Both operands to a binary operator are not of the same type!
  %57 = and <4 x i32> %56, <4 x float> zeroinitializer ; <<4 x i32>> [#uses=1]
Instruction does not dominate all uses!
  %57 = and <4 x i32> %56, <4 x float> zeroinitializer ; <<4 x i32>> [#uses=1]
  store <4 x i32> %57, <4 x i32>* %execution_mask
Both operands to a binary operator are not of the same type!
  %74 = and <4 x i32> %73, <4 x float> zeroinitializer ; <<4 x i32>> [#uses=1]
Instruction does not dominate all uses!
  %74 = and <4 x i32> %73, <4 x float> zeroinitializer ; <<4 x i32>> [#uses=1]
  store <4 x i32> %74, <4 x i32>* %execution_mask14
Both operands to a binary operator are not of the same type!
  %91 = and <4 x i32> %90, <4 x float> zeroinitializer ; <<4 x i32>> [#uses=1]
Instruction does not dominate all uses!
  %91 = and <4 x i32> %90, <4 x float> zeroinitializer ; <<4 x i32>> [#uses=1]
  store <4 x i32> %91, <4 x i32>* %execution_mask30
Both operands to a binary operator are not of the same type!
  %108 = and <4 x i32> %107, <4 x float> zeroinitializer ; <<4 x i32>> [#uses=1]
Instruction does not dominate all uses!
  %108 = and <4 x i32> %107, <4 x float> zeroinitializer ; <<4 x i32>> [#uses=1]
  store <4 x i32> %108, <4 x i32>* %execution_mask46
Broken module found, verification continues.

define void @fs4_variant0_partial(%context* noalias %context, i32 %x, i32 %y, i32, float* noalias %a0, float* noalias %dadx, float* noalias %dady, <16 x i8>** noalias %color_ptr_ptr, i8* noalias %depth, i32 %mask_input, i32* noalias) {
entry:
  %execution_mask67 = alloca <16 x i8>            ; <<16 x i8>*> [#uses=3]
  %execution_mask46 = alloca <4 x i32>            ; <<4 x i32>*> [#uses=5]
  %color45 = alloca <4 x float>                   ; <<4 x float>*> [#uses=2]
  %color44 = alloca <4 x float>                   ; <<4 x float>*> [#uses=2]
  %color43 = alloca <4 x float>                   ; <<4 x float>*> [#uses=2]
  %color42 = alloca <4 x float>                   ; <<4 x float>*> [#uses=2]
  %execution_mask30 = alloca <4 x i32>            ; <<4 x i32>*> [#uses=5]
  %color29 = alloca <4 x float>                   ; <<4 x float>*> [#uses=2]
  %color28 = alloca <4 x float>                   ; <<4 x float>*> [#uses=2]
  %color27 = alloca <4 x float>                   ; <<4 x float>*> [#uses=2]
  %color26 = alloca <4 x float>                   ; <<4 x float>*> [#uses=2]
  %execution_mask14 = alloca <4 x i32>            ; <<4 x i32>*> [#uses=5]
  %color13 = alloca <4 x float>                   ; <<4 x float>*> [#uses=2]
  %color12 = alloca <4 x float>                   ; <<4 x float>*> [#uses=2]
  %color11 = alloca <4 x float>                   ; <<4 x float>*> [#uses=2]
  %color10 = alloca <4 x float>                   ; <<4 x float>*> [#uses=2]
  %execution_mask = alloca <4 x i32>              ; <<4 x i32>*> [#uses=5]
  %color3 = alloca <4 x float>                    ; <<4 x float>*> [#uses=2]
  %color2 = alloca <4 x float>                    ; <<4 x float>*> [#uses=2]
  %color1 = alloca <4 x float>                    ; <<4 x float>*> [#uses=2]
  %color = alloca <4 x float>                     ; <<4 x float>*> [#uses=2]
  %2 = sitofp i32 %x to float                     ; <float> [#uses=6]
  %3 = sitofp i32 %y to float                     ; <float> [#uses=6]
  %4 = getelementptr float* %a0, i32 0            ; <float*> [#uses=1]
  %pos.x.a0 = load float* %4                      ; <float> [#uses=0]
  %5 = insertelement <4 x float> undef, float %2, i32 0 ; <<4 x float>> [#uses=1]
  %6 = insertelement <4 x float> %5, float %2, i32 1 ; <<4 x float>> [#uses=1]
  %7 = insertelement <4 x float> %6, float %2, i32 2 ; <<4 x float>> [#uses=1]
