN ir<%shr> = ashr ir<%mul>, ir<7> ; CHECK: Cost of 0 for VF 4: WIDEN ir<%2> = icmp slt ir<%shr>, ir<127> ; CHECK: Cost of 2 for VF 4: WIDEN-SELECT ir<%spec.select.i> = select ir<%2>, ir<%shr>, ir<127> -; 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<%8> = vector-pointer vp<%next.gep>.1 -; CHECK: Cost of 2 for VF 4: WIDEN store vp<%8>, ir<%conv4> -; 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: 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 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 ; CHECK: Cost of 0 for VF 8: induction instruction %blkCnt.012 = phi i32 [ %dec, %while.body ], [ %blockSize, %while.body.preheader ] @@ -199,28 +199,28 @@ for.inc: ; preds = %for.body, %if.then ; CHECK: Cost of 0 for VF 8: induction instruction %incdec.ptr2 = getelementptr inbounds i8, ptr %pSrcB.addr.09, i32 1 ; CHECK: Cost of 0 for VF 8: induction instruction %pSrcB.addr.09 = phi ptr [ %incdec.ptr2, %while.body ], [ %pSrcB, %while.body.preheader ] ; CHECK: Cost of 1 for VF 8: exit condition instruction %cmp.not = icmp eq i32 %dec, 0 -; 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: EMIT vp<%next.gep> = ptradd ir<%pSrcA>, vp<%3> -; CHECK: Cost of 0 for VF 8: vp<%4> = SCALAR-STEPS vp<%2>, ir<1> -; CHECK: Cost of 0 for VF 8: EMIT vp<%next.gep>.1 = ptradd ir<%pDst>, vp<%4> -; CHECK: Cost of 0 for VF 8: vp<%5> = SCALAR-STEPS vp<%2>, ir<1> -; CHECK: Cost of 0 for VF 8: EMIT vp<%next.gep>.2 = ptradd ir<%pSrcB>, vp<%5> -; CHECK: Cost of 0 for VF 8: vp<%6> = vector-pointer vp<%next.gep> -; CHECK: Cost of 2 for VF 8: WIDEN ir<%0> = load vp<%6> -; CHECK: Cost of 2 for VF 8: WIDEN-CAST ir<%conv1> = sext ir<%0> to i32 -; CHECK: Cost of 0 for VF 8: vp<%7> = vector-pointer vp<%next.gep>.2 -; CHECK: Cost of 2 for VF 8: WIDEN ir<%1> = load vp<%7> -; CHECK: Cost of 2 for VF 8: WIDEN-CAST ir<%conv3> = sext ir<%1> to i32 +; CHECK: Cost of 0 for VF 8: EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<%index.next> +; CHECK: Cost of 0 for VF 8: vp<[[STEPS1:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1> +; CHECK: Cost of 0 for VF 8: EMIT vp<%next.gep> = ptradd ir<%pSrcA>, vp<[[STEPS1]]> +; CHECK: Cost of 0 for VF 8: vp<[[STEPS2:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1> +; CHECK: Cost of 0 for VF 8: EMIT vp<%next.gep>.1 = ptradd ir<%pDst>, vp<[[STEPS2]]> +; CHECK: Cost of 0 for VF 8: vp<[[STEPS3:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1> +; CHECK: Cost of 0 for VF 8: EMIT vp<%next.gep>.2 = ptradd ir<%pSrcB>, vp<[[STEPS3]]> +; CHECK: Cost of 0 for VF 8: vp<[[VEC_PTR1:%.+]]> = vector-pointer vp<%next.gep> +; CHECK: Cost of 2 for VF 8: WIDEN ir<%0> = load vp<[[VEC_PTR1]]> +; CHECK: Cost of 2 for VF 8: WIDEN-CAST ir<%conv1> = sext ir<%0> to i32 +; CHECK: Cost of 0 for VF 8: vp<[[VEC_PTR2:%.+]]> = vector-pointer vp<%next.gep>.2 +; CHECK: Cost of 2 for VF 8: WIDEN ir<%1> = load vp<[[VEC_PTR2]]> +; CHECK: Cost of 2 for VF 8: WIDEN-CAST ir<%conv3> = sext ir<%1> to i32 ; CHECK: Cost of 4 for VF 8: WIDEN ir<%mul> = mul nsw ir<%conv3>, ir<%conv1> ; CHECK: Cost of 4 for VF 8: WIDEN ir<%shr> = ashr ir<%mul>, ir<7> ; CHECK: Cost of 0 for VF 8: WIDEN ir<%2> = icmp slt ir<%shr>, ir<127> ; CHECK: Cost of 4 for VF 8: WIDEN-SELECT ir<%spec.select.i> = select ir<%2>, ir<%shr>, ir<127> -; CHECK: Cost of 2 for VF 8: WIDEN-CAST ir<%conv4> = trunc ir<%spec.select.i> to i8 -; CHECK: Cost of 0 for VF 8: vp<%8> = vector-pointer vp<%next.gep>.1 -; CHECK: Cost of 2 for VF 8: WIDEN store vp<%8>, ir<%conv4> -; 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 2 for VF 8: WIDEN-CAST ir<%conv4> = trunc ir<%spec.select.i> to i8 +; CHECK: Cost of 0 for VF 8: vp<[[VEC_PTR3:%.+]]> = vector-pointer vp<%next.gep>.1 +; CHECK: Cost of 2 for VF 8: WIDEN store vp<[[VEC_PTR3]]>, ir<%conv4> +; CHECK: Cost of 0 for VF 8: EMIT vp<%index.next> = add nuw vp<[[CAN_IV]]>, vp<{{.+}} +; CHECK: Cost of 0 for VF 8: EMIT branch-on-count vp<%index.next>, vp<{{.+}}> ; CHECK: Cost for VF 8: 26 (Estimated cost per lane: 3. ; CHECK: Cost of 1 for VF 16: induction instruction %dec = add i32 %blkCnt.012, -1 ; CHECK: Cost of 0 for VF 16: induction instruction %blkCnt.012 = phi i32 [ %dec, %while.body ], [ %blockSize, %while.body.preheader ] @@ -231,28 +231,28 @@ for.inc: ; preds = %for.body, %if.then ; CHECK: Cost of 0 for VF 16: induction instruction %incdec.ptr2 = getelementptr inbounds i8, ptr %pSrcB.addr.09, i32 1 ; CHECK: Cost of 0 for VF 16: induction instruction %pSrcB.addr.09 = phi ptr [ %incdec.ptr2, %while.body ], [ %pSrcB, %while.body.preheader ] ; CHECK: Cost of 1 for VF 16: exit condition instruction %cmp.not = icmp eq i32 %dec, 0 -; CHECK: Cost of 0 for VF 16: EMIT vp<%2> = CANONICAL-INDUCTION ir<0>, vp<%index.next> -; CHECK: Cost of 0 for VF 16: vp<%3> = SCALAR-STEPS vp<%2>, ir<1> -; CHECK: Cost of 0 for VF 16: EMIT vp<%next.gep> = ptradd ir<%pSrcA>, vp<%3> -; CHECK: Cost of 0 for VF 16: vp<%4> = SCALAR-STEPS vp<%2>, ir<1> -; CHECK: Cost of 0 for VF 16: EMIT vp<%next.gep>.1 = ptradd ir<%pDst>, vp<%4> -; CHECK: Cost of 0 for VF 16: vp<%5> = SCALAR-STEPS vp<%2>, ir<1> -; CHECK: Cost of 0 for VF 16: EMIT vp<%next.gep>.2 = ptradd ir<%pSrcB>, vp<%5> -; CHECK: Cost of 0 for VF 16: vp<%6> = vector-pointer vp<%next.gep> -; CHECK: Cost of 2 for VF 16: WIDEN ir<%0> = load vp<%6> -; CHECK: Cost of 6 for VF 16: WIDEN-CAST ir<%conv1> = sext ir<%0> to i32 -; CHECK: Cost of 0 for VF 16: vp<%7> = vector-pointer vp<%next.gep>.2 -; CHECK: Cost of 2 for VF 16: WIDEN ir<%1> = load vp<%7> -; CHECK: Cost of 6 for VF 16: WIDEN-CAST ir<%conv3> = sext ir<%1> to i32 +; CHECK: Cost of 0 for VF 16: EMIT vp<[[CAN_IV:%.+]]> = CANONICAL-INDUCTION ir<0>, vp<%index.next> +; CHECK: Cost of 0 for VF 16: vp<[[STEPS1:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1> +; CHECK: Cost of 0 for VF 16: EMIT vp<%next.gep> = ptradd ir<%pSrcA>, vp<[[STEPS1]]> +; CHECK: Cost of 0 for VF 16: vp<[[STEPS2:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1> +; CHECK: Cost of 0 for VF 16: EMIT vp<%next.gep>.1 = ptradd ir<%pDst>, vp<[[STEPS2]]> +; CHECK: Cost of 0 for VF 16: vp<[[STEPS3:%.+]]> = SCALAR-STEPS vp<[[CAN_IV]]>, ir<1> +; CHECK: Cost of 0 for VF 16: EMIT vp<%next.gep>.2 = ptradd ir<%pSrcB>, vp<[[STEPS3]]> +; CHECK: Cost of 0 for VF 16: vp<[[VEC_PTR:%.+]]> = vector-pointer vp<%next.gep> +; CHECK: Cost of 2 for VF 16: WIDEN ir<%0> = load vp<[[VEC_PTR]]> +; CHECK: Cost of 6 for VF 16: WIDEN-CAST ir<%conv1> = sext ir<%0> to i32 +; CHECK: Cost of 0 for VF 16: vp<[[VEC_PTR1:%.+]]> = vector-pointer vp<%next.gep>.2 +; CHECK: Cost of 2 for VF 16: WIDEN ir<%1> = load vp<[[VEC_PTR1]]> +; CHECK: Cost of 6 for VF 16: WIDEN-CAST ir<%conv3> = sext ir<%1> to i32 ; CHECK: Cost of 8 for VF 16: WIDEN ir<%mul> = mul nsw ir<%conv3>, ir<%conv1> ; CHECK: Cost of 8 for VF 16: WIDEN ir<%shr> = ashr ir<%mul>, ir<7> ; CHECK: Cost of 0 for VF 16: WIDEN ir<%2> = icmp slt ir<%shr>, ir<127> ; CHECK: Cost of 8 for VF 16: WIDEN-SELECT ir<%spec.select.i> = select ir<%2>, ir<%shr>, ir<127> -; CHECK: Cost of 6 for VF 16: WIDEN-CAST ir<%conv4> = trunc ir<%spec.select.i> to i8 -; CHECK: Cost of 0 for VF 16: vp<%8> = vector-pointer vp<%next.gep>.1 -; CHECK: Cost of 2 for VF 16: WIDEN store vp<%8>, ir<%conv4> -; CHECK: Cost of 0 for VF 16: EMIT vp<%index.next> = add nuw vp<%2>, vp<%0> -; CHECK: Cost of 0 for VF 16: EMIT branch-on-count vp<%index.next>, vp<%1> +; CHECK: Cost of 6 for VF 16: WIDEN-CAST ir<%conv4> = trunc ir<%spec.select.i> to i8 +; CHECK: Cost of 0 for VF 16: vp<[[VEC_PTR2:%.+]]> = vector-pointer vp<%next.gep>.1 +; CHECK: Cost of 2 for VF 16: WIDEN store vp<[[VEC_PTR2]]>, ir<%conv4> +; CHECK: Cost of 0 for VF 16: EMIT vp<%index.next> = add nuw vp<[[CAN_IV]]>, vp<{{.+}}> +; CHECK: Cost of 0 for VF 16: EMIT branch-on-count vp<%index.next>, vp<{{.+}}> ; CHECK: Cost for VF 16: 50 ; CHECK: LV: Selecting VF: 16. define void @cheap_icmp(ptr nocapture readonly %pSrcA, ptr nocapture readonly %pSrcB, ptr nocapture %pDst, i32 %blockSize) #0 { diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-interleaved-cost.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-interleaved-cost.ll index 976c6a9a570a..551b85b7d035 100644 --- a/llvm/test/Transforms/LoopVectorize/ARM/mve-interleaved-cost.ll +++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-interleaved-cost.ll @@ -17,8 +17,8 @@ entry: ; VF_2-LABEL: Checking a loop in 'i8_factor_2' ; VF_2: Found an estimated cost of 12 for VF 2 For instruction: %tmp2 = load i8, ptr %tmp0, align 1 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp3 = load i8, ptr %tmp1, align 1 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i8 %tmp2, ptr %tmp0, align 1 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i8 %tmp3, ptr %tmp1, align 1 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i8 %tmp2, ptr %tmp0, align 1 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i8 %tmp3, ptr %tmp1, align 1 ; VF_4-LABEL: Checking a loop in 'i8_factor_2' ; VF_4: Found an estimated cost of 4 for VF 4 For instruction: %tmp2 = load i8, ptr %tmp0, align 1 ; VF_4-NEXT: Found an estimated cost of 0 for VF 4 For instruction: %tmp3 = load i8, ptr %tmp1, align 1 @@ -58,8 +58,8 @@ entry: ; VF_2-LABEL: Checking a loop in 'i16_factor_2' ; VF_2: Found an estimated cost of 12 for VF 2 For instruction: %tmp2 = load i16, ptr %tmp0, align 2 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp3 = load i16, ptr %tmp1, align 2 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i16 %tmp2, ptr %tmp0, align 2 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i16 %tmp3, ptr %tmp1, align 2 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i16 %tmp2, ptr %tmp0, align 2 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i16 %tmp3, ptr %tmp1, align 2 ; VF_4-LABEL: Checking a loop in 'i16_factor_2' ; VF_4: Found an estimated cost of 4 for VF 4 For instruction: %tmp2 = load i16, ptr %tmp0, align 2 ; VF_4-NEXT: Found an estimated cost of 0 for VF 4 For instruction: %tmp3 = load i16, ptr %tmp1, align 2 @@ -99,8 +99,8 @@ entry: ; VF_2-LABEL: Checking a loop in 'i32_factor_2' ; VF_2: Found an estimated cost of 12 for VF 2 For instruction: %tmp2 = load i32, ptr %tmp0, align 4 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp3 = load i32, ptr %tmp1, align 4 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i32 %tmp2, ptr %tmp0, align 4 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i32 %tmp3, ptr %tmp1, align 4 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i32 %tmp2, ptr %tmp0, align 4 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i32 %tmp3, ptr %tmp1, align 4 ; VF_4-LABEL: Checking a loop in 'i32_factor_2' ; VF_4: Found an estimated cost of 4 for VF 4 For instruction: %tmp2 = load i32, ptr %tmp0, align 4 ; VF_4-NEXT: Found an estimated cost of 0 for VF 4 For instruction: %tmp3 = load i32, ptr %tmp1, align 4 @@ -140,23 +140,23 @@ entry: ; VF_2-LABEL: Checking a loop in 'i64_factor_2' ; VF_2: Found an estimated cost of 22 for VF 2 For instruction: %tmp2 = load i64, ptr %tmp0, align 8 ; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: %tmp3 = load i64, ptr %tmp1, align 8 -; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: store i64 %tmp2, ptr %tmp0, align 8 -; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: store i64 %tmp3, ptr %tmp1, align 8 +; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store i64 %tmp2, ptr %tmp0, align 8 +; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store i64 %tmp3, ptr %tmp1, align 8 ; VF_4-LABEL: Checking a loop in 'i64_factor_2' ; VF_4: Found an estimated cost of 44 for VF 4 For instruction: %tmp2 = load i64, ptr %tmp0, align 8 ; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: %tmp3 = load i64, ptr %tmp1, align 8 -; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: store i64 %tmp2, ptr %tmp0, align 8 -; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: store i64 %tmp3, ptr %tmp1, align 8 +; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store i64 %tmp2, ptr %tmp0, align 8 +; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store i64 %tmp3, ptr %tmp1, align 8 ; VF_8-LABEL: Checking a loop in 'i64_factor_2' ; VF_8: Found an estimated cost of 88 for VF 8 For instruction: %tmp2 = load i64, ptr %tmp0, align 8 ; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: %tmp3 = load i64, ptr %tmp1, align 8 -; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: store i64 %tmp2, ptr %tmp0, align 8 -; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: store i64 %tmp3, ptr %tmp1, align 8 +; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store i64 %tmp2, ptr %tmp0, align 8 +; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store i64 %tmp3, ptr %tmp1, align 8 ; VF_16-LABEL: Checking a loop in 'i64_factor_2' ; VF_16: Found an estimated cost of 176 for VF 16 For instruction: %tmp2 = load i64, ptr %tmp0, align 8 ; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: %tmp3 = load i64, ptr %tmp1, align 8 -; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: store i64 %tmp2, ptr %tmp0, align 8 -; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: store i64 %tmp3, ptr %tmp1, align 8 +; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store i64 %tmp2, ptr %tmp0, align 8 +; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store i64 %tmp3, ptr %tmp1, align 8 for.body: %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ] %tmp0 = getelementptr inbounds %i64.2, ptr %data, i64 %i, i32 0 @@ -181,8 +181,8 @@ entry: ; VF_2-LABEL: Checking a loop in 'f16_factor_2' ; VF_2: Found an estimated cost of 6 for VF 2 For instruction: %tmp2 = load half, ptr %tmp0, align 2 ; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: %tmp3 = load half, ptr %tmp1, align 2 -; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store half %tmp2, ptr %tmp0, align 2 -; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store half %tmp3, ptr %tmp1, align 2 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store half %tmp2, ptr %tmp0, align 2 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store half %tmp3, ptr %tmp1, align 2 ; VF_4-LABEL: Checking a loop in 'f16_factor_2' ; VF_4: Found an estimated cost of 18 for VF 4 For instruction: %tmp2 = load half, ptr %tmp0, align 2 ; VF_4-NEXT: Found an estimated cost of 0 for VF 4 For instruction: %tmp3 = load half, ptr %tmp1, align 2 @@ -263,23 +263,23 @@ entry: ; VF_2-LABEL: Checking a loop in 'f64_factor_2' ; VF_2: Found an estimated cost of 6 for VF 2 For instruction: %tmp2 = load double, ptr %tmp0, align 8 ; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: %tmp3 = load double, ptr %tmp1, align 8 -; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store double %tmp2, ptr %tmp0, align 8 -; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store double %tmp3, ptr %tmp1, align 8 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store double %tmp2, ptr %tmp0, align 8 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store double %tmp3, ptr %tmp1, align 8 ; VF_4-LABEL: Checking a loop in 'f64_factor_2' ; VF_4: Found an estimated cost of 12 for VF 4 For instruction: %tmp2 = load double, ptr %tmp0, align 8 ; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: %tmp3 = load double, ptr %tmp1, align 8 -; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store double %tmp2, ptr %tmp0, align 8 -; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store double %tmp3, ptr %tmp1, align 8 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store double %tmp2, ptr %tmp0, align 8 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store double %tmp3, ptr %tmp1, align 8 ; VF_8-LABEL: Checking a loop in 'f64_factor_2' ; VF_8: Found an estimated cost of 24 for VF 8 For instruction: %tmp2 = load double, ptr %tmp0, align 8 ; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: %tmp3 = load double, ptr %tmp1, align 8 -; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store double %tmp2, ptr %tmp0, align 8 -; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store double %tmp3, ptr %tmp1, align 8 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store double %tmp2, ptr %tmp0, align 8 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store double %tmp3, ptr %tmp1, align 8 ; VF_16-LABEL: Checking a loop in 'f64_factor_2' ; VF_16: Found an estimated cost of 48 for VF 16 For instruction: %tmp2 = load double, ptr %tmp0, align 8 ; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: %tmp3 = load double, ptr %tmp1, align 8 -; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store double %tmp2, ptr %tmp0, align 8 -; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store double %tmp3, ptr %tmp1, align 8 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store double %tmp2, ptr %tmp0, align 8 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store double %tmp3, ptr %tmp1, align 8 for.body: %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ] %tmp0 = getelementptr inbounds %f64.2, ptr %data, i64 %i, i32 0 @@ -309,30 +309,30 @@ entry: ; VF_2: Found an estimated cost of 12 for VF 2 For instruction: %tmp3 = load i8, ptr %tmp0, align 1 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp4 = load i8, ptr %tmp1, align 1 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp5 = load i8, ptr %tmp2, align 1 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i8 %tmp3, ptr %tmp0, align 1 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i8 %tmp4, ptr %tmp1, align 1 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i8 %tmp5, ptr %tmp2, align 1 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i8 %tmp3, ptr %tmp0, align 1 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i8 %tmp4, ptr %tmp1, align 1 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i8 %tmp5, ptr %tmp2, align 1 ; VF_4-LABEL: Checking a loop in 'i8_factor_3' ; VF_4: Found an estimated cost of 24 for VF 4 For instruction: %tmp3 = load i8, ptr %tmp0, align 1 ; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: %tmp4 = load i8, ptr %tmp1, align 1 ; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: %tmp5 = load i8, ptr %tmp2, align 1 -; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: store i8 %tmp3, ptr %tmp0, align 1 -; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: store i8 %tmp4, ptr %tmp1, align 1 -; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: store i8 %tmp5, ptr %tmp2, align 1 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store i8 %tmp3, ptr %tmp0, align 1 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store i8 %tmp4, ptr %tmp1, align 1 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store i8 %tmp5, ptr %tmp2, align 1 ; VF_8-LABEL: Checking a loop in 'i8_factor_3' ; VF_8: Found an estimated cost of 48 for VF 8 For instruction: %tmp3 = load i8, ptr %tmp0, align 1 ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp4 = load i8, ptr %tmp1, align 1 ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp5 = load i8, ptr %tmp2, align 1 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i8 %tmp3, ptr %tmp0, align 1 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i8 %tmp4, ptr %tmp1, align 1 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i8 %tmp5, ptr %tmp2, align 1 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i8 %tmp3, ptr %tmp0, align 1 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i8 %tmp4, ptr %tmp1, align 1 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i8 %tmp5, ptr %tmp2, align 1 ; VF_16-LABEL: Checking a loop in 'i8_factor_3' ; VF_16: Found an estimated cost of 96 for VF 16 For instruction: %tmp3 = load i8, ptr %tmp0, align 1 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp4 = load i8, ptr %tmp1, align 1 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp5 = load i8, ptr %tmp2, align 1 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i8 %tmp3, ptr %tmp0, align 1 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i8 %tmp4, ptr %tmp1, align 1 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i8 %tmp5, ptr %tmp2, align 1 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i8 %tmp3, ptr %tmp0, align 1 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i8 %tmp4, ptr %tmp1, align 1 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i8 %tmp5, ptr %tmp2, align 1 for.body: %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ] %tmp0 = getelementptr inbounds %i8.3, ptr %data, i64 %i, i32 0 @@ -361,30 +361,30 @@ entry: ; VF_2: Found an estimated cost of 12 for VF 2 For instruction: %tmp3 = load i16, ptr %tmp0, align 2 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp4 = load i16, ptr %tmp1, align 2 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp5 = load i16, ptr %tmp2, align 2 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i16 %tmp3, ptr %tmp0, align 2 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i16 %tmp4, ptr %tmp1, align 2 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i16 %tmp5, ptr %tmp2, align 2 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i16 %tmp3, ptr %tmp0, align 2 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i16 %tmp4, ptr %tmp1, align 2 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i16 %tmp5, ptr %tmp2, align 2 ; VF_4-LABEL: Checking a loop in 'i16_factor_3' ; VF_4: Found an estimated cost of 24 for VF 4 For instruction: %tmp3 = load i16, ptr %tmp0, align 2 ; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: %tmp4 = load i16, ptr %tmp1, align 2 ; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: %tmp5 = load i16, ptr %tmp2, align 2 -; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: store i16 %tmp3, ptr %tmp0, align 2 -; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: store i16 %tmp4, ptr %tmp1, align 2 -; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: store i16 %tmp5, ptr %tmp2, align 2 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store i16 %tmp3, ptr %tmp0, align 2 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store i16 %tmp4, ptr %tmp1, align 2 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store i16 %tmp5, ptr %tmp2, align 2 ; VF_8-LABEL: Checking a loop in 'i16_factor_3' ; VF_8: Found an estimated cost of 48 for VF 8 For instruction: %tmp3 = load i16, ptr %tmp0, align 2 ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp4 = load i16, ptr %tmp1, align 2 ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp5 = load i16, ptr %tmp2, align 2 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i16 %tmp3, ptr %tmp0, align 2 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i16 %tmp4, ptr %tmp1, align 2 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i16 %tmp5, ptr %tmp2, align 2 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i16 %tmp3, ptr %tmp0, align 2 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i16 %tmp4, ptr %tmp1, align 2 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i16 %tmp5, ptr %tmp2, align 2 ; VF_16-LABEL: Checking a loop in 'i16_factor_3' ; VF_16: Found an estimated cost of 96 for VF 16 For instruction: %tmp3 = load i16, ptr %tmp0, align 2 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp4 = load i16, ptr %tmp1, align 2 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp5 = load i16, ptr %tmp2, align 2 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i16 %tmp3, ptr %tmp0, align 2 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i16 %tmp4, ptr %tmp1, align 2 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i16 %tmp5, ptr %tmp2, align 2 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i16 %tmp3, ptr %tmp0, align 2 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i16 %tmp4, ptr %tmp1, align 2 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i16 %tmp5, ptr %tmp2, align 2 for.body: %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ] %tmp0 = getelementptr inbounds %i16.3, ptr %data, i64 %i, i32 0 @@ -413,9 +413,9 @@ entry: ; VF_2: Found an estimated cost of 12 for VF 2 For instruction: %tmp3 = load i32, ptr %tmp0, align 4 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp4 = load i32, ptr %tmp1, align 4 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp5 = load i32, ptr %tmp2, align 4 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i32 %tmp3, ptr %tmp0, align 4 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i32 %tmp4, ptr %tmp1, align 4 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i32 %tmp5, ptr %tmp2, align 4 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i32 %tmp3, ptr %tmp0, align 4 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i32 %tmp4, ptr %tmp1, align 4 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i32 %tmp5, ptr %tmp2, align 4 ; VF_4-LABEL: Checking a loop in 'i32_factor_3' ; VF_4: Found an estimated cost of 8 for VF 4 For instruction: %tmp3 = load i32, ptr %tmp0, align 4 ; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: %tmp4 = load i32, ptr %tmp1, align 4 @@ -427,16 +427,16 @@ entry: ; VF_8: Found an estimated cost of 48 for VF 8 For instruction: %tmp3 = load i32, ptr %tmp0, align 4 ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp4 = load i32, ptr %tmp1, align 4 ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp5 = load i32, ptr %tmp2, align 4 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i32 %tmp3, ptr %tmp0, align 4 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i32 %tmp4, ptr %tmp1, align 4 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i32 %tmp5, ptr %tmp2, align 4 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i32 %tmp3, ptr %tmp0, align 4 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i32 %tmp4, ptr %tmp1, align 4 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i32 %tmp5, ptr %tmp2, align 4 ; VF_16-LABEL: Checking a loop in 'i32_factor_3' ; VF_16: Found an estimated cost of 96 for VF 16 For instruction: %tmp3 = load i32, ptr %tmp0, align 4 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp4 = load i32, ptr %tmp1, align 4 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp5 = load i32, ptr %tmp2, align 4 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i32 %tmp3, ptr %tmp0, align 4 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i32 %tmp4, ptr %tmp1, align 4 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i32 %tmp5, ptr %tmp2, align 4 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i32 %tmp3, ptr %tmp0, align 4 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i32 %tmp4, ptr %tmp1, align 4 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i32 %tmp5, ptr %tmp2, align 4 for.body: %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ] %tmp0 = getelementptr inbounds %i32.3, ptr %data, i64 %i, i32 0 @@ -465,30 +465,30 @@ entry: ; VF_2: Found an estimated cost of 22 for VF 2 For instruction: %tmp3 = load i64, ptr %tmp0, align 8 ; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: %tmp4 = load i64, ptr %tmp1, align 8 ; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: %tmp5 = load i64, ptr %tmp2, align 8 -; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: store i64 %tmp3, ptr %tmp0, align 8 -; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: store i64 %tmp4, ptr %tmp1, align 8 -; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: store i64 %tmp5, ptr %tmp2, align 8 +; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store i64 %tmp3, ptr %tmp0, align 8 +; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store i64 %tmp4, ptr %tmp1, align 8 +; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store i64 %tmp5, ptr %tmp2, align 8 ; VF_4-LABEL: Checking a loop in 'i64_factor_3' ; VF_4: Found an estimated cost of 44 for VF 4 For instruction: %tmp3 = load i64, ptr %tmp0, align 8 ; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: %tmp4 = load i64, ptr %tmp1, align 8 ; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: %tmp5 = load i64, ptr %tmp2, align 8 -; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: store i64 %tmp3, ptr %tmp0, align 8 -; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: store i64 %tmp4, ptr %tmp1, align 8 -; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: store i64 %tmp5, ptr %tmp2, align 8 +; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store i64 %tmp3, ptr %tmp0, align 8 +; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store i64 %tmp4, ptr %tmp1, align 8 +; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store i64 %tmp5, ptr %tmp2, align 8 ; VF_8-LABEL: Checking a loop in 'i64_factor_3' ; VF_8: Found an estimated cost of 88 for VF 8 For instruction: %tmp3 = load i64, ptr %tmp0, align 8 ; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: %tmp4 = load i64, ptr %tmp1, align 8 ; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: %tmp5 = load i64, ptr %tmp2, align 8 -; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: store i64 %tmp3, ptr %tmp0, align 8 -; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: store i64 %tmp4, ptr %tmp1, align 8 -; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: store i64 %tmp5, ptr %tmp2, align 8 +; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store i64 %tmp3, ptr %tmp0, align 8 +; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store i64 %tmp4, ptr %tmp1, align 8 +; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store i64 %tmp5, ptr %tmp2, align 8 ; VF_16-LABEL: Checking a loop in 'i64_factor_3' ; VF_16: Found an estimated cost of 176 for VF 16 For instruction: %tmp3 = load i64, ptr %tmp0, align 8 ; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: %tmp4 = load i64, ptr %tmp1, align 8 ; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: %tmp5 = load i64, ptr %tmp2, align 8 -; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: store i64 %tmp3, ptr %tmp0, align 8 -; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: store i64 %tmp4, ptr %tmp1, align 8 -; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: store i64 %tmp5, ptr %tmp2, align 8 +; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store i64 %tmp3, ptr %tmp0, align 8 +; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store i64 %tmp4, ptr %tmp1, align 8 +; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store i64 %tmp5, ptr %tmp2, align 8 for.body: %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ] %tmp0 = getelementptr inbounds %i64.3, ptr %data, i64 %i, i32 0 @@ -517,9 +517,9 @@ entry: ; VF_2: Found an estimated cost of 6 for VF 2 For instruction: %tmp3 = load half, ptr %tmp0, align 2 ; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: %tmp4 = load half, ptr %tmp1, align 2 ; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: %tmp5 = load half, ptr %tmp2, align 2 -; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store half %tmp3, ptr %tmp0, align 2 -; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store half %tmp4, ptr %tmp1, align 2 -; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store half %tmp5, ptr %tmp2, align 2 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store half %tmp3, ptr %tmp0, align 2 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store half %tmp4, ptr %tmp1, align 2 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store half %tmp5, ptr %tmp2, align 2 ; VF_4-LABEL: Checking a loop in 'f16_factor_3' ; VF_4: Found an estimated cost of 28 for VF 4 For instruction: %tmp3 = load half, ptr %tmp0, align 2 ; VF_4-NEXT: Found an estimated cost of 0 for VF 4 For instruction: %tmp4 = load half, ptr %tmp1, align 2 @@ -621,30 +621,30 @@ entry: ; VF_2: Found an estimated cost of 6 for VF 2 For instruction: %tmp3 = load double, ptr %tmp0, align 8 ; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: %tmp4 = load double, ptr %tmp1, align 8 ; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: %tmp5 = load double, ptr %tmp2, align 8 -; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store double %tmp3, ptr %tmp0, align 8 -; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store double %tmp4, ptr %tmp1, align 8 -; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store double %tmp5, ptr %tmp2, align 8 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store double %tmp3, ptr %tmp0, align 8 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store double %tmp4, ptr %tmp1, align 8 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store double %tmp5, ptr %tmp2, align 8 ; VF_4-LABEL: Checking a loop in 'f64_factor_3' ; VF_4: Found an estimated cost of 12 for VF 4 For instruction: %tmp3 = load double, ptr %tmp0, align 8 ; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: %tmp4 = load double, ptr %tmp1, align 8 ; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: %tmp5 = load double, ptr %tmp2, align 8 -; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store double %tmp3, ptr %tmp0, align 8 -; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store double %tmp4, ptr %tmp1, align 8 -; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store double %tmp5, ptr %tmp2, align 8 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store double %tmp3, ptr %tmp0, align 8 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store double %tmp4, ptr %tmp1, align 8 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store double %tmp5, ptr %tmp2, align 8 ; VF_8-LABEL: Checking a loop in 'f64_factor_3' ; VF_8: Found an estimated cost