aboutsummaryrefslogtreecommitdiffstats
path: root/test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll
diff options
context:
space:
mode:
Diffstat (limited to 'test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll')
-rw-r--r--test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll136
1 files changed, 68 insertions, 68 deletions
diff --git a/test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll b/test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll
index 26b2940..2ad6c2e 100644
--- a/test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll
+++ b/test/Transforms/LoopStrengthReduce/ARM/ivchain-ARM.ll
@@ -18,18 +18,18 @@ entry:
loop:
%iv = phi i32* [ %a, %entry ], [ %iv4, %loop ]
%s = phi i32 [ 0, %entry ], [ %s4, %loop ]
- %v = load i32* %iv
- %iv1 = getelementptr inbounds i32* %iv, i32 %x
- %v1 = load i32* %iv1
- %iv2 = getelementptr inbounds i32* %iv1, i32 %x
- %v2 = load i32* %iv2
- %iv3 = getelementptr inbounds i32* %iv2, i32 %x
- %v3 = load i32* %iv3
+ %v = load i32, i32* %iv
+ %iv1 = getelementptr inbounds i32, i32* %iv, i32 %x
+ %v1 = load i32, i32* %iv1
+ %iv2 = getelementptr inbounds i32, i32* %iv1, i32 %x
+ %v2 = load i32, i32* %iv2
+ %iv3 = getelementptr inbounds i32, i32* %iv2, i32 %x
+ %v3 = load i32, i32* %iv3
%s1 = add i32 %s, %v
%s2 = add i32 %s1, %v1
%s3 = add i32 %s2, %v2
%s4 = add i32 %s3, %v3
- %iv4 = getelementptr inbounds i32* %iv3, i32 %x
+ %iv4 = getelementptr inbounds i32, i32* %iv3, i32 %x
%cmp = icmp eq i32* %iv4, %b
br i1 %cmp, label %exit, label %loop
exit:
@@ -52,18 +52,18 @@ entry:
loop:
%iv = phi i32* [ %a, %entry ], [ %iv4, %loop ]
%s = phi i32 [ 0, %entry ], [ %s4, %loop ]
- %v = load i32* %iv
- %iv1 = getelementptr inbounds i32* %iv, i32 %x
- %v1 = load i32* %iv1
- %iv2 = getelementptr inbounds i32* %iv1, i32 %x
- %v2 = load i32* %iv2
- %iv3 = getelementptr inbounds i32* %iv2, i32 %x
- %v3 = load i32* %iv3
+ %v = load i32, i32* %iv
+ %iv1 = getelementptr inbounds i32, i32* %iv, i32 %x
+ %v1 = load i32, i32* %iv1
+ %iv2 = getelementptr inbounds i32, i32* %iv1, i32 %x
+ %v2 = load i32, i32* %iv2
+ %iv3 = getelementptr inbounds i32, i32* %iv2, i32 %x
+ %v3 = load i32, i32* %iv3
%s1 = add i32 %s, %v
%s2 = add i32 %s1, %v1
%s3 = add i32 %s2, %v2
%s4 = add i32 %s3, %v3
- %iv4 = getelementptr inbounds i32* %iv3, i32 %x
+ %iv4 = getelementptr inbounds i32, i32* %iv3, i32 %x
store i32 %s4, i32* %iv
%cmp = icmp eq i32* %iv4, %b
br i1 %cmp, label %exit, label %loop
@@ -103,26 +103,26 @@ for.body: ; preds = %for.body.lr.ph, %fo
%i.010 = phi i32 [ 0, %for.body.lr.ph ], [ %inc, %for.body ]
%res.addr.09 = phi i32* [ %res, %for.body.lr.ph ], [ %add.ptr7, %for.body ]
%0 = bitcast i8* %main.addr.011 to i32*
- %1 = load i32* %0, align 4
- %add.ptr = getelementptr inbounds i8* %main.addr.011, i32 %main_stride
+ %1 = load i32, i32* %0, align 4
+ %add.ptr = getelementptr inbounds i8, i8* %main.addr.011, i32 %main_stride
%2 = bitcast i8* %add.ptr to i32*
- %3 = load i32* %2, align 4
- %add.ptr1 = getelementptr inbounds i8* %main.addr.011, i32 %add.ptr.sum
+ %3 = load i32, i32* %2, align 4
+ %add.ptr1 = getelementptr inbounds i8, i8* %main.addr.011, i32 %add.ptr.sum
%4 = bitcast i8* %add.ptr1 to i32*
- %5 = load i32* %4, align 4