  %8 = insertelement <4 x float> %7, float %2, i32 3 ; <<4 x float>> [#uses=1]
  %pos.x.a = fadd <4 x float> %8, <float 0.000000e+00, float 2.000000e+00, float 0.000000e+00, float 2.000000e+00> ; <<4 x float>> [#uses=4]
  %9 = getelementptr float* %a0, i32 1            ; <float*> [#uses=1]
  %pos.y.a0 = load float* %9                      ; <float> [#uses=0]
  %10 = insertelement <4 x float> undef, float %3, i32 0 ; <<4 x float>> [#uses=1]
  %11 = insertelement <4 x float> %10, float %3, i32 1 ; <<4 x float>> [#uses=1]
  %12 = insertelement <4 x float> %11, float %3, i32 2 ; <<4 x float>> [#uses=1]
  %13 = insertelement <4 x float> %12, float %3, i32 3 ; <<4 x float>> [#uses=1]
  %pos.y.a = fadd <4 x float> %13, <float 0.000000e+00, float 0.000000e+00, float 2.000000e+00, float 2.000000e+00> ; <<4 x float>> [#uses=4]
  %14 = getelementptr float* %dadx, i32 2         ; <float*> [#uses=1]
  %pos.z.dadx = load float* %14                   ; <float> [#uses=3]
  %15 = getelementptr float* %dady, i32 2         ; <float*> [#uses=1]
  %pos.z.dady = load float* %15                   ; <float> [#uses=3]
  %pos.z.dadxy = fadd float %pos.z.dadx, %pos.z.dady ; <float> [#uses=1]
  %16 = getelementptr float* %a0, i32 2           ; <float*> [#uses=1]
  %pos.z.a0 = load float* %16                     ; <float> [#uses=1]
  %17 = insertelement <4 x float> <float 0.000000e+00, float undef, float undef, float undef>, float %pos.z.dadx, i32 1 ; <<4 x float>> [#uses=1]
  %18 = insertelement <4 x float> %17, float %pos.z.dady, i32 2 ; <<4 x float>> [#uses=1]
  %pos.z.dadq = insertelement <4 x float> %18, float %pos.z.dadxy, i32 3 ; <<4 x float>> [#uses=6]
  %19 = fadd <4 x float> %pos.z.dadq, %pos.z.dadq ; <<4 x float>> [#uses=1]
  %20 = fmul float %2, %pos.z.dadx                ; <float> [#uses=1]
  %21 = fmul float %3, %pos.z.dady                ; <float> [#uses=1]
  %22 = fadd float %20, %21                       ; <float> [#uses=1]
  %23 = fadd float %pos.z.a0, %22                 ; <float> [#uses=4]
  %24 = insertelement <4 x float> undef, float %23, i32 0 ; <<4 x float>> [#uses=1]
  %25 = insertelement <4 x float> %24, float %23, i32 1 ; <<4 x float>> [#uses=1]
  %26 = insertelement <4 x float> %25, float %23, i32 2 ; <<4 x float>> [#uses=1]
  %27 = insertelement <4 x float> %26, float %23, i32 3 ; <<4 x float>> [#uses=1]
  %pos.z.a = fadd <4 x float> %27, %19            ; <<4 x float>> [#uses=4]
  %28 = getelementptr float* %dadx, i32 3         ; <float*> [#uses=1]
  %pos.w.dadx = load float* %28                   ; <float> [#uses=3]
  %29 = getelementptr float* %dady, i32 3         ; <float*> [#uses=1]
  %pos.w.dady = load float* %29                   ; <float> [#uses=3]
  %pos.w.dadxy = fadd float %pos.w.dadx, %pos.w.dady ; <float> [#uses=1]
  %30 = getelementptr float* %a0, i32 3           ; <float*> [#uses=1]
  %pos.w.a0 = load float* %30                     ; <float> [#uses=1]
  %31 = insertelement <4 x float> <float 0.000000e+00, float undef, float undef, float undef>, float %pos.w.dadx, i32 1 ; <<4 x float>> [#uses=1]
  %32 = insertelement <4 x float> %31, float %pos.w.dady, i32 2 ; <<4 x float>> [#uses=1]
  %pos.w.dadq = insertelement <4 x float> %32, float %pos.w.dadxy, i32 3 ; <<4 x float>> [#uses=6]
  %33 = fadd <4 x float> %pos.w.dadq, %pos.w.dadq ; <<4 x float>> [#uses=1]
  %34 = fmul float %2, %pos.w.dadx                ; <float> [#uses=1]
  %35 = fmul float %3, %pos.w.dady                ; <float> [#uses=1]
  %36 = fadd float %34, %35                       ; <float> [#uses=1]
  %37 = fadd float %pos.w.a0, %36                 ; <float> [#uses=4]
  %38 = insertelement <4 x float> undef, float %37, i32 0 ; <<4 x float>> [#uses=1]