of 24 for VF 8 For instruction: %tmp3 = load double, ptr %tmp0, align 8 ; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: %tmp4 = load double, ptr %tmp1, align 8 ; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: %tmp5 = load double, ptr %tmp2, align 8 -; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store double %tmp3, ptr %tmp0, align 8 -; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store double %tmp4, ptr %tmp1, align 8 -; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store double %tmp5, ptr %tmp2, align 8 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store double %tmp3, ptr %tmp0, align 8 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store double %tmp4, ptr %tmp1, align 8 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store double %tmp5, ptr %tmp2, align 8 ; VF_16-LABEL: Checking a loop in 'f64_factor_3' ; VF_16: Found an estimated cost of 48 for VF 16 For instruction: %tmp3 = load double, ptr %tmp0, align 8 ; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: %tmp4 = load double, ptr %tmp1, align 8 ; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: %tmp5 = load double, ptr %tmp2, align 8 -; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store double %tmp3, ptr %tmp0, align 8 -; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store double %tmp4, ptr %tmp1, align 8 -; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store double %tmp5, ptr %tmp2, align 8 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store double %tmp3, ptr %tmp0, align 8 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store double %tmp4, ptr %tmp1, align 8 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store double %tmp5, ptr %tmp2, align 8 for.body: %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ] %tmp0 = getelementptr inbounds %f64.3, ptr %data, i64 %i, i32 0 @@ -677,37 +677,37 @@ entry: ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp5 = load i8, ptr %tmp1, align 1 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp6 = load i8, ptr %tmp2, align 1 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp7 = load i8, ptr %tmp3, align 1 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i8 %tmp4, ptr %tmp0, align 1 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i8 %tmp5, ptr %tmp1, align 1 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i8 %tmp6, ptr %tmp2, align 1 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i8 %tmp7, ptr %tmp3, align 1 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i8 %tmp4, ptr %tmp0, align 1 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i8 %tmp5, ptr %tmp1, align 1 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i8 %tmp6, ptr %tmp2, align 1 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i8 %tmp7, ptr %tmp3, align 1 ; VF_4-LABEL: Checking a loop in 'i8_factor_4' ; VF_4: Found an estimated cost of 24 for VF 4 For instruction: %tmp4 = load i8, ptr %tmp0, align 1 ; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: %tmp5 = load i8, ptr %tmp1, align 1 ; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: %tmp6 = load i8, ptr %tmp2, align 1 ; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: %tmp7 = load i8, ptr %tmp3, align 1 -; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: store i8 %tmp4, ptr %tmp0, align 1 -; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: store i8 %tmp5, ptr %tmp1, align 1 -; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: store i8 %tmp6, ptr %tmp2, align 1 -; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: store i8 %tmp7, ptr %tmp3, align 1 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store i8 %tmp4, ptr %tmp0, align 1 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store i8 %tmp5, ptr %tmp1, align 1 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store i8 %tmp6, ptr %tmp2, align 1 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store i8 %tmp7, ptr %tmp3, align 1 ; VF_8-LABEL: Checking a loop in 'i8_factor_4' ; VF_8: Found an estimated cost of 48 for VF 8 For instruction: %tmp4 = load i8, ptr %tmp0, align 1 ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp5 = load i8, ptr %tmp1, align 1 ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp6 = load i8, ptr %tmp2, align 1 ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp7 = load i8, ptr %tmp3, align 1 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i8 %tmp4, ptr %tmp0, align 1 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i8 %tmp5, ptr %tmp1, align 1 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i8 %tmp6, ptr %tmp2, align 1 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i8 %tmp7, ptr %tmp3, align 1 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i8 %tmp4, ptr %tmp0, align 1 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i8 %tmp5, ptr %tmp1, align 1 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i8 %tmp6, ptr %tmp2, align 1 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i8 %tmp7, ptr %tmp3, align 1 ; VF_16-LABEL: Checking a loop in 'i8_factor_4' ; VF_16: Found an estimated cost of 96 for VF 16 For instruction: %tmp4 = load i8, ptr %tmp0, align 1 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp5 = load i8, ptr %tmp1, align 1 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp6 = load i8, ptr %tmp2, align 1 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp7 = load i8, ptr %tmp3, align 1 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i8 %tmp4, ptr %tmp0, align 1 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i8 %tmp5, ptr %tmp1, align 1 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i8 %tmp6, ptr %tmp2, align 1 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i8 %tmp7, ptr %tmp3, align 1 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i8 %tmp4, ptr %tmp0, align 1 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i8 %tmp5, ptr %tmp1, align 1 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i8 %tmp6, ptr %tmp2, align 1 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i8 %tmp7, ptr %tmp3, align 1 for.body: %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ] %tmp0 = getelementptr inbounds %i8.4, ptr %data, i64 %i, i32 0 @@ -740,37 +740,37 @@ entry: ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp5 = load i16, ptr %tmp1, align 2 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp6 = load i16, ptr %tmp2, align 2 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp7 = load i16, ptr %tmp3, align 2 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i16 %tmp4, ptr %tmp0, align 2 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i16 %tmp5, ptr %tmp1, align 2 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i16 %tmp6, ptr %tmp2, align 2 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i16 %tmp7, ptr %tmp3, align 2 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i16 %tmp4, ptr %tmp0, align 2 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i16 %tmp5, ptr %tmp1, align 2 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i16 %tmp6, ptr %tmp2, align 2 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i16 %tmp7, ptr %tmp3, align 2 ; VF_4-LABEL: Checking a loop in 'i16_factor_4' ; VF_4: Found an estimated cost of 24 for VF 4 For instruction: %tmp4 = load i16, ptr %tmp0, align 2 ; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: %tmp5 = load i16, ptr %tmp1, align 2 ; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: %tmp6 = load i16, ptr %tmp2, align 2 ; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: %tmp7 = load i16, ptr %tmp3, align 2 -; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: store i16 %tmp4, ptr %tmp0, align 2 -; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: store i16 %tmp5, ptr %tmp1, align 2 -; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: store i16 %tmp6, ptr %tmp2, align 2 -; VF_4-NEXT: Found an estimated cost of 24 for VF 4 For instruction: store i16 %tmp7, ptr %tmp3, align 2 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store i16 %tmp4, ptr %tmp0, align 2 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store i16 %tmp5, ptr %tmp1, align 2 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store i16 %tmp6, ptr %tmp2, align 2 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store i16 %tmp7, ptr %tmp3, align 2 ; VF_8-LABEL: Checking a loop in 'i16_factor_4' ; VF_8: Found an estimated cost of 48 for VF 8 For instruction: %tmp4 = load i16, ptr %tmp0, align 2 ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp5 = load i16, ptr %tmp1, align 2 ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp6 = load i16, ptr %tmp2, align 2 ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp7 = load i16, ptr %tmp3, align 2 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i16 %tmp4, ptr %tmp0, align 2 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i16 %tmp5, ptr %tmp1, align 2 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i16 %tmp6, ptr %tmp2, align 2 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i16 %tmp7, ptr %tmp3, align 2 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i16 %tmp4, ptr %tmp0, align 2 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i16 %tmp5, ptr %tmp1, align 2 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i16 %tmp6, ptr %tmp2, align 2 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i16 %tmp7, ptr %tmp3, align 2 ; VF_16-LABEL: Checking a loop in 'i16_factor_4' ; VF_16: Found an estimated cost of 96 for VF 16 For instruction: %tmp4 = load i16, ptr %tmp0, align 2 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp5 = load i16, ptr %tmp1, align 2 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp6 = load i16, ptr %tmp2, align 2 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp7 = load i16, ptr %tmp3, align 2 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i16 %tmp4, ptr %tmp0, align 2 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i16 %tmp5, ptr %tmp1, align 2 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i16 %tmp6, ptr %tmp2, align 2 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i16 %tmp7, ptr %tmp3, align 2 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i16 %tmp4, ptr %tmp0, align 2 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i16 %tmp5, ptr %tmp1, align 2 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i16 %tmp6, ptr %tmp2, align 2 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i16 %tmp7, ptr %tmp3, align 2 for.body: %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ] %tmp0 = getelementptr inbounds %i16.4, ptr %data, i64 %i, i32 0 @@ -803,10 +803,10 @@ entry: ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp5 = load i32, ptr %tmp1, align 4 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp6 = load i32, ptr %tmp2, align 4 ; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: %tmp7 = load i32, ptr %tmp3, align 4 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i32 %tmp4, ptr %tmp0, align 4 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i32 %tmp5, ptr %tmp1, align 4 