- %add.ptr2 = getelementptr inbounds i8* %main.addr.011, i32 %add.ptr1.sum
+ %5 = load i32, i32* %4, align 4
+ %add.ptr2 = getelementptr inbounds i8, i8* %main.addr.011, i32 %add.ptr1.sum
%6 = bitcast i8* %add.ptr2 to i32*
- %7 = load i32* %6, align 4
- %add.ptr3 = getelementptr inbounds i8* %main.addr.011, i32 %add.ptr4.sum
+ %7 = load i32, i32* %6, align 4
+ %add.ptr3 = getelementptr inbounds i8, i8* %main.addr.011, i32 %add.ptr4.sum
%8 = bitcast i8* %add.ptr3 to i32*
- %9 = load i32* %8, align 4
+ %9 = load i32, i32* %8, align 4
%add = add i32 %3, %1
%add4 = add i32 %add, %5
%add5 = add i32 %add4, %7
%add6 = add i32 %add5, %9
store i32 %add6, i32* %res.addr.09, align 4
- %add.ptr6 = getelementptr inbounds i8* %main.addr.011, i32 %add.ptr3.sum
- %add.ptr7 = getelementptr inbounds i32* %res.addr.09, i32 %y
+ %add.ptr6 = getelementptr inbounds i8, i8* %main.addr.011, i32 %add.ptr3.sum
+ %add.ptr7 = getelementptr inbounds i32, i32* %res.addr.09, i32 %y
%inc = add i32 %i.010, 1
%cmp = icmp eq i32 %inc, %z
br i1 %cmp, label %for.end, label %for.body
@@ -146,48 +146,48 @@ entry:
for.body: ; preds = %for.body, %entry
%i.07 = phi i32 [ 0, %entry ], [ %inc.3, %for.body ]
- %arrayidx = getelementptr inbounds i8* %a, i32 %i.07
- %0 = load i8* %arrayidx, align 1
+ %arrayidx = getelementptr inbounds i8, i8* %a, i32 %i.07
+ %0 = load i8, i8* %arrayidx, align 1
%conv5 = zext i8 %0 to i32
- %arrayidx1 = getelementptr inbounds i8* %b, i32 %i.07
- %1 = load i8* %arrayidx1, align 1
+ %arrayidx1 = getelementptr inbounds i8, i8* %b, i32 %i.07
+ %1 = load i8, i8* %arrayidx1, align 1
%conv26 = zext i8 %1 to i32
%add = add nsw i32 %conv26, %conv5
%conv3 = trunc i32 %add to i8
- %arrayidx4 = getelementptr inbounds i8* %c, i32 %i.07
+ %arrayidx4 = getelementptr inbounds i8, i8* %c, i32 %i.07
store i8 %conv3, i8* %arrayidx4, align 1
%inc1 = or i32 %i.07, 1
- %arrayidx.1 = getelementptr inbounds i8* %a, i32 %inc1
- %2 = load i8* %arrayidx.1, align 1
+ %arrayidx.1 = getelementptr inbounds i8, i8* %a, i32 %inc1
+ %2 = load i8, i8* %arrayidx.1, align 1
%conv5.1 = zext i8 %2 to i32
- %arrayidx1.1 = getelementptr inbounds i8* %b, i32 %inc1
- %3 = load i8* %arrayidx1.1, align 1
+ %arrayidx1.1 = getelementptr inbounds i8, i8* %b, i32 %inc1
+ %3 = load i8, i8* %arrayidx1.1, align 1
%conv26.1 = zext i8 %3 to i32
%add.1 = add nsw i32 %conv26.1, %conv5.1
%conv3.1 = trunc i32 %add.1 to i8
- %arrayidx4.1 = getelementptr inbounds i8* %c, i32 %inc1
+ %arrayidx4.1 = getelementptr inbounds i8, i8* %c, i32 %inc1
store i8 %conv3.1, i8* %arrayidx4.1, align 1
%inc.12 = or i32 %i.07, 2
- %arrayidx.2 = getelementptr inbounds i8* %a, i32 %inc.12
- %4 = load i8* %arrayidx.2, align 1
+ %arrayidx.2 = getelementptr inbounds i8, i8* %a, i32 %inc.12
+ %4 = load i8, i8* %arrayidx.2, align 1
%conv5.2 = zext i8 %4 to i32
- %arrayidx1.2 = getelementptr inbounds i8* %b, i32 %inc.12
- %5 = load i8* %arrayidx1.2, align 1
+ %arrayidx1.2 = getelementptr inbounds i8, i8* %b, i32 %inc.12
+ %5 = load i8, i8* %arrayidx1.2, align 1
%conv26.2 = zext i8 %5 to i32
%add.2 = add nsw i32 %conv26.2, %conv5.2
%conv3.2 = trunc i32 %add.2 to i8
- %arrayidx4.2 = getelementptr inbounds i8* %c, i32 %inc.12
+ %arrayidx4.2 = getelementptr inbounds i8, i8* %c, i32 %inc.12
