From 407ed5a63de8088afbee6c733f48bf817767fb44 Mon Sep 17 00:00:00 2001
From: Christoffer Lerno <christoffer@aegik.com>
Date: Sat, 8 Oct 2022 14:30:17 +0200
Subject: [PATCH] Fix vector init and float promotion. Fix of 1.0f.

---
 src/compiler/float.c                          |   2 +-
 src/compiler/llvm_codegen_expr.c              |  29 +-
 src/version.h                                 |   2 +-
 test/test_suite/constants/float_type.c3t      |  11 +
 .../vector/vector_init_regression.c3t         | 428 ++++++++++++++++++
 test/test_suite2/constants/float_type.c3t     |  11 +
 .../vector/vector_init_regression.c3t         | 428 ++++++++++++++++++
 7 files changed, 908 insertions(+), 3 deletions(-)
 create mode 100644 test/test_suite/constants/float_type.c3t
 create mode 100644 test/test_suite/vector/vector_init_regression.c3t
 create mode 100644 test/test_suite2/constants/float_type.c3t
 create mode 100644 test/test_suite2/vector/vector_init_regression.c3t

diff --git a/src/compiler/float.c b/src/compiler/float.c
index 059d46f43..e6fbbeefc 100644
--- a/src/compiler/float.c
+++ b/src/compiler/float.c
@@ -150,6 +150,7 @@ Float float_from_string(const char *string, char **error)
 		}
 		switch (i)
 		{
+			case 0:
 			case 32:
 				kind = TYPE_F32;
 				break;
@@ -157,7 +158,6 @@ Float float_from_string(const char *string, char **error)
 				kind = TYPE_F16;
 				break;
 			case 64:
-			case 0:
 				kind = TYPE_F64;
 				break;
 			case 128:
diff --git a/src/compiler/llvm_codegen_expr.c b/src/compiler/llvm_codegen_expr.c
index 459e5f99d..72e431a73 100644
--- a/src/compiler/llvm_codegen_expr.c
+++ b/src/compiler/llvm_codegen_expr.c
@@ -1533,6 +1533,19 @@ static inline void llvm_emit_initialize_reference_const(GenContext *c, BEValue *
 	llvm_emit_inititialize_reference_const(c, ref, initializer);
 
 }
+static inline void llvm_emit_initialize_reference_vector(GenContext *c, BEValue *ref, Type *real_type, Expr **elements)
+{
+	llvm_value_addr(c, ref);
+	LLVMTypeRef llvm_type = llvm_get_type(c, real_type);
+	LLVMValueRef vector_val = LLVMGetUndef(llvm_type);
+	BEValue element_val;
+	FOREACH_BEGIN_IDX(i, Expr *element, elements)
+		llvm_emit_expr(c, &element_val, element);
+		llvm_value_rvalue(c, &element_val);
+		vector_val = LLVMBuildInsertElement(c->builder, vector_val, element_val.value, llvm_const_int(c, type_usize, i), "");
+	FOREACH_END();
+	llvm_store_raw(c, ref, vector_val);
+}
 
 static inline void llvm_emit_initialize_reference_list(GenContext *c, BEValue *ref, Expr *expr)
 {
@@ -1540,10 +1553,17 @@ static inline void llvm_emit_initialize_reference_list(GenContext *c, BEValue *r
 	Type *real_type = type_lowering(ref->type);
 	Expr **elements = expr->initializer_list;
 
+	if (real_type->type_kind == TYPE_VECTOR)
+	{
+		llvm_emit_initialize_reference_vector(c, ref, real_type, elements);
+		return;
+	}
+
 	// Make sure we have an address.
 	llvm_value_addr(c, ref);
 	LLVMValueRef value = ref->value;
 
+
 	// If this is a union, we assume it's initializing the first element.
 	if (real_type->type_kind == TYPE_UNION)
 	{
@@ -1557,6 +1577,8 @@ static inline void llvm_emit_initialize_reference_list(GenContext *c, BEValue *r
 	LLVMTypeRef llvm_type = llvm_get_type(c, real_type);
 	bool is_struct = type_is_union_or_strukt(real_type);
 	bool is_array = real_type->type_kind == TYPE_ARRAY;
+
+	bool is_vector = real_type->type_kind == TYPE_VECTOR;
 	// Now walk through the elements.
 	VECEACH(elements, i)
 	{
@@ -1865,7 +1887,12 @@ static inline void llvm_emit_const_initialize_reference(GenContext *c, BEValue *
 {
 	assert(expr->expr_kind == EXPR_CONST && expr->const_expr.const_kind == CONST_INITIALIZER);
 	ConstInitializer *initializer = expr->const_expr.initializer;
-	assert(!type_flat_is_vector(initializer->type) && "Vectors should be handled elsewhere.");
+	if (initializer->type->type_kind == TYPE_VECTOR)
+	{
+		LLVMValueRef val = llvm_emit_const_initializer(c, initializer);
+		llvm_store_raw(c, ref, val);
+		return;
+	}
 	if (initializer->type->type_kind == TYPE_BITSTRUCT)
 	{
 		llvm_emit_const_initialize_bitstruct_ref(c, ref, initializer);
diff --git a/src/version.h b/src/version.h
index 8fef5002d..212a4e1e7 100644
--- a/src/version.h
+++ b/src/version.h
@@ -1 +1 @@
-#define COMPILER_VERSION "0.3.75"
\ No newline at end of file
+#define COMPILER_VERSION "0.3.76"
\ No newline at end of file
diff --git a/test/test_suite/constants/float_type.c3t b/test/test_suite/constants/float_type.c3t
new file mode 100644
index 000000000..5ee96b84e
--- /dev/null
+++ b/test/test_suite/constants/float_type.c3t
@@ -0,0 +1,11 @@
+// #target: macos-x64
+module test;
+uint f0 = $typeof(1.0f).sizeof;
+uint f32 = $typeof(1.0f32).sizeof;
+uint f64 = $typeof(1.0f64).sizeof;
+
+/* #expect: test.ll
+
+@test_f0 = local_unnamed_addr global i32 4, align 4
+@test_f32 = local_unnamed_addr global i32 4, align 4
+@test_f64 = local_unnamed_addr global i32 8, align 4
\ No newline at end of file
diff --git a/test/test_suite/vector/vector_init_regression.c3t b/test/test_suite/vector/vector_init_regression.c3t
new file mode 100644
index 000000000..2ce7a8a5a
--- /dev/null
+++ b/test/test_suite/vector/vector_init_regression.c3t
@@ -0,0 +1,428 @@
+// #target: macos-x64
+module test;
+extern fn int printf(char* format, ...);
+