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i32 %tmp6, ptr %tmp2, align 4 -; VF_2-NEXT: Found an estimated cost of 12 for VF 2 For instruction: store i32 %tmp7, ptr %tmp3, align 4 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i32 %tmp4, ptr %tmp0, align 4 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i32 %tmp5, ptr %tmp1, align 4 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i32 %tmp6, ptr %tmp2, align 4 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store i32 %tmp7, ptr %tmp3, align 4 ; VF_4-LABEL: Checking a loop in 'i32_factor_4' ; VF_4: Found an estimated cost of 8 for VF 4 For instruction: %tmp4 = load i32, ptr %tmp0, align 4 ; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: %tmp5 = load i32, ptr %tmp1, align 4 @@ -821,19 +821,19 @@ entry: ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp5 = load i32, ptr %tmp1, align 4 ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp6 = load i32, ptr %tmp2, align 4 ; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: %tmp7 = load i32, ptr %tmp3, align 4 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i32 %tmp4, ptr %tmp0, align 4 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i32 %tmp5, ptr %tmp1, align 4 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i32 %tmp6, ptr %tmp2, align 4 -; VF_8-NEXT: Found an estimated cost of 48 for VF 8 For instruction: store i32 %tmp7, ptr %tmp3, align 4 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i32 %tmp4, ptr %tmp0, align 4 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i32 %tmp5, ptr %tmp1, align 4 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i32 %tmp6, ptr %tmp2, align 4 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store i32 %tmp7, ptr %tmp3, align 4 ; VF_16-LABEL: Checking a loop in 'i32_factor_4' ; VF_16: Found an estimated cost of 96 for VF 16 For instruction: %tmp4 = load i32, ptr %tmp0, align 4 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp5 = load i32, ptr %tmp1, align 4 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp6 = load i32, ptr %tmp2, align 4 ; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: %tmp7 = load i32, ptr %tmp3, align 4 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i32 %tmp4, ptr %tmp0, align 4 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i32 %tmp5, ptr %tmp1, align 4 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i32 %tmp6, ptr %tmp2, align 4 -; VF_16-NEXT: Found an estimated cost of 96 for VF 16 For instruction: store i32 %tmp7, ptr %tmp3, align 4 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i32 %tmp4, ptr %tmp0, align 4 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i32 %tmp5, ptr %tmp1, align 4 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i32 %tmp6, ptr %tmp2, align 4 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store i32 %tmp7, ptr %tmp3, align 4 for.body: %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ] %tmp0 = getelementptr inbounds %i32.4, ptr %data, i64 %i, i32 0 @@ -866,37 +866,37 @@ entry: ; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: %tmp5 = load i64, ptr %tmp1, align 8 ; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: %tmp6 = load i64, ptr %tmp2, align 8 ; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: %tmp7 = load i64, ptr %tmp3, align 8 -; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: store i64 %tmp4, ptr %tmp0, align 8 -; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: store i64 %tmp5, ptr %tmp1, align 8 -; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: store i64 %tmp6, ptr %tmp2, align 8 -; VF_2-NEXT: Found an estimated cost of 22 for VF 2 For instruction: store i64 %tmp7, ptr %tmp3, align 8 +; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store i64 %tmp4, ptr %tmp0, align 8 +; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store i64 %tmp5, ptr %tmp1, align 8 +; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store i64 %tmp6, ptr %tmp2, align 8 +; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store i64 %tmp7, ptr %tmp3, align 8 ; VF_4-LABEL: Checking a loop in 'i64_factor_4' ; VF_4: Found an estimated cost of 44 for VF 4 For instruction: %tmp4 = load i64, ptr %tmp0, align 8 ; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: %tmp5 = load i64, ptr %tmp1, align 8 ; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: %tmp6 = load i64, ptr %tmp2, align 8 ; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: %tmp7 = load i64, ptr %tmp3, align 8 -; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: store i64 %tmp4, ptr %tmp0, align 8 -; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: store i64 %tmp5, ptr %tmp1, align 8 -; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: store i64 %tmp6, ptr %tmp2, align 8 -; VF_4-NEXT: Found an estimated cost of 44 for VF 4 For instruction: store i64 %tmp7, ptr %tmp3, align 8 +; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store i64 %tmp4, ptr %tmp0, align 8 +; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store i64 %tmp5, ptr %tmp1, align 8 +; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store i64 %tmp6, ptr %tmp2, align 8 +; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store i64 %tmp7, ptr %tmp3, align 8 ; VF_8-LABEL: Checking a loop in 'i64_factor_4' ; VF_8: Found an estimated cost of 88 for VF 8 For instruction: %tmp4 = load i64, ptr %tmp0, align 8 ; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: %tmp5 = load i64, ptr %tmp1, align 8 ; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: %tmp6 = load i64, ptr %tmp2, align 8 ; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: %tmp7 = load i64, ptr %tmp3, align 8 -; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: store i64 %tmp4, ptr %tmp0, align 8 -; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: store i64 %tmp5, ptr %tmp1, align 8 -; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: store i64 %tmp6, ptr %tmp2, align 8 -; VF_8-NEXT: Found an estimated cost of 88 for VF 8 For instruction: store i64 %tmp7, ptr %tmp3, align 8 +; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store i64 %tmp4, ptr %tmp0, align 8 +; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store i64 %tmp5, ptr %tmp1, align 8 +; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store i64 %tmp6, ptr %tmp2, align 8 +; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store i64 %tmp7, ptr %tmp3, align 8 ; VF_16-LABEL: Checking a loop in 'i64_factor_4' ; VF_16: Found an estimated cost of 176 for VF 16 For instruction: %tmp4 = load i64, ptr %tmp0, align 8 ; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: %tmp5 = load i64, ptr %tmp1, align 8 ; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: %tmp6 = load i64, ptr %tmp2, align 8 ; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: %tmp7 = load i64, ptr %tmp3, align 8 -; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: store i64 %tmp4, ptr %tmp0, align 8 -; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: store i64 %tmp5, ptr %tmp1, align 8 -; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: store i64 %tmp6, ptr %tmp2, align 8 -; VF_16-NEXT: Found an estimated cost of 176 for VF 16 For instruction: store i64 %tmp7, ptr %tmp3, align 8 +; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store i64 %tmp4, ptr %tmp0, align 8 +; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store i64 %tmp5, ptr %tmp1, align 8 +; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store i64 %tmp6, ptr %tmp2, align 8 +; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store i64 %tmp7, ptr %tmp3, align 8 for.body: %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ] %tmp0 = getelementptr inbounds %i64.4, ptr %data, i64 %i, i32 0 @@ -1055,37 +1055,37 @@ entry: ; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: %tmp5 = load double, ptr %tmp1, align 8 ; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: %tmp6 = load double, ptr %tmp2, align 8 ; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: %tmp7 = load double, ptr %tmp3, align 8 -; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store double %tmp4, ptr %tmp0, align 8 -; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store double %tmp5, ptr %tmp1, align 8 -; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store double %tmp6, ptr %tmp2, align 8 -; VF_2-NEXT: Found an estimated cost of 6 for VF 2 For instruction: store double %tmp7, ptr %tmp3, align 8 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store double %tmp4, ptr %tmp0, align 8 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store double %tmp5, ptr %tmp1, align 8 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store double %tmp6, ptr %tmp2, align 8 +; VF_2-NEXT: Found an estimated cost of 4 for VF 2 For instruction: store double %tmp7, ptr %tmp3, align 8 ; VF_4-LABEL: Checking a loop in 'f64_factor_4' ; VF_4: Found an estimated cost of 12 for VF 4 For instruction: %tmp4 = load double, ptr %tmp0, align 8 ; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: %tmp5 = load double, ptr %tmp1, align 8 ; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: %tmp6 = load double, ptr %tmp2, align 8 ; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: %tmp7 = load double, ptr %tmp3, align 8 -; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store double %tmp4, ptr %tmp0, align 8 -; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store double %tmp5, ptr %tmp1, align 8 -; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store double %tmp6, ptr %tmp2, align 8 -; VF_4-NEXT: Found an estimated cost of 12 for VF 4 For instruction: store double %tmp7, ptr %tmp3, align 8 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store double %tmp4, ptr %tmp0, align 8 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store double %tmp5, ptr %tmp1, align 8 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store double %tmp6, ptr %tmp2, align 8 +; VF_4-NEXT: Found an estimated cost of 8 for VF 4 For instruction: store double %tmp7, ptr %tmp3, align 8 ; VF_8-LABEL: Checking a loop in 'f64_factor_4' ; VF_8: Found an estimated cost of 24 for VF 8 For instruction: %tmp4 = load double, ptr %tmp0, align 8 ; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: %tmp5 = load double, ptr %tmp1, align 8 ; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: %tmp6 = load double, ptr %tmp2, align 8 ; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: %tmp7 = load double, ptr %tmp3, align 8 -; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store double %tmp4, ptr %tmp0, align 8 -; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store double %tmp5, ptr %tmp1, align 8 -; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store double %tmp6, ptr %tmp2, align 8 -; VF_8-NEXT: Found an estimated cost of 24 for VF 8 For instruction: store double %tmp7, ptr %tmp3, align 8 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store double %tmp4, ptr %tmp0, align 8 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store double %tmp5, ptr %tmp1, align 8 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store double %tmp6, ptr %tmp2, align 8 +; VF_8-NEXT: Found an estimated cost of 16 for VF 8 For instruction: store double %tmp7, ptr %tmp3, align 8 ; VF_16-LABEL: Checking a loop in 'f64_factor_4' ; VF_16: Found an estimated cost of 48 for VF 16 For instruction: %tmp4 = load double, ptr %tmp0, align 8 ; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: %tmp5 = load double, ptr %tmp1, align 8 ; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: %tmp6 = load double, ptr %tmp2, align 8 ; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: %tmp7 = load double, ptr %tmp3, align 8 -; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store double %tmp4, ptr %tmp0, align 8 -; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store double %tmp5, ptr %tmp1, align 8 -; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store double %tmp6, ptr %tmp2, align 8 -; VF_16-NEXT: Found an estimated cost of 48 for VF 16 For instruction: store double %tmp7, ptr %tmp3, align 8 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store double %tmp4, ptr %tmp0, align 8 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store double %tmp5, ptr %tmp1, align 8 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store double %tmp6, ptr %tmp2, align 8 +; VF_16-NEXT: Found an estimated cost of 32 for VF 16 For instruction: store double %tmp7, ptr %tmp3, align 8 for.body: %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ] %tmp0 = getelementptr inbounds %f64.4, ptr %data, i64 %i, i32 0 diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-multiexit.