1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py 2 ; RUN: opt < %s -sroa -instcombine | \ 3 ; RUN: llc -march=x86 -mcpu=yonah | not grep sub.*esp 4 5 ; This checks that various insert/extract idiom work without going to the 6 ; stack. 7 target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32" 8 9 define void @test(<4 x float>* %F, float %f) { 10 entry: 11 %tmp = load <4 x float>, <4 x float>* %F ; <<4 x float>> [#uses=2] 12 %tmp3 = fadd <4 x float> %tmp, %tmp ; <<4 x float>> [#uses=1] 13 %tmp10 = insertelement <4 x float> %tmp3, float %f, i32 0 ; <<4 x float>> [#uses=2] 14 %tmp6 = fadd <4 x float> %tmp10, %tmp10 ; <<4 x float>> [#uses=1] 15 store <4 x float> %tmp6, <4 x float>* %F 16 ret void 17 } 18 19 define void @test2(<4 x float>* %F, float %f) { 20 entry: 21 %G = alloca <4 x float>, align 16 ; <<4 x float>*> [#uses=3] 22 %tmp = load <4 x float>, <4 x float>* %F ; <<4 x float>> [#uses=2] 23 %tmp3 = fadd <4 x float> %tmp, %tmp ; <<4 x float>> [#uses=1] 24 store <4 x float> %tmp3, <4 x float>* %G 25 %tmp.upgrd.1 = getelementptr <4 x float>, <4 x float>* %G, i32 0, i32 2 ; <float*> [#uses=1] 26 store float %f, float* %tmp.upgrd.1 27 %tmp4 = load <4 x float>, <4 x float>* %G ; <<4 x float>> [#uses=2] 28 %tmp6 = fadd <4 x float> %tmp4, %tmp4 ; <<4 x float>> [#uses=1] 29 store <4 x float> %tmp6, <4 x float>* %F 30 ret void 31 } 32 33 define void @test3(<4 x float>* %F, float* %f) { 34 entry: 35 %G = alloca <4 x float>, align 16 ; <<4 x float>*> [#uses=2] 36 %tmp = load <4 x float>, <4 x float>* %F ; <<4 x float>> [#uses=2] 37 %tmp3 = fadd <4 x float> %tmp, %tmp ; <<4 x float>> [#uses=1] 38 store <4 x float> %tmp3, <4 x float>* %G 39 %tmp.upgrd.2 = getelementptr <4 x float>, <4 x float>* %G, i32 0, i32 2 ; <float*> [#uses=1] 40 %tmp.upgrd.3 = load float, float* %tmp.upgrd.2 ; <float> [#uses=1] 41 store float %tmp.upgrd.3, float* %f 42 ret void 43 } 44 45 define void @test4(<4 x float>* %F, float* %f) { 46 entry: 47 %tmp = load <4 x float>, <4 x float>* %F ; <<4 x float>> [#uses=2] 48 %tmp5.lhs = extractelement <4 x float> %tmp, i32 0 ; <float> [#uses=1] 49 %tmp5.rhs = extractelement <4 x float> %tmp, i32 0 ; <float> [#uses=1] 50 %tmp5 = fadd float %tmp5.lhs, %tmp5.rhs ; <float> [#uses=1] 51 store float %tmp5, float* %f 52 ret void 53 } 54