+fn void main() {
+    float radians = 3.1415 / 4;
+    float[<3>] axis = {0.0, 0.0, 1.0};
+
+    float cosr = (float) $$cos(radians);
+    float sinr = (float) $$sin(radians);
+    float x = axis[0];
+    float y = axis[1];
+    float z = axis[2];
+
+    float[<4>][4] a = {};
+    a[0] = {
+        cosr + (x * x) * (float) (1.0 - cosr),
+        (x * y) * (float) (1.0 - cosr) - (z * sinr),
+        (x * z) * (float) (1.0 - cosr) + (y * sinr),
+        0.0
+    };
+
+    a[1] = {
+        (y * x) * (float) (1.0 - cosr) + (z * sinr),
+        cosr + (y * y) * (float) (1.0 - cosr),
+        (y * z) * (float) (1.0 - cosr) - (x * sinr),
+        0.0
+    };
+
+    a[2] = {
+        (z * x) * (float) (1.0 - cosr) - (y * sinr),
+        (z * y) * (float) (1.0 - cosr) + (x * sinr),
+        cosr + (z * z) * (float) (1.0 - cosr),
+        0.0
+    };
+
+    a[3] = {
+        0.0,
+        0.0,
+        0.0,
+        1.0
+    };
+
+    float[<4>][4] b = {
+        {
+            cosr + (x * x) * (float) (1.0 - cosr),
+            (x * y) * (float) (1.0 - cosr) - (z * sinr),
+            (x * z) * (float) (1.0 - cosr) + (y * sinr),
+            0.0
+        },
+        {
+            (y * x) * (float) (1.0 - cosr) + (z * sinr),
+            cosr + (y * y) * (float) (1.0 - cosr),
+            (y * z) * (float) (1.0 - cosr) - (x * sinr),
+            0.0
+        },
+        {
+            (z * x) * (float) (1.0 - cosr) - (y * sinr),
+            (z * y) * (float) (1.0 - cosr) + (x * sinr),
+            cosr + (z * z) * (float) (1.0 - cosr),
+            0.0
+        },
+        {
+            0.0,
+            0.0,
+            0.0,
+            1.0
+        }
+    };
+
+    foreach(v : a) {
+        printf("A: %f %f %f %f\n", v[0], v[1], v[2], v[3]);
+    }
+
+    printf("\n");
+
+    foreach(v : b) {
+        printf("B: %f %f %f %f\n", v[0], v[1], v[2], v[3]);
+    }
+}
+
+/* #expect: test.ll
+
+define void @test_main() #0 {
+entry:
+  %radians = alloca float, align 4
+  %axis = alloca <3 x float>, align 16
+  %cosr = alloca float, align 4
+  %sinr = alloca float, align 4
+  %x = alloca float, align 4
+  %y = alloca float, align 4
+  %z = alloca float, align 4
+  %a = alloca [4 x <4 x float>], align 16
+  %b = alloca [4 x <4 x float>], align 16
+  %.anon = alloca i64, align 8
+  %v = alloca <4 x float>, align 16
+  %.anon120 = alloca i64, align 8
+  %v124 = alloca <4 x float>, align 16
+  store float 0x3FE921CAC0000000, float* %radians, align 4
+  store <3 x float> <float 0.000000e+00, float 0.000000e+00, float 1.000000e+00>, <3 x float>* %axis, align 16
+  %0 = load float, float* %radians, align 4
+  %1 = call float @llvm.cos.f32(float %0)
+  store float %1, float* %cosr, align 4
+  %2 = load float, float* %radians, align 4
+  %3 = call float @llvm.sin.f32(float %2)
+  store float %3, float* %sinr, align 4
+  %4 = load <3 x float>, <3 x float>* %axis, align 16
+  %5 = extractelement <3 x float> %4, i64 0
+  store float %5, float* %x, align 4
+  %6 = load <3 x float>, <3 x float>* %axis, align 16
+  %7 = extractelement <3 x float> %6, i64 1
+  store float %7, float* %y, align 4
+  %8 = load <3 x float>, <3 x float>* %axis, align 16
+  %9 = extractelement <3 x float> %8, i64 2
+  store float %9, float* %z, align 4
+  %10 = bitcast [4 x <4 x float>]* %a to i8*
+  call void @llvm.memset.p0i8.i64(i8* align 16 %10, i8 0, i64 64, i1 false)
+  %11 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %a, i64 0, i64 0
+  %12 = load float, float* %cosr, align 4
+  %13 = load float, float* %x, align 4
+  %14 = load float, float* %x, align 4
+  %fmul = fmul float %13, %14
+  %15 = load float, float* %cosr, align 4
+  %fpfpext = fpext float %15 to double
+  %fsub = fsub double 1.000000e+00, %fpfpext
+  %fpfptrunc = fptrunc double %fsub to float
+  %fmul1 = fmul float %fmul, %fpfptrunc
+  %fadd = fadd float %12, %fmul1
+  %16 = insertelement <4 x float> undef, float %fadd, i64 0
+  %17 = load float, float* %x, align 4
+  %18 = load float, float* %y, align 4
+  %fmul2 = fmul float %17, %18
+  %19 = load float, float* %cosr, align 4
+  %fpfpext3 = fpext float %19 to double
+  %fsub4 = fsub double 1.000000e+00, %fpfpext3
+  %fpfptrunc5 = fptrunc double %fsub4 to float
+  %fmul6 = fmul float %fmul2, %fpfptrunc5
+  %20 = load float, float* %z, align 4
+  %21 = load float, float* %sinr, align 4
+  %fmul7 = fmul float %20, %21
+  %fsub8 = fsub float %fmul6, %fmul7
+  %22 = insertelement <4 x float> %16, float %fsub8, i64 1
+  %23 = load float, float* %x, align 4
+  %24 = load float, float* %z, align 4
+  %fmul9 = fmul float %23, %24
+  %25 = load float, float* %cosr, align 4
+  %fpfpext10 = fpext float %25 to double
+  %fsub11 = fsub double 1.000000e+00, %fpfpext10
+  %fpfptrunc12 = fptrunc double %fsub11 to float
+  %fmul13 = fmul float %fmul9, %fpfptrunc12
+  %26 = load float, float* %y, align 4
+  %27 = load float, float* %sinr, align 4
+  %fmul14 = fmul float %26, %27
+  %fadd15 = fadd float %fmul13, %fmul14
+  %28 = insertelement <4 x float> %22, float %fadd15, i64 2
+  %29 = insertelement <4 x float> %28, float 0.000000e+00, i64 3
+  store <4 x float> %29, <4 x float>* %11, align 16
+  %30 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %a, i64 0, i64 1
+  %31 = load float, float* %y, align 4