  %39 = insertelement <4 x float> %38, float %37, i32 1 ; <<4 x float>> [#uses=1]
  %40 = insertelement <4 x float> %39, float %37, i32 2 ; <<4 x float>> [#uses=1]
  %41 = insertelement <4 x float> %40, float %37, i32 3 ; <<4 x float>> [#uses=1]
  %pos.w.a = fadd <4 x float> %41, %33            ; <<4 x float>> [#uses=4]
  %42 = getelementptr i8* %depth, i32 0           ; <i8*> [#uses=0]
  %context.stencil_ref_front_ptr = getelementptr %context* %context, i32 0, i32 2 ; <i32*> [#uses=1]
  %context.stencil_ref_front = load i32* %context.stencil_ref_front_ptr ; <i32> [#uses=0]
  %context.stencil_ref_back_ptr = getelementptr %context* %context, i32 0, i32 3 ; <i32*> [#uses=1]
  %context.stencil_ref_back = load i32* %context.stencil_ref_back_ptr ; <i32> [#uses=0]
  %context.constants_ptr = getelementptr %context* %context, i32 0, i32 0 ; <float**> [#uses=1]
  %context.constants = load float** %context.constants_ptr ; <float*> [#uses=0]
  store <4 x float> zeroinitializer, <4 x float>* %color
  store <4 x float> zeroinitializer, <4 x float>* %color1
  store <4 x float> zeroinitializer, <4 x float>* %color2
  store <4 x float> zeroinitializer, <4 x float>* %color3
  %43 = lshr i32 %mask_input, 0                   ; <i32> [#uses=4]
  %44 = insertelement <4 x i32> undef, i32 %43, i32 0 ; <<4 x i32>> [#uses=1]
  %45 = insertelement <4 x i32> %44, i32 %43, i32 1 ; <<4 x i32>> [#uses=1]
  %46 = insertelement <4 x i32> %45, i32 %43, i32 2 ; <<4 x i32>> [#uses=1]
  %47 = insertelement <4 x i32> %46, i32 %43, i32 3 ; <<4 x i32>> [#uses=1]
  %48 = and <4 x i32> %47, <i32 1, i32 2, i32 16, i32 32> ; <<4 x i32>> [#uses=1]
  %49 = icmp ne <4 x i32> %48, zeroinitializer    ; <<4 x i1>> [#uses=1]
  %50 = sext <4 x i1> %49 to <4 x i32>            ; <<4 x i32>> [#uses=1]
  store <4 x i32> zeroinitializer, <4 x i32>* %execution_mask
  store <4 x i32> %50, <4 x i32>* %execution_mask
  %51 = shufflevector <4 x float> %pos.x.a, <4 x float> undef, <4 x i32> zeroinitializer ; <<4 x float>> [#uses=1]
  %pos.x = fadd <4 x float> %51, <float 0.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00> ; <<4 x float>> [#uses=0]
  %52 = shufflevector <4 x float> %pos.y.a, <4 x float> undef, <4 x i32> zeroinitializer ; <<4 x float>> [#uses=1]
  %pos.y = fadd <4 x float> %52, <float 0.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00> ; <<4 x float>> [#uses=0]
  %53 = shufflevector <4 x float> %pos.z.a, <4 x float> undef, <4 x i32> zeroinitializer ; <<4 x float>> [#uses=1]
  %54 = fadd <4 x float> %53, %pos.z.dadq         ; <<4 x float>> [#uses=1]
  %pos.z = call <4 x float> @llvm.x86.sse.min.ps(<4 x float> %54, <4 x float> <float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00>) ; <<4 x float>> [#uses=0]
  %55 = shufflevector <4 x float> %pos.w.a, <4 x float> undef, <4 x i32> zeroinitializer ; <<4 x float>> [#uses=1]
  %pos.w = fadd <4 x float> %55, %pos.w.dadq      ; <<4 x float>> [#uses=0]
  %56 = load <4 x i32>* %execution_mask           ; <<4 x i32>> [#uses=1]
  %57 = and <4 x i32> %56, <4 x float> zeroinitializer ; <<4 x i32>> [#uses=1]
  store <4 x i32> %57, <4 x i32>* %execution_mask
  br label %skip

skip:                                             ; preds = %entry
  %58 = load <4 x i32>* %execution_mask           ; <<4 x i32>> [#uses=1]
  %59 = getelementptr i8* %depth, i32 4           ; <i8*> [#uses=0]
  %context.stencil_ref_front_ptr4 = getelementptr %context* %context, i32 0, i32 2 ; <i32*> [#uses=1]
  %context.stencil_ref_front5 = load i32* %context.stencil_ref_front_ptr4 ; <i32> [#uses=0]
  %context.stencil_ref_back_ptr6 = getelementptr %context* %context, i32 0, i32 3 ; <i32*> [#uses=1]
