CodeGen peephole: fold redundant phys reg copies

[oota-llvm.git] / test / CodeGen / X86 / scalar_widen_div.ll

diff --git a/test/CodeGen/X86/scalar_widen_div.ll b/test/CodeGen/X86/scalar_widen_div.ll
index e99ea9356a64e27ef179d4d0bdb993136f1e88b6..1671f8f891086c410b01538e1acfe4a89ca34a2c 100644 (file)
--- a/test/CodeGen/X86/scalar_widen_div.ll
+++ b/test/CodeGen/X86/scalar_widen_div.ll
@@ -1,4 +1,4 @@
-; RUN: llc < %s -march=x86-64 -mattr=+sse42 |  FileCheck %s
+; RUN: llc < %s -march=x86-64 -mattr=+sse4.2 |  FileCheck %s
 
 ; Verify when widening a divide/remainder operation, we only generate a
 ; divide/rem per element since divide/remainder can trap.
@@ -17,17 +17,17 @@ entry:
   store <2 x i32> addrspace(1)* %nsource, <2 x i32> addrspace(1)** %nsource.addr
   store <2 x i32> addrspace(1)* %dsource, <2 x i32> addrspace(1)** %dsource.addr
   store <2 x i32> addrspace(1)* %qdest, <2 x i32> addrspace(1)** %qdest.addr
-  %tmp = load <2 x i32> addrspace(1)** %qdest.addr
-  %tmp1 = load i32* %index
-  %arrayidx = getelementptr <2 x i32> addrspace(1)* %tmp, i32 %tmp1
-  %tmp2 = load <2 x i32> addrspace(1)** %nsource.addr
-  %tmp3 = load i32* %index
-  %arrayidx4 = getelementptr <2 x i32> addrspace(1)* %tmp2, i32 %tmp3
-  %tmp5 = load <2 x i32> addrspace(1)* %arrayidx4
-  %tmp6 = load <2 x i32> addrspace(1)** %dsource.addr
-  %tmp7 = load i32* %index
-  %arrayidx8 = getelementptr <2 x i32> addrspace(1)* %tmp6, i32 %tmp7
-  %tmp9 = load <2 x i32> addrspace(1)* %arrayidx8
+  %tmp = load <2 x i32> addrspace(1)*, <2 x i32> addrspace(1)** %qdest.addr
+  %tmp1 = load i32, i32* %index
+  %arrayidx = getelementptr <2 x i32>, <2 x i32> addrspace(1)* %tmp, i32 %tmp1
+  %tmp2 = load <2 x i32> addrspace(1)*, <2 x i32> addrspace(1)** %nsource.addr
+  %tmp3 = load i32, i32* %index
+  %arrayidx4 = getelementptr <2 x i32>, <2 x i32> addrspace(1)* %tmp2, i32 %tmp3
+  %tmp5 = load <2 x i32>, <2 x i32> addrspace(1)* %arrayidx4
+  %tmp6 = load <2 x i32> addrspace(1)*, <2 x i32> addrspace(1)** %dsource.addr
+  %tmp7 = load i32, i32* %index
+  %arrayidx8 = getelementptr <2 x i32>, <2 x i32> addrspace(1)* %tmp6, i32 %tmp7
+  %tmp9 = load <2 x i32>, <2 x i32> addrspace(1)* %arrayidx8
   %tmp10 = sdiv <2 x i32> %tmp5, %tmp9
   store <2 x i32> %tmp10, <2 x i32> addrspace(1)* %arrayidx
   ret void
@@ -180,10 +180,10 @@ bb.nph:
 
 for.body:
   %i.014 = phi i32 [ 0, %bb.nph ], [ %inc, %for.body ] 
-  %arrayidx11 = getelementptr <3 x i32>* %dest, i32 %i.014
-  %tmp4 = load <3 x i32>* %arrayidx11 ; <<3 x i32>> [#uses=1]
-  %arrayidx7 = getelementptr inbounds <3 x i32>* %old, i32 %i.014
-  %tmp8 = load <3 x i32>* %arrayidx7 ; <<3 x i32>> [#uses=1]
+  %arrayidx11 = getelementptr <3 x i32>, <3 x i32>* %dest, i32 %i.014
+  %tmp4 = load <3 x i32>, <3 x i32>* %arrayidx11 ; <<3 x i32>> [#uses=1]
+  %arrayidx7 = getelementptr inbounds <3 x i32>, <3 x i32>* %old, i32 %i.014
+  %tmp8 = load <3 x i32>, <3 x i32>* %arrayidx7 ; <<3 x i32>> [#uses=1]
   %div = sdiv <3 x i32> %tmp4, %tmp8
   store <3 x i32> %div, <3 x i32>* %arrayidx11
   %inc = add nsw i32 %i.014, 1