+  %32 = load float, float* %x, align 4
+  %fmul16 = fmul float %31, %32
+  %33 = load float, float* %cosr, align 4
+  %fpfpext17 = fpext float %33 to double
+  %fsub18 = fsub double 1.000000e+00, %fpfpext17
+  %fpfptrunc19 = fptrunc double %fsub18 to float
+  %fmul20 = fmul float %fmul16, %fpfptrunc19
+  %34 = load float, float* %z, align 4
+  %35 = load float, float* %sinr, align 4
+  %fmul21 = fmul float %34, %35
+  %fadd22 = fadd float %fmul20, %fmul21
+  %36 = insertelement <4 x float> undef, float %fadd22, i64 0
+  %37 = load float, float* %cosr, align 4
+  %38 = load float, float* %y, align 4
+  %39 = load float, float* %y, align 4
+  %fmul23 = fmul float %38, %39
+  %40 = load float, float* %cosr, align 4
+  %fpfpext24 = fpext float %40 to double
+  %fsub25 = fsub double 1.000000e+00, %fpfpext24
+  %fpfptrunc26 = fptrunc double %fsub25 to float
+  %fmul27 = fmul float %fmul23, %fpfptrunc26
+  %fadd28 = fadd float %37, %fmul27
+  %41 = insertelement <4 x float> %36, float %fadd28, i64 1
+  %42 = load float, float* %y, align 4
+  %43 = load float, float* %z, align 4
+  %fmul29 = fmul float %42, %43
+  %44 = load float, float* %cosr, align 4
+  %fpfpext30 = fpext float %44 to double
+  %fsub31 = fsub double 1.000000e+00, %fpfpext30
+  %fpfptrunc32 = fptrunc double %fsub31 to float
+  %fmul33 = fmul float %fmul29, %fpfptrunc32
+  %45 = load float, float* %x, align 4
+  %46 = load float, float* %sinr, align 4
+  %fmul34 = fmul float %45, %46
+  %fsub35 = fsub float %fmul33, %fmul34
+  %47 = insertelement <4 x float> %41, float %fsub35, i64 2
+  %48 = insertelement <4 x float> %47, float 0.000000e+00, i64 3
+  store <4 x float> %48, <4 x float>* %30, align 16
+  %49 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %a, i64 0, i64 2
+  %50 = load float, float* %z, align 4
+  %51 = load float, float* %x, align 4
+  %fmul36 = fmul float %50, %51
+  %52 = load float, float* %cosr, align 4
+  %fpfpext37 = fpext float %52 to double
+  %fsub38 = fsub double 1.000000e+00, %fpfpext37
+  %fpfptrunc39 = fptrunc double %fsub38 to float
+  %fmul40 = fmul float %fmul36, %fpfptrunc39
+  %53 = load float, float* %y, align 4
+  %54 = load float, float* %sinr, align 4
+  %fmul41 = fmul float %53, %54
+  %fsub42 = fsub float %fmul40, %fmul41
+  %55 = insertelement <4 x float> undef, float %fsub42, i64 0
+  %56 = load float, float* %z, align 4
+  %57 = load float, float* %y, align 4
+  %fmul43 = fmul float %56, %57
+  %58 = load float, float* %cosr, align 4
+  %fpfpext44 = fpext float %58 to double
+  %fsub45 = fsub double 1.000000e+00, %fpfpext44
+  %fpfptrunc46 = fptrunc double %fsub45 to float
+  %fmul47 = fmul float %fmul43, %fpfptrunc46
+  %59 = load float, float* %x, align 4
+  %60 = load float, float* %sinr, align 4
+  %fmul48 = fmul float %59, %60
+  %fadd49 = fadd float %fmul47, %fmul48
+  %61 = insertelement <4 x float> %55, float %fadd49, i64 1
+  %62 = load float, float* %cosr, align 4
+  %63 = load float, float* %z, align 4
+  %64 = load float, float* %z, align 4
+  %fmul50 = fmul float %63, %64
+  %65 = load float, float* %cosr, align 4
+  %fpfpext51 = fpext float %65 to double
+  %fsub52 = fsub double 1.000000e+00, %fpfpext51
+  %fpfptrunc53 = fptrunc double %fsub52 to float
+  %fmul54 = fmul float %fmul50, %fpfptrunc53
+  %fadd55 = fadd float %62, %fmul54
+  %66 = insertelement <4 x float> %61, float %fadd55, i64 2
+  %67 = insertelement <4 x float> %66, float 0.000000e+00, i64 3
+  store <4 x float> %67, <4 x float>* %49, align 16
+  %68 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %a, i64 0, i64 3
+  store <4 x float> <float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 1.000000e+00>, <4 x float>* %68, align 16
+  %69 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %b, i64 0, i64 0
+  %70 = load float, float* %cosr, align 4
+  %71 = load float, float* %x, align 4
+  %72 = load float, float* %x, align 4
+  %fmul56 = fmul float %71, %72
+  %73 = load float, float* %cosr, align 4
+  %fpfpext57 = fpext float %73 to double
+  %fsub58 = fsub double 1.000000e+00, %fpfpext57
+  %fpfptrunc59 = fptrunc double %fsub58 to float
+  %fmul60 = fmul float %fmul56, %fpfptrunc59
+  %fadd61 = fadd float %70, %fmul60
+  %74 = insertelement <4 x float> undef, float %fadd61, i64 0
+  %75 = load float, float* %x, align 4
+  %76 = load float, float* %y, align 4
+  %fmul62 = fmul float %75, %76
+  %77 = load float, float* %cosr, align 4
+  %fpfpext63 = fpext float %77 to double
+  %fsub64 = fsub double 1.000000e+00, %fpfpext63
+  %fpfptrunc65 = fptrunc double %fsub64 to float
+  %fmul66 = fmul float %fmul62, %fpfptrunc65
+  %78 = load float, float* %z, align 4
+  %79 = load float, float* %sinr, align 4
+  %fmul67 = fmul float %78, %79
+  %fsub68 = fsub float %fmul66, %fmul67
+  %80 = insertelement <4 x float> %74, float %fsub68, i64 1
+  %81 = load float, float* %x, align 4
+  %82 = load float, float* %z, align 4
+  %fmul69 = fmul float %81, %82
+  %83 = load float, float* %cosr, align 4
+  %fpfpext70 = fpext float %83 to double
+  %fsub71 = fsub double 1.000000e+00, %fpfpext70
+  %fpfptrunc72 = fptrunc double %fsub71 to float
+  %fmul73 = fmul float %fmul69, %fpfptrunc72
+  %84 = load float, float* %y, align 4
+  %85 = load float, float* %sinr, align 4
+  %fmul74 = fmul float %84, %85