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-multiexit.ll index 789a97c052a9..cc2fbb1b0df7 100644 --- a/llvm/test/Transforms/LoopVectorize/ARM/mve-multiexit.ll +++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-multiexit.ll @@ -42,7 +42,7 @@ define void @multiple_exits_unique_exit_block(ptr %A, ptr %B, i32 %N) #0 { ; CHECK: middle.block: ; CHECK-NEXT: br label [[SCALAR_PH]] ; CHECK: scalar.ph: -; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ], [ 0, [[VECTOR_MEMCHECK]] ] +; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[VECTOR_MEMCHECK]] ], [ 0, [[ENTRY:%.*]] ] ; CHECK-NEXT: br label [[LOOP_HEADER:%.*]] ; CHECK: loop.header: ; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY:%.*]] ] @@ -116,7 +116,7 @@ define i32 @multiple_exits_multiple_exit_blocks(ptr %A, ptr %B, i32 %N) #0 { ; CHECK: middle.block: ; CHECK-NEXT: br label [[SCALAR_PH]] ; CHECK: scalar.ph: -; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ], [ 0, [[VECTOR_MEMCHECK]] ] +; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], [[MIDDLE_BLOCK]] ], [ 0, [[VECTOR_MEMCHECK]] ], [ 0, [[ENTRY:%.*]] ] ; CHECK-NEXT: br label [[LOOP_HEADER:%.*]] ; CHECK: loop.header: ; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], [[FOR_BODY:%.*]] ] diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-qabs.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-qabs.ll index fde594d04836..0595bb31a58d 100644 --- a/llvm/test/Transforms/LoopVectorize/ARM/mve-qabs.ll +++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-qabs.ll @@ -49,14 +49,14 @@ define void @arm_abs_q7(ptr nocapture readonly %pSrc, ptr nocapture %pDst, i32 % ; CHECK-NEXT: [[PSRC_ADDR_022:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[BLKCNT_021:%.*]] = phi i32 [ [[DEC:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[PDST_ADDR_020:%.*]] = phi ptr [ [[INCDEC_PTR13:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL6]], [[SCALAR_PH]] ] -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[PSRC_ADDR_022]], i32 1 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[PSRC_ADDR_022]], i32 1 ; CHECK-NEXT: [[TMP7:%.*]] = load i8, ptr [[PSRC_ADDR_022]], align 1 ; CHECK-NEXT: [[CMP1:%.*]] = icmp sgt i8 [[TMP7]], 0 ; CHECK-NEXT: [[CMP5:%.*]] = icmp eq i8 [[TMP7]], -128 ; CHECK-NEXT: [[SUB:%.*]] = sub i8 0, [[TMP7]] ; CHECK-NEXT: [[COND:%.*]] = select i1 [[CMP5]], i8 127, i8 [[SUB]] ; CHECK-NEXT: [[COND11:%.*]] = select i1 [[CMP1]], i8 [[TMP7]], i8 [[COND]] -; CHECK-NEXT: [[INCDEC_PTR13]] = getelementptr inbounds i8, ptr [[PDST_ADDR_020]], i32 1 +; CHECK-NEXT: [[INCDEC_PTR13]] = getelementptr inbounds nuw i8, ptr [[PDST_ADDR_020]], i32 1 ; CHECK-NEXT: store i8 [[COND11]], ptr [[PDST_ADDR_020]], align 1 ; CHECK-NEXT: [[DEC]] = add i32 [[BLKCNT_021]], -1 ; CHECK-NEXT: [[CMP_NOT:%.*]] = icmp eq i32 [[DEC]], 0 @@ -144,14 +144,14 @@ define void @arm_abs_q15(ptr nocapture readonly %pSrc, ptr nocapture %pDst, i32 ; CHECK-NEXT: [[PSRC_ADDR_023:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[BLKCNT_022:%.*]] = phi i32 [ [[DEC:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[PDST_ADDR_021:%.*]] = phi ptr [ [[INCDEC_PTR13:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL6]], [[SCALAR_PH]] ] -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[PSRC_ADDR_023]], i32 2 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[PSRC_ADDR_023]], i32 2 ; CHECK-NEXT: [[TMP9:%.*]] = load i16, ptr [[PSRC_ADDR_023]], align 2 ; CHECK-NEXT: [[CMP1:%.*]] = icmp sgt i16 [[TMP9]], 0 ; CHECK-NEXT: [[CMP5:%.*]] = icmp eq i16 [[TMP9]], -32768 ; CHECK-NEXT: [[SUB:%.*]] = sub i16 0, [[TMP9]] ; CHECK-NEXT: [[COND:%.*]] = select i1 [[CMP5]], i16 32767, i16 [[SUB]] ; CHECK-NEXT: [[COND11:%.*]] = select i1 [[CMP1]], i16 [[TMP9]], i16 [[COND]] -; CHECK-NEXT: [[INCDEC_PTR13]] = getelementptr inbounds i8, ptr [[PDST_ADDR_021]], i32 2 +; CHECK-NEXT: [[INCDEC_PTR13]] = getelementptr inbounds nuw i8, ptr [[PDST_ADDR_021]], i32 2 ; CHECK-NEXT: store i16 [[COND11]], ptr [[PDST_ADDR_021]], align 2 ; CHECK-NEXT: [[DEC]] = add i32 [[BLKCNT_022]], -1 ; CHECK-NEXT: [[CMP_NOT:%.*]] = icmp eq i32 [[DEC]], 0 @@ -239,14 +239,14 @@ define void @arm_abs_q31(ptr nocapture readonly %pSrc, ptr nocapture %pDst, i32 ; CHECK-NEXT: [[PSRC_ADDR_017:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[BLKCNT_016:%.*]] = phi i32 [ [[DEC:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[PDST_ADDR_015:%.*]] = phi ptr [ [[INCDEC_PTR7:%.*]], [[WHILE_BODY]] ], [ [[BC_RESUME_VAL6]], [[SCALAR_PH]] ] -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[PSRC_ADDR_017]], i32 4 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[PSRC_ADDR_017]], i32 4 ; CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[PSRC_ADDR_017]], align 4 ; CHECK-NEXT: [[CMP1:%.*]] = icmp sgt i32 [[TMP9]], 0 ; CHECK-NEXT: [[CMP2:%.*]] = icmp eq i32 [[TMP9]], -2147483648 ; CHECK-NEXT: [[SUB:%.*]] = sub nsw i32 0, [[TMP9]] ; CHECK-NEXT: [[COND:%.*]] = select i1 [[CMP2]], i32 2147483647, i32 [[SUB]] ; CHECK-NEXT: [[COND6:%.*]] = select i1 [[CMP1]], i32 [[TMP9]], i32 [[COND]] -; CHECK-NEXT: [[INCDEC_PTR7]] = getelementptr inbounds i8, ptr [[PDST_ADDR_015]], i32 4 +; CHECK-NEXT: [[INCDEC_PTR7]] = getelementptr inbounds nuw i8, ptr [[PDST_ADDR_015]], i32 4 ; CHECK-NEXT: store i32 [[COND6]], ptr [[PDST_ADDR_015]], align 4 ; CHECK-NEXT: [[DEC]] = add i32 [[BLKCNT_016]], -1 ; CHECK-NEXT: [[CMP_NOT:%.*]] = icmp eq i32 [[DEC]], 0 diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions.ll index baad7a84a891..c459dd28fdae 100644 --- a/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions.ll +++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-reductions.ll @@ -13,7 +13,7 @@ define i64 @add_i64_i64(ptr nocapture readonly %x, i32 %n) #0 { ; CHECK: for.body: ; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY:%.*]] ] ; CHECK-NEXT: [[R_07:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY]] ] -; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[X:%.*]], i32 [[I_08]] +; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i64, ptr [[X:%.*]], i32 [[I_08]] ; CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr [[ARRAYIDX]], align 8 ; CHECK-NEXT: [[ADD]] = add nsw i64 [[TMP0]], [[R_07]] ; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1 @@ -76,7 +76,7 @@ define i64 @add_i32_i64(ptr nocapture readonly %x, i32 %n) #0 { ; CHECK: for.body: ; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[R_07:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ] -; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[X]], i32 [[I_08]] +; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[X]], i32 [[I_08]] ; CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 ; CHECK-NEXT: [[CONV:%.*]] = sext i32 [[TMP5]] to i64 ; CHECK-NEXT: [[ADD]] = add nsw i64 [[R_07]], [[CONV]] @@ -141,7 +141,7 @@ define i64 @add_i16_i64(ptr nocapture readonly %x, i32 %n) #0 { ; CHECK: for.body: ; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[R_07:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ] -; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[X]], i32 [[I_08]] +; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i16, ptr [[X]], i32 [[I_08]] ; CHECK-NEXT: [[TMP5:%.*]] = load i16, ptr [[ARRAYIDX]], align 2 ; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP5]] to i64 ; CHECK-NEXT: [[ADD]] = add nsw i64 [[R_07]], [[CONV]] @@ -206,7 +206,7 @@ define i64 @add_i8_i64(ptr nocapture readonly %x, i32 %n) #0 { ; CHECK: for.body: ; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[R_07:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ] -; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[X]], i32 [[I_08]] +; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i32 [[I_08]] ; CHECK-NEXT: [[TMP5:%.*]] = load i8, ptr [[ARRAYIDX]], align 1 ; CHECK-NEXT: [[CONV:%.*]] = zext i8 [[TMP5]] to i64 ; CHECK-NEXT: [[ADD]] = add nuw nsw i64 [[R_07]], [[CONV]] @@ -519,9 +519,9 @@ define i64 @mla_i64_i64(ptr nocapture readonly %x, ptr nocapture readonly %y, i3 ; CHECK: for.body: ; CHECK-NEXT: [[I_010:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY:%.*]] ] ; CHECK-NEXT: [[R_09:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY]] ] -; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i64, ptr [[X:%.*]], i32 [[I_010]] +; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i64, ptr [[X:%.*]], i32 [[I_010]] ; CHECK-NEXT: [[TMP0:%.*]] = load i64, ptr [[ARRAYIDX]], align 8 -; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i64, ptr [[Y:%.*]], i32 [[I_010]] +; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i64, ptr [[Y:%.*]], i32 [[I_010]] ; CHECK-NEXT: [[TMP1:%.*]] = load i64, ptr [[ARRAYIDX1]], align 8 ; CHECK-NEXT: [[MUL:%.*]] = mul nsw i64 [[TMP1]], [[TMP0]] ; CHECK-NEXT: [[ADD]] = add nsw i64 [[MUL]], [[R_09]] @@ -591,9 +591,9 @@ define i64 @mla_i32_i64(ptr nocapture readonly %x, ptr nocapture readonly %y, i3 ; CHECK: for.body: ; CHECK-NEXT: [[I_010:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[R_09:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ] -; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[X]], i32 [[I_010]] +; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[X]], i32 [[I_010]] ; CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 -; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i32, ptr [[Y]], i32 [[I_010]] +; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i32, ptr [[Y]], i32 [[I_010]] ; CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[ARRAYIDX1]], align 4 ; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], [[TMP7]] ; CHECK-NEXT: [[CONV:%.