store i8 %conv3.2, i8* %arrayidx4.2, align 1
%inc.23 = or i32 %i.07, 3
- %arrayidx.3 = getelementptr inbounds i8* %a, i32 %inc.23
- %6 = load i8* %arrayidx.3, align 1
+ %arrayidx.3 = getelementptr inbounds i8, i8* %a, i32 %inc.23
+ %6 = load i8, i8* %arrayidx.3, align 1
%conv5.3 = zext i8 %6 to i32
- %arrayidx1.3 = getelementptr inbounds i8* %b, i32 %inc.23
- %7 = load i8* %arrayidx1.3, align 1
+ %arrayidx1.3 = getelementptr inbounds i8, i8* %b, i32 %inc.23
+ %7 = load i8, i8* %arrayidx1.3, align 1
%conv26.3 = zext i8 %7 to i32
%add.3 = add nsw i32 %conv26.3, %conv5.3
%conv3.3 = trunc i32 %add.3 to i8
- %arrayidx4.3 = getelementptr inbounds i8* %c, i32 %inc.23
+ %arrayidx4.3 = getelementptr inbounds i8, i8* %c, i32 %inc.23
store i8 %conv3.3, i8* %arrayidx4.3, align 1
%inc.3 = add nsw i32 %i.07, 4
%exitcond.3 = icmp eq i32 %inc.3, 400
@@ -240,31 +240,31 @@ define hidden void @testNeon(i8* %ref_data, i32 %ref_stride, i32 %limit, <16 x i
%result.03 = phi <16 x i8> [ zeroinitializer, %.lr.ph ], [ %41, %11 ]
%.012 = phi <16 x i8>* [ %data, %.lr.ph ], [ %43, %11 ]
%12 = tail call <1 x i64> @llvm.arm.neon.vld1.v1i64(i8* %.05, i32 1) nounwind
- %13 = getelementptr inbounds i8* %.05, i32 %ref_stride
+ %13 = getelementptr inbounds i8, i8* %.05, i32 %ref_stride
%14 = tail call <1 x i64> @llvm.arm.neon.vld1.v1i64(i8* %13, i32 1) nounwind
%15 = shufflevector <1 x i64> %12, <1 x i64> %14, <2 x i32> <i32 0, i32 1>
%16 = bitcast <2 x i64> %15 to <16 x i8>
- %17 = getelementptr inbounds <16 x i8>* %.012, i32 1
+ %17 = getelementptr inbounds <16 x i8>, <16 x i8>* %.012, i32 1
store <16 x i8> %16, <16 x i8>* %.012, align 4
- %18 = getelementptr inbounds i8* %.05, i32 %2
+ %18 = getelementptr inbounds i8, i8* %.05, i32 %2
%19 = tail call <1 x i64> @llvm.arm.neon.vld1.v1i64(i8* %18, i32 1) nounwind
- %20 = getelementptr inbounds i8* %.05, i32 %3
+ %20 = getelementptr inbounds i8, i8* %.05, i32 %3
%21 = tail call <1 x i64> @llvm.arm.neon.vld1.v1i64(i8* %20, i32 1) nounwind
%22 = shufflevector <1 x i64> %19, <1 x i64> %21, <2 x i32> <i32 0, i32 1>
%23 = bitcast <2 x i64> %22 to <16 x i8>
- %24 = getelementptr inbounds <16 x i8>* %.012, i32 2
+ %24 = getelementptr inbounds <16 x i8>, <16 x i8>* %.012, i32 2
store <16 x i8> %23, <16 x i8>* %17, align 4
- %25 = getelementptr inbounds i8* %.05, i32 %4
+ %25 = getelementptr inbounds i8, i8* %.05, i32 %4
%26 = tail call <1 x i64> @llvm.arm.neon.vld1.v1i64(i8* %25, i32 1) nounwind
- %27 = getelementptr inbounds i8* %.05, i32 %5
+ %27 = getelementptr inbounds i8, i8* %.05, i32 %5
%28 = tail call <1 x i64> @llvm.arm.neon.vld1.v1i64(i8* %27, i32 1) nounwind
%29 = shufflevector <1 x i64> %26, <1 x i64> %28, <2 x i32> <i32 0, i32 1>
%30 = bitcast <2 x i64> %29 to <16 x i8>
- %31 = getelementptr inbounds <16 x i8>* %.012, i32 3
+ %31 = getelementptr inbounds <16 x i8>, <16 x i8>* %.012, i32 3
store <16 x i8> %30, <16 x i8>* %24, align 4
- %32 = getelementptr inbounds i8* %.05, i32 %6
+ %32 = getelementptr inbounds i8, i8* %.05, i32 %6
%33 = tail call <1 x i64> @llvm.arm.neon.vld1.v1i64(i8* %32, i32 1) nounwind
- %34 = getelementptr inbounds i8* %.05, i32 %7
+ %34 = getelementptr inbounds i8, i8* %.05, i32 %7