+  %fadd75 = fadd float %fmul73, %fmul74
+  %86 = insertelement <4 x float> %80, float %fadd75, i64 2
+  %87 = insertelement <4 x float> %86, float 0.000000e+00, i64 3
+  store <4 x float> %87, <4 x float>* %69, align 16
+  %88 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %b, i64 0, i64 1
+  %89 = load float, float* %y, align 4
+  %90 = load float, float* %x, align 4
+  %fmul76 = fmul float %89, %90
+  %91 = load float, float* %cosr, align 4
+  %fpfpext77 = fpext float %91 to double
+  %fsub78 = fsub double 1.000000e+00, %fpfpext77
+  %fpfptrunc79 = fptrunc double %fsub78 to float
+  %fmul80 = fmul float %fmul76, %fpfptrunc79
+  %92 = load float, float* %z, align 4
+  %93 = load float, float* %sinr, align 4
+  %fmul81 = fmul float %92, %93
+  %fadd82 = fadd float %fmul80, %fmul81
+  %94 = insertelement <4 x float> undef, float %fadd82, i64 0
+  %95 = load float, float* %cosr, align 4
+  %96 = load float, float* %y, align 4
+  %97 = load float, float* %y, align 4
+  %fmul83 = fmul float %96, %97
+  %98 = load float, float* %cosr, align 4
+  %fpfpext84 = fpext float %98 to double
+  %fsub85 = fsub double 1.000000e+00, %fpfpext84
+  %fpfptrunc86 = fptrunc double %fsub85 to float
+  %fmul87 = fmul float %fmul83, %fpfptrunc86
+  %fadd88 = fadd float %95, %fmul87
+  %99 = insertelement <4 x float> %94, float %fadd88, i64 1
+  %100 = load float, float* %y, align 4
+  %101 = load float, float* %z, align 4
+  %fmul89 = fmul float %100, %101
+  %102 = load float, float* %cosr, align 4
+  %fpfpext90 = fpext float %102 to double
+  %fsub91 = fsub double 1.000000e+00, %fpfpext90
+  %fpfptrunc92 = fptrunc double %fsub91 to float
+  %fmul93 = fmul float %fmul89, %fpfptrunc92
+  %103 = load float, float* %x, align 4
+  %104 = load float, float* %sinr, align 4
+  %fmul94 = fmul float %103, %104
+  %fsub95 = fsub float %fmul93, %fmul94
+  %105 = insertelement <4 x float> %99, float %fsub95, i64 2
+  %106 = insertelement <4 x float> %105, float 0.000000e+00, i64 3
+  store <4 x float> %106, <4 x float>* %88, align 16
+  %107 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %b, i64 0, i64 2
+  %108 = load float, float* %z, align 4
+  %109 = load float, float* %x, align 4
+  %fmul96 = fmul float %108, %109
+  %110 = load float, float* %cosr, align 4
+  %fpfpext97 = fpext float %110 to double
+  %fsub98 = fsub double 1.000000e+00, %fpfpext97
+  %fpfptrunc99 = fptrunc double %fsub98 to float
+  %fmul100 = fmul float %fmul96, %fpfptrunc99
+  %111 = load float, float* %y, align 4
+  %112 = load float, float* %sinr, align 4
+  %fmul101 = fmul float %111, %112
+  %fsub102 = fsub float %fmul100, %fmul101
+  %113 = insertelement <4 x float> undef, float %fsub102, i64 0
+  %114 = load float, float* %z, align 4
+  %115 = load float, float* %y, align 4
+  %fmul103 = fmul float %114, %115
+  %116 = load float, float* %cosr, align 4
+  %fpfpext104 = fpext float %116 to double
+  %fsub105 = fsub double 1.000000e+00, %fpfpext104
+  %fpfptrunc106 = fptrunc double %fsub105 to float
+  %fmul107 = fmul float %fmul103, %fpfptrunc106
+  %117 = load float, float* %x, align 4
+  %118 = load float, float* %sinr, align 4
+  %fmul108 = fmul float %117, %118
+  %fadd109 = fadd float %fmul107, %fmul108
+  %119 = insertelement <4 x float> %113, float %fadd109, i64 1
+  %120 = load float, float* %cosr, align 4
+  %121 = load float, float* %z, align 4
+  %122 = load float, float* %z, align 4
+  %fmul110 = fmul float %121, %122
+  %123 = load float, float* %cosr, align 4
+  %fpfpext111 = fpext float %123 to double
+  %fsub112 = fsub double 1.000000e+00, %fpfpext111
+  %fpfptrunc113 = fptrunc double %fsub112 to float
+  %fmul114 = fmul float %fmul110, %fpfptrunc113
+  %fadd115 = fadd float %120, %fmul114
+  %124 = insertelement <4 x float> %119, float %fadd115, i64 2
+  %125 = insertelement <4 x float> %124, float 0.000000e+00, i64 3
+  store <4 x float> %125, <4 x float>* %107, align 16
+  %126 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %b, i64 0, i64 3
+  store <4 x float> <float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 1.000000e+00>, <4 x float>* %126, align 16
+  store i64 0, i64* %.anon, align 8
+  br label %loop.cond
+
+loop.cond:                                        ; preds = %loop.body, %entry
+  %127 = load i64, i64* %.anon, align 8
+  %gt = icmp ugt i64 4, %127
+  br i1 %gt, label %loop.body, label %loop.exit
+
+loop.body:                                        ; preds = %loop.cond
+  %128 = load i64, i64* %.anon, align 8
+  %129 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %a, i64 0, i64 %128
+  %130 = load <4 x float>, <4 x float>* %129, align 16
+  store <4 x float> %130, <4 x float>* %v, align 16
+  %131 = load <4 x float>, <4 x float>* %v, align 16
+  %132 = extractelement <4 x float> %131, i64 0
+  %fpfpext116 = fpext float %132 to double
+  %133 = load <4 x float>, <4 x float>* %v, align 16
+  %134 = extractelement <4 x float> %133, i64 1
+  %fpfpext117 = fpext float %134 to double
+  %135 = load <4 x float>, <4 x float>* %v, align 16
+  %136 = extractelement <4 x float> %135, i64 2
+  %fpfpext118 = fpext float %136 to double
+  %137 = load <4 x float>, <4 x float>* %v, align 16
+  %138 = extractelement <4 x float> %137, i64 3
+  %fpfpext119 = fpext float %138 to double