  %context.stencil_ref_back7 = load i32* %context.stencil_ref_back_ptr6 ; <i32> [#uses=0]
  %context.constants_ptr8 = getelementptr %context* %context, i32 0, i32 0 ; <float**> [#uses=1]
  %context.constants9 = load float** %context.constants_ptr8 ; <float*> [#uses=0]
  store <4 x float> zeroinitializer, <4 x float>* %color10
  store <4 x float> zeroinitializer, <4 x float>* %color11
  store <4 x float> zeroinitializer, <4 x float>* %color12
  store <4 x float> zeroinitializer, <4 x float>* %color13
  %60 = lshr i32 %mask_input, 2                   ; <i32> [#uses=4]
  %61 = insertelement <4 x i32> undef, i32 %60, i32 0 ; <<4 x i32>> [#uses=1]
  %62 = insertelement <4 x i32> %61, i32 %60, i32 1 ; <<4 x i32>> [#uses=1]
  %63 = insertelement <4 x i32> %62, i32 %60, i32 2 ; <<4 x i32>> [#uses=1]
  %64 = insertelement <4 x i32> %63, i32 %60, i32 3 ; <<4 x i32>> [#uses=1]
  %65 = and <4 x i32> %64, <i32 1, i32 2, i32 16, i32 32> ; <<4 x i32>> [#uses=1]
  %66 = icmp ne <4 x i32> %65, zeroinitializer    ; <<4 x i1>> [#uses=1]
  %67 = sext <4 x i1> %66 to <4 x i32>            ; <<4 x i32>> [#uses=1]
  store <4 x i32> zeroinitializer, <4 x i32>* %execution_mask14
  store <4 x i32> %67, <4 x i32>* %execution_mask14
  %68 = shufflevector <4 x float> %pos.x.a, <4 x float> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1> ; <<4 x float>> [#uses=1]
  %pos.x16 = fadd <4 x float> %68, <float 0.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00> ; <<4 x float>> [#uses=0]
  %69 = shufflevector <4 x float> %pos.y.a, <4 x float> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1> ; <<4 x float>> [#uses=1]
  %pos.y17 = fadd <4 x float> %69, <float 0.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00> ; <<4 x float>> [#uses=0]
  %70 = shufflevector <4 x float> %pos.z.a, <4 x float> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1> ; <<4 x float>> [#uses=1]
  %71 = fadd <4 x float> %70, %pos.z.dadq         ; <<4 x float>> [#uses=1]
  %pos.z18 = call <4 x float> @llvm.x86.sse.min.ps(<4 x float> %71, <4 x float> <float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00>) ; <<4 x float>> [#uses=0]
  %72 = shufflevector <4 x float> %pos.w.a, <4 x float> undef, <4 x i32> <i32 1, i32 1, i32 1, i32 1> ; <<4 x float>> [#uses=1]
  %pos.w19 = fadd <4 x float> %72, %pos.w.dadq    ; <<4 x float>> [#uses=0]
  %73 = load <4 x i32>* %execution_mask14         ; <<4 x i32>> [#uses=1]
  %74 = and <4 x i32> %73, <4 x float> zeroinitializer ; <<4 x i32>> [#uses=1]
  store <4 x i32> %74, <4 x i32>* %execution_mask14
  br label %skip15

skip15:                                           ; preds = %skip
  %75 = load <4 x i32>* %execution_mask14         ; <<4 x i32>> [#uses=1]
  %76 = getelementptr i8* %depth, i32 8           ; <i8*> [#uses=0]
  %context.stencil_ref_front_ptr20 = getelementptr %context* %context, i32 0, i32 2 ; <i32*> [#uses=1]
  %context.stencil_ref_front21 = load i32* %context.stencil_ref_front_ptr20 ; <i32> [#uses=0]
  %context.stencil_ref_back_ptr22 = getelementptr %context* %context, i32 0, i32 3 ; <i32*> [#uses=1]
  %context.stencil_ref_back23 = load i32* %context.stencil_ref_back_ptr22 ; <i32> [#uses=0]
  %context.constants_ptr24 = getelementptr %context* %context, i32 0, i32 0 ; <float**> [#uses=1]
  %context.constants25 = load float** %context.constants_ptr24 ; <float*> [#uses=0]
  store <4 x float> zeroinitializer, <4 x float>* %color26
  store <4 x float> zeroinitializer, <4 x float>* %color27
  store <4 x float> zeroinitializer, <4 x float>* %color28
  store <4 x float> zeroinitializer, <4 x float>* %color29
  %77 = lshr i32 %mask_input, 8                   ; <i32> [#uses=4]
  %78 = insertelement <4 x i32> undef, i32 %77, i32 0 ; <<4 x i32>> [#uses=1]