%35 = tail call <1 x i64> @llvm.arm.neon.vld1.v1i64(i8* %34, i32 1) nounwind
%36 = shufflevector <1 x i64> %33, <1 x i64> %35, <2 x i32> <i32 0, i32 1>
%37 = bitcast <2 x i64> %36 to <16 x i8>
@@ -273,14 +273,14 @@ define hidden void @testNeon(i8* %ref_data, i32 %ref_stride, i32 %limit, <16 x i
%39 = add <16 x i8> %38, %30
%40 = add <16 x i8> %39, %37
%41 = add <16 x i8> %result.03, %40
- %42 = getelementptr i8* %.05, i32 %9
- %43 = getelementptr inbounds <16 x i8>* %.012, i32 -64
+ %42 = getelementptr i8, i8* %.05, i32 %9
+ %43 = getelementptr inbounds <16 x i8>, <16 x i8>* %.012, i32 -64
%44 = add nsw i32 %counter.04, 1
%exitcond = icmp eq i32 %44, %limit
br i1 %exitcond, label %._crit_edge, label %11
._crit_edge: ; preds = %11
- %scevgep = getelementptr <16 x i8>* %data, i32 %10
+ %scevgep = getelementptr <16 x i8>, <16 x i8>* %data, i32 %10
br label %45
; <label>:45 ; preds = %._crit_edge, %0
@@ -327,20 +327,20 @@ entry:
for.body: ; preds = %for.body, %entry
%i.0110 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
%src.addr = phi i8* [ %src, %entry ], [ %add.ptr45, %for.body ]
- %add.ptr = getelementptr inbounds i8* %src.addr, i32 %idx.neg
+ %add.ptr = getelementptr inbounds i8, i8* %src.addr, i32 %idx.neg
%vld1 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr, i32 1)
- %add.ptr3 = getelementptr inbounds i8* %src.addr, i32 %idx.neg2
+ %add.ptr3 = getelementptr inbounds i8, i8* %src.addr, i32 %idx.neg2
%vld2 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr3, i32 1)
- %add.ptr7 = getelementptr inbounds i8* %src.addr, i32 %idx.neg6
+ %add.ptr7 = getelementptr inbounds i8, i8* %src.addr, i32 %idx.neg6
%vld3 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr7, i32 1)
- %add.ptr11 = getelementptr inbounds i8* %src.addr, i32 %idx.neg10
+ %add.ptr11 = getelementptr inbounds i8, i8* %src.addr, i32 %idx.neg10
%vld4 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr11, i32 1)
%vld5 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %src.addr, i32 1)
- %add.ptr17 = getelementptr inbounds i8* %src.addr, i32 %stride
+ %add.ptr17 = getelementptr inbounds i8, i8* %src.addr, i32 %stride
%vld6 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr17, i32 1)
- %add.ptr20 = getelementptr inbounds i8* %src.addr, i32 %mul5
+ %add.ptr20 = getelementptr inbounds i8, i8* %src.addr, i32 %mul5
%vld7 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr20, i32 1)
- %add.ptr23 = getelementptr inbounds i8* %src.addr, i32 %mul1
+ %add.ptr23 = getelementptr inbounds i8, i8* %src.addr, i32 %mul1
%vld8 = tail call <8 x i8> @llvm.arm.neon.vld1.v8i8(i8* %add.ptr23, i32 1)
%vadd1 = tail call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %vld1, <8 x i8> %vld2) nounwind
%vadd2 = tail call <8 x i8> @llvm.arm.neon.vhaddu.v8i8(<8 x i8> %vld2, <8 x i8> %vld3) nounwind
@@ -355,7 +355,7 @@ for.body: ; preds = %for.body, %entry
tail call void @llvm.arm.neon.vst1.v8i8(i8* %add.ptr17, <8 x i8> %vadd5, i32 1)
tail call void @llvm.arm.neon.vst1.v8i8(i8* %add.ptr20, <8 x i8> %vadd6, i32 1)
%inc = add nsw i32 %i.0110, 1
- %add.ptr45 = getelementptr inbounds i8* %src.addr, i32 8
+ %add.ptr45 = getelementptr inbounds i8, i8* %src.addr, i32 8
%exitcond = icmp eq i32 %inc, 4
br i1 %exitcond, label %for.end, label %for.body