+  %139 = call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([16 x i8], [16 x i8]* @.str, i32 0, i32 0), double %fpfpext116, double %fpfpext117, double %fpfpext118, double %fpfpext119)
+  %140 = load i64, i64* %.anon, align 8
+  %add = add i64 %140, 1
+  store i64 %add, i64* %.anon, align 8
+  br label %loop.cond
+
+loop.exit:                                        ; preds = %loop.cond
+  %141 = call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([2 x i8], [2 x i8]* @.str.1, i32 0, i32 0))
+  store i64 0, i64* %.anon120, align 8
+  br label %loop.cond121
+
+loop.cond121:                                     ; preds = %loop.body123, %loop.exit
+  %142 = load i64, i64* %.anon120, align 8
+  %gt122 = icmp ugt i64 4, %142
+  br i1 %gt122, label %loop.body123, label %loop.exit130
+
+loop.body123:                                     ; preds = %loop.cond121
+  %143 = load i64, i64* %.anon120, align 8
+  %144 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %b, i64 0, i64 %143
+  %145 = load <4 x float>, <4 x float>* %144, align 16
+  store <4 x float> %145, <4 x float>* %v124, align 16
+  %146 = load <4 x float>, <4 x float>* %v124, align 16
+  %147 = extractelement <4 x float> %146, i64 0
+  %fpfpext125 = fpext float %147 to double
+  %148 = load <4 x float>, <4 x float>* %v124, align 16
+  %149 = extractelement <4 x float> %148, i64 1
+  %fpfpext126 = fpext float %149 to double
+  %150 = load <4 x float>, <4 x float>* %v124, align 16
+  %151 = extractelement <4 x float> %150, i64 2
+  %fpfpext127 = fpext float %151 to double
+  %152 = load <4 x float>, <4 x float>* %v124, align 16
+  %153 = extractelement <4 x float> %152, i64 3
+  %fpfpext128 = fpext float %153 to double
+  %154 = call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([16 x i8], [16 x i8]* @.str.2, i32 0, i32 0), double %fpfpext125, double %fpfpext126, double %fpfpext127, double %fpfpext128)
+  %155 = load i64, i64* %.anon120, align 8
+  %add129 = add i64 %155, 1
+  store i64 %add129, i64* %.anon120, align 8
+  br label %loop.cond121
+
+loop.exit130:                                     ; preds = %loop.cond121
+  ret void
+}
\ No newline at end of file
diff --git a/test/test_suite2/constants/float_type.c3t b/test/test_suite2/constants/float_type.c3t
new file mode 100644
index 000000000..5ee96b84e
--- /dev/null
+++ b/test/test_suite2/constants/float_type.c3t
@@ -0,0 +1,11 @@
+// #target: macos-x64
+module test;
+uint f0 = $typeof(1.0f).sizeof;
+uint f32 = $typeof(1.0f32).sizeof;
+uint f64 = $typeof(1.0f64).sizeof;
+
+/* #expect: test.ll
+
+@test_f0 = local_unnamed_addr global i32 4, align 4
+@test_f32 = local_unnamed_addr global i32 4, align 4
+@test_f64 = local_unnamed_addr global i32 8, align 4
\ No newline at end of file
diff --git a/test/test_suite2/vector/vector_init_regression.c3t b/test/test_suite2/vector/vector_init_regression.c3t
new file mode 100644
index 000000000..2ce7a8a5a
--- /dev/null
+++ b/test/test_suite2/vector/vector_init_regression.c3t
@@ -0,0 +1,428 @@
+// #target: macos-x64
+module test;
+extern fn int printf(char* format, ...);
+
+fn void main() {
+    float radians = 3.1415 / 4;
+    float[<3>] axis = {0.0, 0.0, 1.0};
+
+    float cosr = (float) $$cos(radians);
+    float sinr = (float) $$sin(radians);
+    float x = axis[0];
+    float y = axis[1];
+    float z = axis[2];
+
+    float[<4>][4] a = {};
+    a[0] = {
+        cosr + (x * x) * (float) (1.0 - cosr),
+        (x * y) * (float) (1.0 - cosr) - (z * sinr),
+        (x * z) * (float) (1.0 - cosr) + (y * sinr),
+        0.0
+    };
+
+    a[1] = {
+        (y * x) * (float) (1.0 - cosr) + (z * sinr),
+        cosr + (y * y) * (float) (1.0 - cosr),
+        (y * z) * (float) (1.0 - cosr) - (x * sinr),
+        0.0
+    };
+
+    a[2] = {
+        (z * x) * (float) (1.0 - cosr) - (y * sinr),
+        (z * y) * (float) (1.0 - cosr) + (x * sinr),
+        cosr + (z * z) * (float) (1.0 - cosr),
+        0.0
+    };
+
+    a[3] = {
+        0.0,
+        0.0,
+        0.0,
+        1.0
+    };
+
+    float[<4>][4] b = {
+        {
+            cosr + (x * x) * (float) (1.0 - cosr),
+            (x * y) * (float) (1.0 - cosr) - (z * sinr),
+            (x * z) * (float) (1.0 - cosr) + (y * sinr),
+            0.0
+        },
+        {
+            (y * x) * (float) (1.0 - cosr) + (z * sinr),
+            cosr + (y * y) * (float) (1.0 - cosr),
+            (y * z) * (float) (1.0 - cosr) - (x * sinr),
+            0.0
+        },
+        {
+            (z * x) * (float) (1.0 - cosr) - (y * sinr),
+            (z * y) * (float) (1.0 - cosr) + (x * sinr),
+            cosr + (z * z) * (float) (1.0 - cosr),
+            0.0
+        },
+        {
+            0.0,
+            0.0,
+            0.0,
+            1.0
+        }
+    };
+
+    foreach(v : a) {
+        printf("A: %f %f %f %f\n", v[0], v[1], v[2], v[3]);
+    }
+
+    printf("\n");
+
+    foreach(v : b) {
+        printf("B: %f %f %f %f\n", v[0], v[1], v[2], v[3]);
+    }
+}
+
+/* #expect: test.ll
+
+define void @test_main() #0 {
+entry:
+  %radians = alloca float, align 4
+  %axis = alloca <3 x float>, align 16
+  %cosr = alloca float, align 4
+  %sinr = alloca float, align 4
+  %x = alloca float, align 4
+  %y = alloca float, align 4
+  %z = alloca float, align 4