  %79 = insertelement <4 x i32> %78, i32 %77, i32 1 ; <<4 x i32>> [#uses=1]
  %80 = insertelement <4 x i32> %79, i32 %77, i32 2 ; <<4 x i32>> [#uses=1]
  %81 = insertelement <4 x i32> %80, i32 %77, i32 3 ; <<4 x i32>> [#uses=1]
  %82 = and <4 x i32> %81, <i32 1, i32 2, i32 16, i32 32> ; <<4 x i32>> [#uses=1]
  %83 = icmp ne <4 x i32> %82, zeroinitializer    ; <<4 x i1>> [#uses=1]
  %84 = sext <4 x i1> %83 to <4 x i32>            ; <<4 x i32>> [#uses=1]
  store <4 x i32> zeroinitializer, <4 x i32>* %execution_mask30
  store <4 x i32> %84, <4 x i32>* %execution_mask30
  %85 = shufflevector <4 x float> %pos.x.a, <4 x float> undef, <4 x i32> <i32 2, i32 2, i32 2, i32 2> ; <<4 x float>> [#uses=1]
  %pos.x32 = fadd <4 x float> %85, <float 0.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00> ; <<4 x float>> [#uses=0]
  %86 = shufflevector <4 x float> %pos.y.a, <4 x float> undef, <4 x i32> <i32 2, i32 2, i32 2, i32 2> ; <<4 x float>> [#uses=1]
  %pos.y33 = fadd <4 x float> %86, <float 0.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00> ; <<4 x float>> [#uses=0]
  %87 = shufflevector <4 x float> %pos.z.a, <4 x float> undef, <4 x i32> <i32 2, i32 2, i32 2, i32 2> ; <<4 x float>> [#uses=1]
  %88 = fadd <4 x float> %87, %pos.z.dadq         ; <<4 x float>> [#uses=1]
  %pos.z34 = call <4 x float> @llvm.x86.sse.min.ps(<4 x float> %88, <4 x float> <float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00>) ; <<4 x float>> [#uses=0]
  %89 = shufflevector <4 x float> %pos.w.a, <4 x float> undef, <4 x i32> <i32 2, i32 2, i32 2, i32 2> ; <<4 x float>> [#uses=1]
  %pos.w35 = fadd <4 x float> %89, %pos.w.dadq    ; <<4 x float>> [#uses=0]
  %90 = load <4 x i32>* %execution_mask30         ; <<4 x i32>> [#uses=1]
  %91 = and <4 x i32> %90, <4 x float> zeroinitializer ; <<4 x i32>> [#uses=1]
  store <4 x i32> %91, <4 x i32>* %execution_mask30
  br label %skip31

skip31:                                           ; preds = %skip15
  %92 = load <4 x i32>* %execution_mask30         ; <<4 x i32>> [#uses=1]
  %93 = getelementptr i8* %depth, i32 12          ; <i8*> [#uses=0]
  %context.stencil_ref_front_ptr36 = getelementptr %context* %context, i32 0, i32 2 ; <i32*> [#uses=1]
  %context.stencil_ref_front37 = load i32* %context.stencil_ref_front_ptr36 ; <i32> [#uses=0]
  %context.stencil_ref_back_ptr38 = getelementptr %context* %context, i32 0, i32 3 ; <i32*> [#uses=1]
  %context.stencil_ref_back39 = load i32* %context.stencil_ref_back_ptr38 ; <i32> [#uses=0]
  %context.constants_ptr40 = getelementptr %context* %context, i32 0, i32 0 ; <float**> [#uses=1]
  %context.constants41 = load float** %context.constants_ptr40 ; <float*> [#uses=0]
  store <4 x float> zeroinitializer, <4 x float>* %color42
  store <4 x float> zeroinitializer, <4 x float>* %color43
  store <4 x float> zeroinitializer, <4 x float>* %color44
  store <4 x float> zeroinitializer, <4 x float>* %color45
  %94 = lshr i32 %mask_input, 10                  ; <i32> [#uses=4]
  %95 = insertelement <4 x i32> undef, i32 %94, i32 0 ; <<4 x i32>> [#uses=1]
  %96 = insertelement <4 x i32> %95, i32 %94, i32 1 ; <<4 x i32>> [#uses=1]
  %97 = insertelement <4 x i32> %96, i32 %94, i32 2 ; <<4 x i32>> [#uses=1]
  %98 = insertelement <4 x i32> %97, i32 %94, i32 3 ; <<4 x i32>> [#uses=1]
  %99 = and <4 x i32> %98, <i32 1, i32 2, i32 16, i32 32> ; <<4 x i32>> [#uses=1]
  %100 = icmp ne <4 x i32> %99, zeroinitializer   ; <<4 x i1>> [#uses=1]
  %101 = sext <4 x i1> %100 to <4 x i32>          ; <<4 x i32>> [#uses=1]
  store <4 x i32> zeroinitializer, <4 x i32>* %execution_mask46
  store <4 x i32> %101, <4 x i32>* %execution_mask46