*]] = sext i32 [[MUL]] to i64 @@ -667,10 +667,10 @@ define i64 @mla_i16_i64(ptr nocapture readonly %x, ptr nocapture readonly %y, i3 ; CHECK: for.body: ; CHECK-NEXT: [[I_012:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[R_011:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ] -; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[X]], i32 [[I_012]] +; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i16, ptr [[X]], i32 [[I_012]] ; CHECK-NEXT: [[TMP9:%.*]] = load i16, ptr [[ARRAYIDX]], align 2 ; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP9]] to i32 -; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i16, ptr [[Y]], i32 [[I_012]] +; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i16, ptr [[Y]], i32 [[I_012]] ; CHECK-NEXT: [[TMP10:%.*]] = load i16, ptr [[ARRAYIDX1]], align 2 ; CHECK-NEXT: [[CONV2:%.*]] = sext i16 [[TMP10]] to i32 ; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[CONV2]], [[CONV]] @@ -747,10 +747,10 @@ define i64 @mla_i8_i64(ptr nocapture readonly %x, ptr nocapture readonly %y, i32 ; CHECK: for.body: ; CHECK-NEXT: [[I_012:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[R_011:%.*]] = phi i64 [ [[ADD:%.*]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ] -; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8, ptr [[X]], i32 [[I_012]] +; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i8, ptr [[X]], i32 [[I_012]] ; CHECK-NEXT: [[TMP9:%.*]] = load i8, ptr [[ARRAYIDX]], align 1 ; CHECK-NEXT: [[CONV:%.*]] = zext i8 [[TMP9]] to i32 -; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, ptr [[Y]], i32 [[I_012]] +; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i8, ptr [[Y]], i32 [[I_012]] ; CHECK-NEXT: [[TMP10:%.*]] = load i8, ptr [[ARRAYIDX1]], align 1 ; CHECK-NEXT: [[CONV2:%.*]] = zext i8 [[TMP10]] to i32 ; CHECK-NEXT: [[MUL:%.*]] = mul nuw nsw i32 [[CONV2]], [[CONV]] @@ -1482,7 +1482,7 @@ define i64 @mla_xx_sext_zext(ptr nocapture noundef readonly %x, i32 %n) #0 { ; CHECK: for.body: ; CHECK-NEXT: [[I_011:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[S_010:%.*]] = phi i64 [ [[ADD]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ] -; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[X]], i32 [[I_011]] +; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i16, ptr [[X]], i32 [[I_011]] ; CHECK-NEXT: [[TMP7:%.*]] = load i16, ptr [[ARRAYIDX]], align 2 ; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP7]] to i32 ; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[CONV]], [[CONV]] @@ -1558,7 +1558,7 @@ define i64 @mla_and_add_together_16_64(ptr nocapture noundef readonly %x, i32 no ; CHECK-NEXT: [[I_019:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[T_018:%.*]] = phi i64 [ [[ADD]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[S_017:%.*]] = phi i32 [ [[ADD6]], [[FOR_BODY]] ], [ [[BC_MERGE_RDX2]], [[SCALAR_PH]] ] -; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[X]], i32 [[I_019]] +; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i16, ptr [[X]], i32 [[I_019]] ; CHECK-NEXT: [[TMP9:%.*]] = load i16, ptr [[ARRAYIDX]], align 2 ; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP9]] to i32 ; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[CONV]], [[CONV]] @@ -1606,20 +1606,20 @@ define i64 @interleave_doublereduct_i16_i64(ptr %x, ptr %y, i32 %n) { ; CHECK: for.body: ; CHECK-NEXT: [[I_025:%.*]] = phi i32 [ [[ADD13:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY]] ] ; CHECK-NEXT: [[T_024:%.*]] = phi i64 [ [[ADD12]], [[FOR_BODY]] ], [ 0, [[ENTRY]] ] -; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[X:%.*]], i32 [[I_025]] +; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i16, ptr [[X:%.*]], i32 [[I_025]] ; CHECK-NEXT: [[TMP0:%.*]] = load i16, ptr [[ARRAYIDX]], align 2 ; CHECK-NEXT: [[CONV:%.*]] = sext i16 [[TMP0]] to i32 -; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i16, ptr [[Y:%.*]], i32 [[I_025]] +; CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i16, ptr [[Y:%.*]], i32 [[I_025]] ; CHECK-NEXT: [[TMP1:%.*]] = load i16, ptr [[ARRAYIDX1]], align 2 ; CHECK-NEXT: [[CONV2:%.*]] = sext i16 [[TMP1]] to i32 ; CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[CONV2]], [[CONV]] ; CHECK-NEXT: [[CONV3:%.*]] = sext i32 [[MUL]] to i64 ; CHECK-NEXT: [[ADD:%.*]] = add nsw i64 [[T_024]], [[CONV3]] ; CHECK-NEXT: [[ADD4:%.*]] = or disjoint i32 [[I_025]], 1 -; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, ptr [[X]], i32 [[ADD4]] +; CHECK-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds nuw i16, ptr [[X]], i32 [[ADD4]] ; CHECK-NEXT: [[TMP2:%.*]] = load i16, ptr [[ARRAYIDX5]], align 2 ; CHECK-NEXT: [[CONV6:%.*]] = sext i16 [[TMP2]] to i32 -; CHECK-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i16, ptr [[Y]], i32 [[ADD4]] +; CHECK-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds nuw i16, ptr [[Y]], i32 [[ADD4]] ; CHECK-NEXT: [[TMP3:%.*]] = load i16, ptr [[ARRAYIDX8]], align 2 ; CHECK-NEXT: [[CONV9:%.*]] = sext i16 [[TMP3]] to i32 ; CHECK-NEXT: [[MUL10:%.*]] = mul nsw i32 [[CONV9]], [[CONV6]] @@ -1677,7 +1677,7 @@ define i64 @test_std_q31(ptr %x, i32 %n) #0 { ; CHECK-NEXT: [[S_014:%.*]] = phi i64 [ [[ADD]], [[FOR_BODY1]] ], [ 0, [[ENTRY:%.*]] ] ; CHECK-NEXT: [[I_013:%.*]] = phi i32 [ [[ADD4:%.*]], [[FOR_BODY1]] ], [ 0, [[ENTRY]] ] ; CHECK-NEXT: [[T_012:%.*]] = phi i64 [ [[ADD3]], [[FOR_BODY1]] ], [ 0, [[ENTRY]] ] -; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[X:%.*]], i32 [[I_013]] +; CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[X:%.*]], i32 [[I_013]] ; CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 ; CHECK-NEXT: [[SHR:%.*]] = ashr i32 [[TMP0]], 8 ; CHECK-NEXT: [[CONV:%.*]] = sext i32 [[SHR]] to i64 diff --git a/llvm/test/Transforms/LoopVectorize/ARM/mve-selectandorcost.ll b/llvm/test/Transforms/LoopVectorize/ARM/mve-selectandorcost.ll index 025b8738f782..eaec02038fde 100644 --- a/llvm/test/Transforms/LoopVectorize/ARM/mve-selectandorcost.ll +++ b/llvm/test/Transforms/LoopVectorize/ARM/mve-selectandorcost.ll @@ -65,9 +65,9 @@ define float @test(ptr nocapture readonly %pA, ptr nocapture readonly %pB, i32 % ; CHECK-NEXT: [[PB_ADDR_019:%.*]] = phi ptr [ [[INCDEC_PTR1:%.*]], [[IF_END]] ], [ [[BC_RESUME_VAL2]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[BLOCKSIZE_ADDR_018:%.*]] = phi i32 [ [[DEC:%.*]], [[IF_END]] ], [ [[BC_RESUME_VAL4]], [[SCALAR_PH]] ] ; CHECK-NEXT: [[ACCUM_017:%.*]] = phi float [ [[ACCUM_1:%.*]], [[IF_END]] ], [ [[BC_MERGE_RDX]], [[SCALAR_PH]] ] -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[PA_ADDR_020]], i32 4 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[PA_ADDR_020]], i32 4 ; CHECK-NEXT: [[TMP13:%.*]] = load float, ptr [[PA_ADDR_020]], align 4 -; CHECK-NEXT: [[INCDEC_PTR1]] = getelementptr inbounds i8, ptr [[PB_ADDR_019]], i32 4 +; CHECK-NEXT: [[INCDEC_PTR1]] = getelementptr inbounds nuw i8, ptr [[PB_ADDR_019]], i32 4 ; CHECK-NEXT: [[TMP14:%.*]] = load float, ptr [[PB_ADDR_019]], align 4 ; CHECK-NEXT: [[CMP2:%.*]] = fcmp fast une float [[TMP13]], 0.000000e+00 ; CHECK-NEXT: [[CMP3:%.*]] = fcmp fast une float [[TMP14]], 0.000000e+00 diff --git a/llvm/test/Transforms/LoopVectorize/ARM/pointer_iv.ll b/llvm/test/Transforms/LoopVectorize/ARM/pointer_iv.ll index d49639db3bbf..07ba7745b34a 100644 --- a/llvm/test/Transforms/LoopVectorize/ARM/pointer_iv.ll +++ b/llvm/test/Transforms/LoopVectorize/ARM/pointer_iv.ll @@ -69,10 +69,10 @@ define hidden void @pointer_phi_v4i32_add2(ptr noalias nocapture readonly %A, pt ; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 996, [[VECTOR_BODY]] ] ; CHECK-NEXT: [[B_ADDR_07:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ] ; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[A_ADDR_09]], align 4 -; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_09]], i32 8 +; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds nuw i8, ptr [[A_ADDR_09]], i32 8 ; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP2]], [[Y]] ; CHECK-NEXT: store i32 [[ADD]], ptr [[B_ADDR_07]], align 4 -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_07]], i32 4 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[B_ADDR_07]], i32 4 ; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1 ; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000 ; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]] @@ -123,10 +123,10 @@ define hidden void @pointer_phi_v4i32_add3(ptr noalias nocapture readonly %A, pt ; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 996, [[VECTOR_BODY]] ] ; CHECK-NEXT: [[B_ADDR_07:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ] ; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[A_ADDR_09]], align 4 -; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_09]], i32 12 +; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds nuw i8, ptr [[A_ADDR_09]], i32 12 ; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP3]], [[Y]] ; CHECK-NEXT: store i32 [[ADD]], ptr [[B_ADDR_07]], align 4 -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_07]], i32 4 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[B_ADDR_07]], i32 4 ; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1 ; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000 ; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]] @@ -219,10 +219,10 @@ define hidden void @pointer_phi_v8i16_add2(ptr noalias nocapture readonly %A, pt ; CHECK-NEXT: [[I_010:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 992, [[VECTOR_BODY]] ] ; CHECK-NEXT: [[B_ADDR_09:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ] ; CHECK-NEXT: [[L1:%.*]] = load i16, ptr [[A_ADDR_011]], align 2 -; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_011]], i32 4 +; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds nuw i8, ptr [[A_ADDR_011]], i32 4 ; CHECK-NEXT: [[CONV1:%.*]] = add i16 [[L1]], [[TMP0]] ; CHECK-NEXT: store i16 [[CONV1]], ptr [[B_ADDR_09]], align 2 -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_09]], i32 2 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[B_ADDR_09]], i32 2 ; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_010]], 1 ; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000 ; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP9:![0-9]+]] @@ -258,10 +258,10 @@ define hidden void @pointer_phi_v8i16_add3(ptr noalias nocapture readonly %A, pt ; CHECK-NEXT: [[I_010:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[INC:%.*]], [[FOR_BODY]] ] ; CHECK-NEXT: [[B_ADDR_09:%.*]] = phi ptr [ [[B:%.*]], [[ENTRY]] ], [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ] ; CHECK-NEXT: [[L1:%.*]] = load i16, ptr [[A_ADDR_011]], align 2 -; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_011]], i32 6 +; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds nuw i8, ptr [[A_ADDR_011]], i32 6 ; CHECK-NEXT: [[CONV1:%.