+  %a = alloca [4 x <4 x float>], align 16
+  %b = alloca [4 x <4 x float>], align 16
+  %.anon = alloca i64, align 8
+  %v = alloca <4 x float>, align 16
+  %.anon120 = alloca i64, align 8
+  %v124 = alloca <4 x float>, align 16
+  store float 0x3FE921CAC0000000, float* %radians, align 4
+  store <3 x float> <float 0.000000e+00, float 0.000000e+00, float 1.000000e+00>, <3 x float>* %axis, align 16
+  %0 = load float, float* %radians, align 4
+  %1 = call float @llvm.cos.f32(float %0)
+  store float %1, float* %cosr, align 4
+  %2 = load float, float* %radians, align 4
+  %3 = call float @llvm.sin.f32(float %2)
+  store float %3, float* %sinr, align 4
+  %4 = load <3 x float>, <3 x float>* %axis, align 16
+  %5 = extractelement <3 x float> %4, i64 0
+  store float %5, float* %x, align 4
+  %6 = load <3 x float>, <3 x float>* %axis, align 16
+  %7 = extractelement <3 x float> %6, i64 1
+  store float %7, float* %y, align 4
+  %8 = load <3 x float>, <3 x float>* %axis, align 16
+  %9 = extractelement <3 x float> %8, i64 2
+  store float %9, float* %z, align 4
+  %10 = bitcast [4 x <4 x float>]* %a to i8*
+  call void @llvm.memset.p0i8.i64(i8* align 16 %10, i8 0, i64 64, i1 false)
+  %11 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %a, i64 0, i64 0
+  %12 = load float, float* %cosr, align 4
+  %13 = load float, float* %x, align 4
+  %14 = load float, float* %x, align 4
+  %fmul = fmul float %13, %14
+  %15 = load float, float* %cosr, align 4
+  %fpfpext = fpext float %15 to double
+  %fsub = fsub double 1.000000e+00, %fpfpext
+  %fpfptrunc = fptrunc double %fsub to float
+  %fmul1 = fmul float %fmul, %fpfptrunc
+  %fadd = fadd float %12, %fmul1
+  %16 = insertelement <4 x float> undef, float %fadd, i64 0
+  %17 = load float, float* %x, align 4
+  %18 = load float, float* %y, align 4
+  %fmul2 = fmul float %17, %18
+  %19 = load float, float* %cosr, align 4
+  %fpfpext3 = fpext float %19 to double
+  %fsub4 = fsub double 1.000000e+00, %fpfpext3
+  %fpfptrunc5 = fptrunc double %fsub4 to float
+  %fmul6 = fmul float %fmul2, %fpfptrunc5
+  %20 = load float, float* %z, align 4
+  %21 = load float, float* %sinr, align 4
+  %fmul7 = fmul float %20, %21
+  %fsub8 = fsub float %fmul6, %fmul7
+  %22 = insertelement <4 x float> %16, float %fsub8, i64 1
+  %23 = load float, float* %x, align 4
+  %24 = load float, float* %z, align 4
+  %fmul9 = fmul float %23, %24
+  %25 = load float, float* %cosr, align 4
+  %fpfpext10 = fpext float %25 to double
+  %fsub11 = fsub double 1.000000e+00, %fpfpext10
+  %fpfptrunc12 = fptrunc double %fsub11 to float
+  %fmul13 = fmul float %fmul9, %fpfptrunc12
+  %26 = load float, float* %y, align 4
+  %27 = load float, float* %sinr, align 4
+  %fmul14 = fmul float %26, %27
+  %fadd15 = fadd float %fmul13, %fmul14
+  %28 = insertelement <4 x float> %22, float %fadd15, i64 2
+  %29 = insertelement <4 x float> %28, float 0.000000e+00, i64 3
+  store <4 x float> %29, <4 x float>* %11, align 16
+  %30 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %a, i64 0, i64 1
+  %31 = load float, float* %y, align 4
+  %32 = load float, float* %x, align 4
+  %fmul16 = fmul float %31, %32
+  %33 = load float, float* %cosr, align 4
+  %fpfpext17 = fpext float %33 to double
+  %fsub18 = fsub double 1.000000e+00, %fpfpext17
+  %fpfptrunc19 = fptrunc double %fsub18 to float
+  %fmul20 = fmul float %fmul16, %fpfptrunc19
+  %34 = load float, float* %z, align 4
+  %35 = load float, float* %sinr, align 4
+  %fmul21 = fmul float %34, %35
+  %fadd22 = fadd float %fmul20, %fmul21
+  %36 = insertelement <4 x float> undef, float %fadd22, i64 0
+  %37 = load float, float* %cosr, align 4
+  %38 = load float, float* %y, align 4
+  %39 = load float, float* %y, align 4
+  %fmul23 = fmul float %38, %39
+  %40 = load float, float* %cosr, align 4
+  %fpfpext24 = fpext float %40 to double
+  %fsub25 = fsub double 1.000000e+00, %fpfpext24
+  %fpfptrunc26 = fptrunc double %fsub25 to float
+  %fmul27 = fmul float %fmul23, %fpfptrunc26
+  %fadd28 = fadd float %37, %fmul27
+  %41 = insertelement <4 x float> %36, float %fadd28, i64 1
+  %42 = load float, float* %y, align 4
+  %43 = load float, float* %z, align 4
+  %fmul29 = fmul float %42, %43
+  %44 = load float, float* %cosr, align 4
+  %fpfpext30 = fpext float %44 to double
+  %fsub31 = fsub double 1.000000e+00, %fpfpext30
+  %fpfptrunc32 = fptrunc double %fsub31 to float
+  %fmul33 = fmul float %fmul29, %fpfptrunc32
+  %45 = load float, float* %x, align 4
+  %46 = load float, float* %sinr, align 4
+  %fmul34 = fmul float %45, %46
+  %fsub35 = fsub float %fmul33, %fmul34
+  %47 = insertelement <4 x float> %41, float %fsub35, i64 2
+  %48 = insertelement <4 x float> %47, float 0.000000e+00, i64 3
+  store <4 x float> %48, <4 x float>* %30, align 16
+  %49 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %a, i64 0, i64 2
+  %50 = load float, float* %z, align 4
+  %51 = load float, float* %x, align 4
+  %fmul36 = fmul float %50, %51
+  %52 = load float, float* %cosr, align 4
+  %fpfpext37 = fpext float %52 to double
+  %fsub38 = fsub double 1.000000e+00, %fpfpext37
+  %fpfptrunc39 = fptrunc double %fsub38 to float
+  %fmul40 = fmul float %fmul36, %fpfptrunc39
+  %53 = load float, float* %y, align 4
+  %54 = load float, float* %sinr, align 4