  %102 = shufflevector <4 x float> %pos.x.a, <4 x float> undef, <4 x i32> <i32 3, i32 3, i32 3, i32 3> ; <<4 x float>> [#uses=1]
  %pos.x48 = fadd <4 x float> %102, <float 0.000000e+00, float 1.000000e+00, float 0.000000e+00, float 1.000000e+00> ; <<4 x float>> [#uses=0]
  %103 = shufflevector <4 x float> %pos.y.a, <4 x float> undef, <4 x i32> <i32 3, i32 3, i32 3, i32 3> ; <<4 x float>> [#uses=1]
  %pos.y49 = fadd <4 x float> %103, <float 0.000000e+00, float 0.000000e+00, float 1.000000e+00, float 1.000000e+00> ; <<4 x float>> [#uses=0]
  %104 = shufflevector <4 x float> %pos.z.a, <4 x float> undef, <4 x i32> <i32 3, i32 3, i32 3, i32 3> ; <<4 x float>> [#uses=1]
  %105 = fadd <4 x float> %104, %pos.z.dadq       ; <<4 x float>> [#uses=1]
  %pos.z50 = call <4 x float> @llvm.x86.sse.min.ps(<4 x float> %105, <4 x float> <float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00>) ; <<4 x float>> [#uses=0]
  %106 = shufflevector <4 x float> %pos.w.a, <4 x float> undef, <4 x i32> <i32 3, i32 3, i32 3, i32 3> ; <<4 x float>> [#uses=1]
  %pos.w51 = fadd <4 x float> %106, %pos.w.dadq   ; <<4 x float>> [#uses=0]
  %107 = load <4 x i32>* %execution_mask46        ; <<4 x i32>> [#uses=1]
  %108 = and <4 x i32> %107, <4 x float> zeroinitializer ; <<4 x i32>> [#uses=1]
  store <4 x i32> %108, <4 x i32>* %execution_mask46
  br label %skip47

skip47:                                           ; preds = %skip31
  %109 = load <4 x i32>* %execution_mask46        ; <<4 x i32>> [#uses=1]
  %fs_color_vals = load <4 x float>* %color       ; <<4 x float>> [#uses=1]
  %fs_color_vals52 = load <4 x float>* %color10   ; <<4 x float>> [#uses=1]
  %fs_color_vals53 = load <4 x float>* %color26   ; <<4 x float>> [#uses=1]
  %fs_color_vals54 = load <4 x float>* %color42   ; <<4 x float>> [#uses=1]
  %110 = fmul <4 x float> %fs_color_vals, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %111 = fmul <4 x float> %fs_color_vals52, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %112 = fmul <4 x float> %fs_color_vals53, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %113 = fmul <4 x float> %fs_color_vals54, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %114 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %110) ; <<4 x i32>> [#uses=1]
  %115 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %111) ; <<4 x i32>> [#uses=1]
  %116 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %112) ; <<4 x i32>> [#uses=1]
  %117 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %113) ; <<4 x i32>> [#uses=1]
  %118 = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> %114, <4 x i32> %115) ; <<8 x i16>> [#uses=1]
  %119 = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> %116, <4 x i32> %117) ; <<8 x i16>> [#uses=1]
  %res.r = call <16 x i8> @llvm.x86.sse2.packuswb.128(<8 x i16> %118, <8 x i16> %119) ; <<16 x i8>> [#uses=1]
  %fs_color_vals55 = load <4 x float>* %color1    ; <<4 x float>> [#uses=1]
  %fs_color_vals56 = load <4 x float>* %color11   ; <<4 x float>> [#uses=1]
  %fs_color_vals57 = load <4 x float>* %color27   ; <<4 x float>> [#uses=1]
  %fs_color_vals58 = load <4 x float>* %color43   ; <<4 x float>> [#uses=1]
  %120 = fmul <4 x float> %fs_color_vals55, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %121 = fmul <4 x float> %fs_color_vals56, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %122 = fmul <4 x float> %fs_color_vals57, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %123 = fmul <4 x float> %fs_color_vals58, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %124 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %120) ; <<4 x i32>> [#uses=1]
  %125 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %121) ; <<4 x i32>> [#uses=1]
