new testcase, should be able to eliminate the alloca and memcpy

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@36428 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/test/Transforms/ScalarRepl/memcpy-from-global.ll b/test/Transforms/ScalarRepl/memcpy-from-global.ll
new file mode 100644 (file)
index 0000000..ee77e1f
--- /dev/null
+++ b/test/Transforms/ScalarRepl/memcpy-from-global.ll
@@ -0,0 +1,33 @@
+; RUN: llvm-as < %s | opt -scalarrepl | llvm-dis | not grep {call.*memcpy}
+@C.0.1248 = internal constant [128 x float] [ float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float -1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float -1.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float -1.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00 ], align 32               ; <[128 x float]*> [#uses=1]
+
+define float @grad4(i32 %hash, float %x, float %y, float %z, float %w) {
+entry:
+       %lookupTable = alloca [128 x float], align 16           ; <[128 x float]*> [#uses=5]
+       %lookupTable1 = bitcast [128 x float]* %lookupTable to i8*              ; <i8*> [#uses=1]
+       call void @llvm.memcpy.i32( i8* %lookupTable1, i8* bitcast ([128 x float]* @C.0.1248 to i8*), i32 512, i32 16 )
+       %tmp3 = shl i32 %hash, 2                ; <i32> [#uses=1]
+       %tmp5 = and i32 %tmp3, 124              ; <i32> [#uses=4]
+       %tmp753 = getelementptr [128 x float]* %lookupTable, i32 0, i32 %tmp5           ; <float*> [#uses=1]
+       %tmp9 = load float* %tmp753             ; <float> [#uses=1]
+       %tmp11 = mul float %tmp9, %x            ; <float> [#uses=1]
+       %tmp13 = add float %tmp11, 0.000000e+00         ; <float> [#uses=1]
+       %tmp17.sum52 = or i32 %tmp5, 1          ; <i32> [#uses=1]
+       %tmp1851 = getelementptr [128 x float]* %lookupTable, i32 0, i32 %tmp17.sum52           ; <float*> [#uses=1]
+       %tmp19 = load float* %tmp1851           ; <float> [#uses=1]
+       %tmp21 = mul float %tmp19, %y           ; <float> [#uses=1]
+       %tmp23 = add float %tmp21, %tmp13               ; <float> [#uses=1]
+       %tmp27.sum50 = or i32 %tmp5, 2          ; <i32> [#uses=1]
+       %tmp2849 = getelementptr [128 x float]* %lookupTable, i32 0, i32 %tmp27.sum50           ; <float*> [#uses=1]
+       %tmp29 = load float* %tmp2849           ; <float> [#uses=1]
+       %tmp31 = mul float %tmp29, %z           ; <float> [#uses=1]
+       %tmp33 = add float %tmp31, %tmp23               ; <float> [#uses=1]
+       %tmp37.sum48 = or i32 %tmp5, 3          ; <i32> [#uses=1]
+       %tmp3847 = getelementptr [128 x float]* %lookupTable, i32 0, i32 %tmp37.sum48           ; <float*> [#uses=1]
+       %tmp39 = load float* %tmp3847           ; <float> [#uses=1]
+       %tmp41 = mul float %tmp39, %w           ; <float> [#uses=1]
+       %tmp43 = add float %tmp41, %tmp33               ; <float> [#uses=1]
+       ret float %tmp43
+}
+
+declare void @llvm.memcpy.i32(i8*, i8*, i32, i32)