+  %fmul41 = fmul float %53, %54
+  %fsub42 = fsub float %fmul40, %fmul41
+  %55 = insertelement <4 x float> undef, float %fsub42, i64 0
+  %56 = load float, float* %z, align 4
+  %57 = load float, float* %y, align 4
+  %fmul43 = fmul float %56, %57
+  %58 = load float, float* %cosr, align 4
+  %fpfpext44 = fpext float %58 to double
+  %fsub45 = fsub double 1.000000e+00, %fpfpext44
+  %fpfptrunc46 = fptrunc double %fsub45 to float
+  %fmul47 = fmul float %fmul43, %fpfptrunc46
+  %59 = load float, float* %x, align 4
+  %60 = load float, float* %sinr, align 4
+  %fmul48 = fmul float %59, %60
+  %fadd49 = fadd float %fmul47, %fmul48
+  %61 = insertelement <4 x float> %55, float %fadd49, i64 1
+  %62 = load float, float* %cosr, align 4
+  %63 = load float, float* %z, align 4
+  %64 = load float, float* %z, align 4
+  %fmul50 = fmul float %63, %64
+  %65 = load float, float* %cosr, align 4
+  %fpfpext51 = fpext float %65 to double
+  %fsub52 = fsub double 1.000000e+00, %fpfpext51
+  %fpfptrunc53 = fptrunc double %fsub52 to float
+  %fmul54 = fmul float %fmul50, %fpfptrunc53
+  %fadd55 = fadd float %62, %fmul54
+  %66 = insertelement <4 x float> %61, float %fadd55, i64 2
+  %67 = insertelement <4 x float> %66, float 0.000000e+00, i64 3
+  store <4 x float> %67, <4 x float>* %49, align 16
+  %68 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %a, i64 0, i64 3
+  store <4 x float> <float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 1.000000e+00>, <4 x float>* %68, align 16
+  %69 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %b, i64 0, i64 0
+  %70 = load float, float* %cosr, align 4
+  %71 = load float, float* %x, align 4
+  %72 = load float, float* %x, align 4
+  %fmul56 = fmul float %71, %72
+  %73 = load float, float* %cosr, align 4
+  %fpfpext57 = fpext float %73 to double
+  %fsub58 = fsub double 1.000000e+00, %fpfpext57
+  %fpfptrunc59 = fptrunc double %fsub58 to float
+  %fmul60 = fmul float %fmul56, %fpfptrunc59
+  %fadd61 = fadd float %70, %fmul60
+  %74 = insertelement <4 x float> undef, float %fadd61, i64 0
+  %75 = load float, float* %x, align 4
+  %76 = load float, float* %y, align 4
+  %fmul62 = fmul float %75, %76
+  %77 = load float, float* %cosr, align 4
+  %fpfpext63 = fpext float %77 to double
+  %fsub64 = fsub double 1.000000e+00, %fpfpext63
+  %fpfptrunc65 = fptrunc double %fsub64 to float
+  %fmul66 = fmul float %fmul62, %fpfptrunc65
+  %78 = load float, float* %z, align 4
+  %79 = load float, float* %sinr, align 4
+  %fmul67 = fmul float %78, %79
+  %fsub68 = fsub float %fmul66, %fmul67
+  %80 = insertelement <4 x float> %74, float %fsub68, i64 1
+  %81 = load float, float* %x, align 4
+  %82 = load float, float* %z, align 4
+  %fmul69 = fmul float %81, %82
+  %83 = load float, float* %cosr, align 4
+  %fpfpext70 = fpext float %83 to double
+  %fsub71 = fsub double 1.000000e+00, %fpfpext70
+  %fpfptrunc72 = fptrunc double %fsub71 to float
+  %fmul73 = fmul float %fmul69, %fpfptrunc72
+  %84 = load float, float* %y, align 4
+  %85 = load float, float* %sinr, align 4
+  %fmul74 = fmul float %84, %85
+  %fadd75 = fadd float %fmul73, %fmul74
+  %86 = insertelement <4 x float> %80, float %fadd75, i64 2
+  %87 = insertelement <4 x float> %86, float 0.000000e+00, i64 3
+  store <4 x float> %87, <4 x float>* %69, align 16
+  %88 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %b, i64 0, i64 1
+  %89 = load float, float* %y, align 4
+  %90 = load float, float* %x, align 4
+  %fmul76 = fmul float %89, %90
+  %91 = load float, float* %cosr, align 4
+  %fpfpext77 = fpext float %91 to double
+  %fsub78 = fsub double 1.000000e+00, %fpfpext77
+  %fpfptrunc79 = fptrunc double %fsub78 to float
+  %fmul80 = fmul float %fmul76, %fpfptrunc79
+  %92 = load float, float* %z, align 4
+  %93 = load float, float* %sinr, align 4
+  %fmul81 = fmul float %92, %93
+  %fadd82 = fadd float %fmul80, %fmul81
+  %94 = insertelement <4 x float> undef, float %fadd82, i64 0
+  %95 = load float, float* %cosr, align 4
+  %96 = load float, float* %y, align 4
+  %97 = load float, float* %y, align 4
+  %fmul83 = fmul float %96, %97
+  %98 = load float, float* %cosr, align 4
+  %fpfpext84 = fpext float %98 to double
+  %fsub85 = fsub double 1.000000e+00, %fpfpext84
+  %fpfptrunc86 = fptrunc double %fsub85 to float
+  %fmul87 = fmul float %fmul83, %fpfptrunc86
+  %fadd88 = fadd float %95, %fmul87
+  %99 = insertelement <4 x float> %94, float %fadd88, i64 1
+  %100 = load float, float* %y, align 4
+  %101 = load float, float* %z, align 4
+  %fmul89 = fmul float %100, %101
+  %102 = load float, float* %cosr, align 4
+  %fpfpext90 = fpext float %102 to double
+  %fsub91 = fsub double 1.000000e+00, %fpfpext90
+  %fpfptrunc92 = fptrunc double %fsub91 to float
+  %fmul93 = fmul float %fmul89, %fpfptrunc92
+  %103 = load float, float* %x, align 4
+  %104 = load float, float* %sinr, align 4
+  %fmul94 = fmul float %103, %104
+  %fsub95 = fsub float %fmul93, %fmul94
+  %105 = insertelement <4 x float> %99, float %fsub95, i64 2
+  %106 = insertelement <4 x float> %105, float 0.000000e+00, i64 3
+  store <4 x float> %106, <4 x float>* %88, align 16
+  %107 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %b, i64 0, i64 2
+  %108 = load float, float* %z, align 4
+  %109 = load float, float* %x, align 4