  %126 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %122) ; <<4 x i32>> [#uses=1]
  %127 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %123) ; <<4 x i32>> [#uses=1]
  %128 = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> %124, <4 x i32> %125) ; <<8 x i16>> [#uses=1]
  %129 = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> %126, <4 x i32> %127) ; <<8 x i16>> [#uses=1]
  %res.g = call <16 x i8> @llvm.x86.sse2.packuswb.128(<8 x i16> %128, <8 x i16> %129) ; <<16 x i8>> [#uses=1]
  %fs_color_vals59 = load <4 x float>* %color2    ; <<4 x float>> [#uses=1]
  %fs_color_vals60 = load <4 x float>* %color12   ; <<4 x float>> [#uses=1]
  %fs_color_vals61 = load <4 x float>* %color28   ; <<4 x float>> [#uses=1]
  %fs_color_vals62 = load <4 x float>* %color44   ; <<4 x float>> [#uses=1]
  %130 = fmul <4 x float> %fs_color_vals59, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %131 = fmul <4 x float> %fs_color_vals60, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %132 = fmul <4 x float> %fs_color_vals61, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %133 = fmul <4 x float> %fs_color_vals62, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %134 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %130) ; <<4 x i32>> [#uses=1]
  %135 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %131) ; <<4 x i32>> [#uses=1]
  %136 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %132) ; <<4 x i32>> [#uses=1]
  %137 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %133) ; <<4 x i32>> [#uses=1]
  %138 = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> %134, <4 x i32> %135) ; <<8 x i16>> [#uses=1]
  %139 = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> %136, <4 x i32> %137) ; <<8 x i16>> [#uses=1]
  %res.b = call <16 x i8> @llvm.x86.sse2.packuswb.128(<8 x i16> %138, <8 x i16> %139) ; <<16 x i8>> [#uses=1]
  %fs_color_vals63 = load <4 x float>* %color3    ; <<4 x float>> [#uses=1]
  %fs_color_vals64 = load <4 x float>* %color13   ; <<4 x float>> [#uses=1]
  %fs_color_vals65 = load <4 x float>* %color29   ; <<4 x float>> [#uses=1]
  %fs_color_vals66 = load <4 x float>* %color45   ; <<4 x float>> [#uses=1]
  %140 = fmul <4 x float> %fs_color_vals63, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %141 = fmul <4 x float> %fs_color_vals64, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %142 = fmul <4 x float> %fs_color_vals65, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %143 = fmul <4 x float> %fs_color_vals66, <float 2.550000e+02, float 2.550000e+02, float 2.550000e+02, float 2.550000e+02> ; <<4 x float>> [#uses=1]
  %144 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %140) ; <<4 x i32>> [#uses=1]
  %145 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %141) ; <<4 x i32>> [#uses=1]
  %146 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %142) ; <<4 x i32>> [#uses=1]
  %147 = call <4 x i32> @llvm.x86.sse2.cvtps2dq(<4 x float> %143) ; <<4 x i32>> [#uses=1]
  %148 = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> %144, <4 x i32> %145) ; <<8 x i16>> [#uses=1]
  %149 = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> %146, <4 x i32> %147) ; <<8 x i16>> [#uses=1]
  %res.a = call <16 x i8> @llvm.x86.sse2.packuswb.128(<8 x i16> %148, <8 x i16> %149) ; <<16 x i8>> [#uses=1]
  %150 = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> %58, <4 x i32> %75) ; <<8 x i16>> [#uses=1]
  %151 = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> %92, <4 x i32> %109) ; <<8 x i16>> [#uses=1]
  %152 = call <16 x i8> @llvm.x86.sse2.packsswb.128(<8 x i16> %150, <8 x i16> %151) ; <<16 x i8>> [#uses=9]
  %153 = getelementptr <16 x i8>** %color_ptr_ptr, i32 0 ; <<16 x i8>**> [#uses=1]