*]] = add i16 [[L1]], [[TMP0]] ; CHECK-NEXT: store i16 [[CONV1]], ptr [[B_ADDR_09]], align 2 -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_09]], i32 2 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[B_ADDR_09]], i32 2 ; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_010]], 1 ; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000 ; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]] @@ -311,10 +311,10 @@ define hidden void @pointer_phi_v16i8_add1(ptr noalias nocapture readonly %A, pt ; CHECK-NEXT: [[I_09:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 992, [[VECTOR_BODY]] ] ; CHECK-NEXT: [[B_ADDR_08:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ] ; CHECK-NEXT: [[TMP3:%.*]] = load i8, ptr [[A_ADDR_010]], align 1 -; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_010]], i32 1 +; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds nuw i8, ptr [[A_ADDR_010]], i32 1 ; CHECK-NEXT: [[CONV1:%.*]] = add i8 [[TMP3]], [[TMP0]] ; CHECK-NEXT: store i8 [[CONV1]], ptr [[B_ADDR_08]], align 1 -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_08]], i32 1 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[B_ADDR_08]], i32 1 ; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_09]], 1 ; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000 ; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP11:![0-9]+]] @@ -367,10 +367,10 @@ define hidden void @pointer_phi_v16i8_add2(ptr noalias nocapture readonly %A, pt ; CHECK-NEXT: [[I_09:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 992, [[VECTOR_BODY]] ] ; CHECK-NEXT: [[B_ADDR_08:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ] ; CHECK-NEXT: [[TMP3:%.*]] = load i8, ptr [[A_ADDR_010]], align 1 -; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_010]], i32 2 +; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds nuw i8, ptr [[A_ADDR_010]], i32 2 ; CHECK-NEXT: [[CONV1:%.*]] = add i8 [[TMP3]], [[TMP0]] ; CHECK-NEXT: store i8 [[CONV1]], ptr [[B_ADDR_08]], align 1 -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_08]], i32 1 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[B_ADDR_08]], i32 1 ; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_09]], 1 ; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000 ; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP13:![0-9]+]] @@ -407,10 +407,10 @@ define hidden void @pointer_phi_v16i8_add3(ptr noalias nocapture readonly %A, pt ; CHECK-NEXT: [[I_09:%.*]] = phi i32 [ 0, [[ENTRY]] ], [ [[INC:%.*]], [[FOR_BODY]] ] ; CHECK-NEXT: [[B_ADDR_08:%.*]] = phi ptr [ [[B:%.*]], [[ENTRY]] ], [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ] ; CHECK-NEXT: [[TMP1:%.*]] = load i8, ptr [[A_ADDR_010]], align 1 -; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_010]], i32 3 +; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds nuw i8, ptr [[A_ADDR_010]], i32 3 ; CHECK-NEXT: [[CONV1:%.*]] = add i8 [[TMP1]], [[TMP0]] ; CHECK-NEXT: store i8 [[CONV1]], ptr [[B_ADDR_08]], align 1 -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_08]], i32 1 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[B_ADDR_08]], i32 1 ; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_09]], 1 ; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000 ; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]] @@ -502,10 +502,10 @@ define hidden void @pointer_phi_v4f32_add2(ptr noalias nocapture readonly %A, pt ; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 996, [[VECTOR_BODY]] ] ; CHECK-NEXT: [[B_ADDR_07:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ] ; CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[A_ADDR_09]], align 4 -; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_09]], i32 8 +; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds nuw i8, ptr [[A_ADDR_09]], i32 8 ; CHECK-NEXT: [[ADD:%.*]] = fadd fast float [[TMP2]], [[Y]] ; CHECK-NEXT: store float [[ADD]], ptr [[B_ADDR_07]], align 4 -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_07]], i32 4 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[B_ADDR_07]], i32 4 ; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1 ; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000 ; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP16:![0-9]+]] @@ -556,10 +556,10 @@ define hidden void @pointer_phi_v4f32_add3(ptr noalias nocapture readonly %A, pt ; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 996, [[VECTOR_BODY]] ] ; CHECK-NEXT: [[B_ADDR_07:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ] ; CHECK-NEXT: [[TMP3:%.*]] = load float, ptr [[A_ADDR_09]], align 4 -; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_09]], i32 12 +; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds nuw i8, ptr [[A_ADDR_09]], i32 12 ; CHECK-NEXT: [[ADD:%.*]] = fadd fast float [[TMP3]], [[Y]] ; CHECK-NEXT: store float [[ADD]], ptr [[B_ADDR_07]], align 4 -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_07]], i32 4 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[B_ADDR_07]], i32 4 ; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1 ; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000 ; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP18:![0-9]+]] @@ -649,10 +649,10 @@ define hidden void @pointer_phi_v4half_add2(ptr noalias nocapture readonly %A, p ; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 992, [[VECTOR_BODY]] ] ; CHECK-NEXT: [[B_ADDR_07:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ] ; CHECK-NEXT: [[TMP2:%.*]] = load half, ptr [[A_ADDR_09]], align 4 -; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_09]], i32 4 +; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds nuw i8, ptr [[A_ADDR_09]], i32 4 ; CHECK-NEXT: [[ADD:%.*]] = fadd fast half [[TMP2]], [[Y]] ; CHECK-NEXT: store half [[ADD]], ptr [[B_ADDR_07]], align 4 -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_07]], i32 2 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[B_ADDR_07]], i32 2 ; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1 ; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000 ; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP21:![0-9]+]] @@ -703,10 +703,10 @@ define hidden void @pointer_phi_v4half_add3(ptr noalias nocapture readonly %A, p ; CHECK-NEXT: [[I_08:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 992, [[VECTOR_BODY]] ] ; CHECK-NEXT: [[B_ADDR_07:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ] ; CHECK-NEXT: [[TMP2:%.*]] = load half, ptr [[A_ADDR_09]], align 4 -; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_09]], i32 6 +; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds nuw i8, ptr [[A_ADDR_09]], i32 6 ; CHECK-NEXT: [[ADD:%.*]] = fadd fast half [[TMP2]], [[Y]] ; CHECK-NEXT: store half [[ADD]], ptr [[B_ADDR_07]], align 4 -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_07]], i32 2 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[B_ADDR_07]], i32 2 ; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_08]], 1 ; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000 ; CHECK-NEXT: br i1 [[EXITCOND]], label [[END:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP23:![0-9]+]] @@ -767,10 +767,10 @@ define hidden void @pointer_phi_v4i32_uf2(ptr noalias nocapture readonly %A, ptr ; CHECK-NEXT: [[I_07:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 9992, [[VECTOR_BODY]] ] ; CHECK-NEXT: [[B_ADDR_06:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ] ; CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[A_ADDR_08]], align 4 -; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_08]], i32 24 +; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds nuw i8, ptr [[A_ADDR_08]], i32 24 ; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], [[Y]] ; CHECK-NEXT: store i32 [[ADD]], ptr [[B_ADDR_06]], align 4 -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_06]], i32 4 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[B_ADDR_06]], i32 4 ; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_07]], 1 ; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 10000 ; CHECK-NEXT: br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP25:![0-9]+]] @@ -842,10 +842,10 @@ define hidden void @pointer_phi_v4i32_uf4(ptr noalias nocapture readonly %A, ptr ; CHECK-NEXT: [[I_07:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 9984, [[VECTOR_BODY]] ] ; CHECK-NEXT: [[B_ADDR_06:%.*]] = phi ptr [ [[INCDEC_PTR:%.*]], [[FOR_BODY]] ], [ [[IND_END2]], [[VECTOR_BODY]] ] ; CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[A_ADDR_08]], align 4 -; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds i8, ptr [[A_ADDR_08]], i32 24 +; CHECK-NEXT: [[ADD_PTR]] = getelementptr inbounds nuw i8, ptr [[A_ADDR_08]], i32 24 ; CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[Y]] ; CHECK-NEXT: store i32 [[ADD]], ptr [[B_ADDR_06]], align 4 -; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds i8, ptr [[B_ADDR_06]], i32 4 +; CHECK-NEXT: [[INCDEC_PTR]] = getelementptr inbounds nuw i8, ptr [[B_ADDR_06]], i32 4 ; CHECK-NEXT: [[INC]] = add nuw nsw i32 [[I_07]], 1 ; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 10000 ; CHECK-NEXT: br i1 [[EXITCOND]], label [[FOR_COND_CLEANUP:%.*]], label [[FOR_BODY]], !llvm.loop [[LOOP27:![0-9]+]] @@ -907,20 +907,20 @@ define hidden void @mult_ptr_iv(ptr noalias nocapture readonly %x, ptr noalias n ; CHECK-NEXT: [[X_ADDR_050:%.*]] = phi ptr [ [[INCDEC_PTR2:%.*]], [[FOR_BODY]] ], [ [[X]], [[ENTRY]] ] ; CHECK-NEXT: [[Z_ADDR_049:%.*]] = phi ptr [ [[INCDEC_PTR34:%.*]], [[FOR_BODY]] ], [ [[Z]], [[ENTRY]] ] ; CHECK-NEXT: [[I_048:%.*]] = phi i32 [ [[INC:%.*]], [[FOR_BODY]] ], [ 0, [[ENTRY]] ] -; CHECK-NEXT: [[INCDEC_PTR:%.*]] = getelementptr inbounds i8, ptr [[X_ADDR_050]], i32 1 +; CHECK-NEXT: [[INCDEC_PTR:%.*]] = getelementptr inbounds nuw i8, ptr [[X_ADDR_050]], i32 1 ; CHECK-NEXT: [[TMP10:%.*]] = load i8, ptr [[X_ADDR_050]], align 1 -; CHECK-NEXT: [[INCDEC_PTR1:%.*]] = getelementptr inbounds i8, ptr [[X_ADDR_050]], i32 2 +; CHECK-NEXT: [[INCDEC_PTR1:%.*]] = getelementptr inbounds nuw i8, ptr [[X_ADDR_050]], i32 2 ; CHECK-NEXT: [[TMP11:%.*]] = load i8, ptr [[INCDEC_PTR]], align 1 -; CHECK-NEXT: [[INCDEC_PTR2]] = getelementptr inbounds i8, ptr [[X_ADDR_050]], i32 3 +; CHECK-NEXT: [[INCDEC_PTR2]] = getelementptr inbounds nuw i8, ptr [[X_ADDR_050]], i32 3 ; CHECK-NEXT: [[TMP12:%.*]] = load i8, ptr [[INCDEC_PTR1]], align 1 ; CHECK-NEXT: [[MUL:%.*]] = mul i8 [[TMP10]], 10 ; CHECK-NEXT: [[MUL1:%.*]] = mul i8 [[TMP10]], [[TMP11]] ; CHECK-NEXT: [[MUL2:%.*]] = mul i8 [[TMP10]], [[TMP12]] -; CHECK-NEXT: [[INCDEC_PTR32:%.*]] = getelementptr inbounds i8, ptr [[Z_ADDR_049]], i32 1 +; CHECK-NEXT: [[INCDEC_PTR32:%.*]] = getelementptr inbounds nuw i8, ptr [[Z_ADDR_049]], i32 1 ; CHECK-NEXT: store i8 [[MUL]], ptr [[Z_ADDR_049]], align 1 -; CHECK-NEXT: [[INCDEC_PTR33:%.*]] = getelementptr inbounds i8, ptr [[Z_ADDR_049]], i32 2 +; CHECK-NEXT: [[INCDEC_PTR33:%.*]] = getelementptr inbounds nuw i8, ptr [[Z_ADDR_049]], i32 2 ; CHECK-NEXT: store i8 [[MUL1]], ptr [[INCDEC_PTR32]], align 1 -; CHECK-NEXT: [[INCDEC_PTR34]] = getelementptr inbounds i8, ptr [[Z_ADDR_049]], i32 3 +; CHECK-NEXT: [[INCDEC_PTR34]] = getelementptr inbounds nuw i8, ptr [[Z_ADDR_049]], i32 3 ; CHECK-NEXT: store i8 [[MUL2]], ptr [[INCDEC_PTR33]], align 1 ; CHECK-NEXT: [[INC]] = add nuw i32 [[I_048]], 1 ; CHECK-NEXT: [[EXITCOND:%.*]] = icmp eq i32 [[INC]], 1000 |