+  %fmul96 = fmul float %108, %109
+  %110 = load float, float* %cosr, align 4
+  %fpfpext97 = fpext float %110 to double
+  %fsub98 = fsub double 1.000000e+00, %fpfpext97
+  %fpfptrunc99 = fptrunc double %fsub98 to float
+  %fmul100 = fmul float %fmul96, %fpfptrunc99
+  %111 = load float, float* %y, align 4
+  %112 = load float, float* %sinr, align 4
+  %fmul101 = fmul float %111, %112
+  %fsub102 = fsub float %fmul100, %fmul101
+  %113 = insertelement <4 x float> undef, float %fsub102, i64 0
+  %114 = load float, float* %z, align 4
+  %115 = load float, float* %y, align 4
+  %fmul103 = fmul float %114, %115
+  %116 = load float, float* %cosr, align 4
+  %fpfpext104 = fpext float %116 to double
+  %fsub105 = fsub double 1.000000e+00, %fpfpext104
+  %fpfptrunc106 = fptrunc double %fsub105 to float
+  %fmul107 = fmul float %fmul103, %fpfptrunc106
+  %117 = load float, float* %x, align 4
+  %118 = load float, float* %sinr, align 4
+  %fmul108 = fmul float %117, %118
+  %fadd109 = fadd float %fmul107, %fmul108
+  %119 = insertelement <4 x float> %113, float %fadd109, i64 1
+  %120 = load float, float* %cosr, align 4
+  %121 = load float, float* %z, align 4
+  %122 = load float, float* %z, align 4
+  %fmul110 = fmul float %121, %122
+  %123 = load float, float* %cosr, align 4
+  %fpfpext111 = fpext float %123 to double
+  %fsub112 = fsub double 1.000000e+00, %fpfpext111
+  %fpfptrunc113 = fptrunc double %fsub112 to float
+  %fmul114 = fmul float %fmul110, %fpfptrunc113
+  %fadd115 = fadd float %120, %fmul114
+  %124 = insertelement <4 x float> %119, float %fadd115, i64 2
+  %125 = insertelement <4 x float> %124, float 0.000000e+00, i64 3
+  store <4 x float> %125, <4 x float>* %107, align 16
+  %126 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %b, i64 0, i64 3
+  store <4 x float> <float 0.000000e+00, float 0.000000e+00, float 0.000000e+00, float 1.000000e+00>, <4 x float>* %126, align 16
+  store i64 0, i64* %.anon, align 8
+  br label %loop.cond
+
+loop.cond:                                        ; preds = %loop.body, %entry
+  %127 = load i64, i64* %.anon, align 8
+  %gt = icmp ugt i64 4, %127
+  br i1 %gt, label %loop.body, label %loop.exit
+
+loop.body:                                        ; preds = %loop.cond
+  %128 = load i64, i64* %.anon, align 8
+  %129 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %a, i64 0, i64 %128
+  %130 = load <4 x float>, <4 x float>* %129, align 16
+  store <4 x float> %130, <4 x float>* %v, align 16
+  %131 = load <4 x float>, <4 x float>* %v, align 16
+  %132 = extractelement <4 x float> %131, i64 0
+  %fpfpext116 = fpext float %132 to double
+  %133 = load <4 x float>, <4 x float>* %v, align 16
+  %134 = extractelement <4 x float> %133, i64 1
+  %fpfpext117 = fpext float %134 to double
+  %135 = load <4 x float>, <4 x float>* %v, align 16
+  %136 = extractelement <4 x float> %135, i64 2
+  %fpfpext118 = fpext float %136 to double
+  %137 = load <4 x float>, <4 x float>* %v, align 16
+  %138 = extractelement <4 x float> %137, i64 3
+  %fpfpext119 = fpext float %138 to double
+  %139 = call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([16 x i8], [16 x i8]* @.str, i32 0, i32 0), double %fpfpext116, double %fpfpext117, double %fpfpext118, double %fpfpext119)
+  %140 = load i64, i64* %.anon, align 8
+  %add = add i64 %140, 1
+  store i64 %add, i64* %.anon, align 8
+  br label %loop.cond
+
+loop.exit:                                        ; preds = %loop.cond
+  %141 = call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([2 x i8], [2 x i8]* @.str.1, i32 0, i32 0))
+  store i64 0, i64* %.anon120, align 8
+  br label %loop.cond121
+
+loop.cond121:                                     ; preds = %loop.body123, %loop.exit
+  %142 = load i64, i64* %.anon120, align 8
+  %gt122 = icmp ugt i64 4, %142
+  br i1 %gt122, label %loop.body123, label %loop.exit130
+
+loop.body123:                                     ; preds = %loop.cond121
+  %143 = load i64, i64* %.anon120, align 8
+  %144 = getelementptr inbounds [4 x <4 x float>], [4 x <4 x float>]* %b, i64 0, i64 %143
+  %145 = load <4 x float>, <4 x float>* %144, align 16
+  store <4 x float> %145, <4 x float>* %v124, align 16
+  %146 = load <4 x float>, <4 x float>* %v124, align 16
+  %147 = extractelement <4 x float> %146, i64 0
+  %fpfpext125 = fpext float %147 to double
+  %148 = load <4 x float>, <4 x float>* %v124, align 16
+  %149 = extractelement <4 x float> %148, i64 1
+  %fpfpext126 = fpext float %149 to double
+  %150 = load <4 x float>, <4 x float>* %v124, align 16
+  %151 = extractelement <4 x float> %150, i64 2
+  %fpfpext127 = fpext float %151 to double
+  %152 = load <4 x float>, <4 x float>* %v124, align 16
+  %153 = extractelement <4 x float> %152, i64 3
+  %fpfpext128 = fpext float %153 to double
+  %154 = call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([16 x i8], [16 x i8]* @.str.2, i32 0, i32 0), double %fpfpext125, double %fpfpext126, double %fpfpext127, double %fpfpext128)
+  %155 = load i64, i64* %.anon120, align 8
+  %add129 = add i64 %155, 1
+  store i64 %add129, i64* %.anon120, align 8
+  br label %loop.cond121
+
+loop.exit130:                                     ; preds = %loop.cond121
+  ret void
+}
\ No newline at end of file