  %color_ptr0 = load <16 x i8>** %153             ; <<16 x i8>*> [#uses=8]
  store <16 x i8> zeroinitializer, <16 x i8>* %execution_mask67
  store <16 x i8> %152, <16 x i8>* %execution_mask67
  %context.blend_color_ptr = getelementptr %context* %context, i32 0, i32 4 ; <i8**> [#uses=1]
  %context.blend_color = load i8** %context.blend_color_ptr ; <i8*> [#uses=1]
  %154 = bitcast i8* %context.blend_color to <16 x i8>* ; <<16 x i8>*> [#uses=4]
  %155 = getelementptr <16 x i8>* %154, i32 0     ; <<16 x i8>*> [#uses=1]
  %con.r = load <16 x i8>* %155                   ; <<16 x i8>> [#uses=0]
  %156 = getelementptr <16 x i8>* %color_ptr0, i32 0 ; <<16 x i8>*> [#uses=1]
  %dst.r = load <16 x i8>* %156                   ; <<16 x i8>> [#uses=1]
  %157 = getelementptr <16 x i8>* %154, i32 1     ; <<16 x i8>*> [#uses=1]
  %con.g = load <16 x i8>* %157                   ; <<16 x i8>> [#uses=0]
  %158 = getelementptr <16 x i8>* %color_ptr0, i32 1 ; <<16 x i8>*> [#uses=1]
  %dst.g = load <16 x i8>* %158                   ; <<16 x i8>> [#uses=1]
  %159 = getelementptr <16 x i8>* %154, i32 2     ; <<16 x i8>*> [#uses=1]
  %con.b = load <16 x i8>* %159                   ; <<16 x i8>> [#uses=0]
  %160 = getelementptr <16 x i8>* %color_ptr0, i32 2 ; <<16 x i8>*> [#uses=1]
  %dst.b = load <16 x i8>* %160                   ; <<16 x i8>> [#uses=1]
  %161 = getelementptr <16 x i8>* %154, i32 3     ; <<16 x i8>*> [#uses=1]
  %con.a = load <16 x i8>* %161                   ; <<16 x i8>> [#uses=0]
  %162 = getelementptr <16 x i8>* %color_ptr0, i32 3 ; <<16 x i8>*> [#uses=1]
  %dst.a = load <16 x i8>* %162                   ; <<16 x i8>> [#uses=1]
  %163 = and <16 x i8> %res.r, %152               ; <<16 x i8>> [#uses=1]
  %164 = xor <16 x i8> %152, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1> ; <<16 x i8>> [#uses=1]
  %165 = and <16 x i8> %dst.r, %164               ; <<16 x i8>> [#uses=1]
  %166 = or <16 x i8> %163, %165                  ; <<16 x i8>> [#uses=1]
  %167 = getelementptr <16 x i8>* %color_ptr0, i32 0 ; <<16 x i8>*> [#uses=1]
  store <16 x i8> %166, <16 x i8>* %167
  %168 = and <16 x i8> %res.g, %152               ; <<16 x i8>> [#uses=1]
  %169 = xor <16 x i8> %152, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1> ; <<16 x i8>> [#uses=1]
  %170 = and <16 x i8> %dst.g, %169               ; <<16 x i8>> [#uses=1]
  %171 = or <16 x i8> %168, %170                  ; <<16 x i8>> [#uses=1]
  %172 = getelementptr <16 x i8>* %color_ptr0, i32 1 ; <<16 x i8>*> [#uses=1]
  store <16 x i8> %171, <16 x i8>* %172
  %173 = and <16 x i8> %res.b, %152               ; <<16 x i8>> [#uses=1]
  %174 = xor <16 x i8> %152, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1> ; <<16 x i8>> [#uses=1]
  %175 = and <16 x i8> %dst.b, %174               ; <<16 x i8>> [#uses=1]
  %176 = or <16 x i8> %173, %175                  ; <<16 x i8>> [#uses=1]
  %177 = getelementptr <16 x i8>* %color_ptr0, i32 2 ; <<16 x i8>*> [#uses=1]
  store <16 x i8> %176, <16 x i8>* %177
  %178 = and <16 x i8> %res.a, %152               ; <<16 x i8>> [#uses=1]
  %179 = xor <16 x i8> %152, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1> ; <<16 x i8>> [#uses=1]
  %180 = and <16 x i8> %dst.a, %179               ; <<16 x i8>> [#uses=1]
  %181 = or <16 x i8> %178, %180                  ; <<16 x i8>> [#uses=1]
  %182 = getelementptr <16 x i8>* %color_ptr0, i32 3 ; <<16 x i8>*> [#uses=1]
  store <16 x i8> %181, <16 x i8>* %182
  br label %skip68

skip68:                                           ; preds = %skip47
  %183 = load <16 x i8>* %execution_mask67        ; <<16 x i8>> [#uses=0]
  call void @llvm.x86.mmx.emms()
  ret void
}