diff --git a/src/compiler/enums.h b/src/compiler/enums.h
index 473b01cb6..7473c9bfb 100644
--- a/src/compiler/enums.h
+++ b/src/compiler/enums.h
@@ -826,6 +826,7 @@ typedef enum
 	BUILTIN_MEMSET,
 	BUILTIN_MEMMOVE,
 	BUILTIN_MIN,
+	BUILTIN_MULADD,
 	BUILTIN_NEARBYINT,
 	BUILTIN_POPCOUNT,
 	BUILTIN_POW,
diff --git a/src/compiler/llvm_codegen.c b/src/compiler/llvm_codegen.c
index f169a0db1..5d3981dec 100644
--- a/src/compiler/llvm_codegen.c
+++ b/src/compiler/llvm_codegen.c
@@ -654,6 +654,7 @@ static void llvm_codegen_setup()
 	intrinsic_id.memmove = lookup_intrinsic("llvm.memmove");
 	intrinsic_id.minimum = lookup_intrinsic("llvm.minimum");
 	intrinsic_id.minnum = lookup_intrinsic("llvm.minnum");
+	intrinsic_id.muladd = lookup_intrinsic("llvm.fmuladd");
 	intrinsic_id.nearbyint = lookup_intrinsic("llvm.nearbyint");
 	intrinsic_id.pow = lookup_intrinsic("llvm.pow");
 	intrinsic_id.powi = lookup_intrinsic("llvm.powi");
diff --git a/src/compiler/llvm_codegen_builtins.c b/src/compiler/llvm_codegen_builtins.c
index e01e41377..63dd4c858 100644
--- a/src/compiler/llvm_codegen_builtins.c
+++ b/src/compiler/llvm_codegen_builtins.c
@@ -507,6 +507,9 @@ void llvm_emit_builtin_call(GenContext *c, BEValue *result_value, Expr *expr)
 		case BUILTIN_LOG10:
 			llvm_emit_simple_builtin(c, result_value, expr, intrinsic_id.log10);
 			return;
+		case BUILTIN_MULADD:
+			llvm_emit_simple_builtin(c, result_value, expr, intrinsic_id.muladd);
+			return;
 		case BUILTIN_POW:
 			llvm_emit_simple_builtin(c, result_value, expr, intrinsic_id.pow);
 			return;
diff --git a/src/compiler/llvm_codegen_internal.h b/src/compiler/llvm_codegen_internal.h
index dd94f1331..6693c68c9 100644
--- a/src/compiler/llvm_codegen_internal.h
+++ b/src/compiler/llvm_codegen_internal.h
@@ -149,6 +149,7 @@ typedef struct
 	unsigned memmove;
 	unsigned minimum;
 	unsigned minnum;
+	unsigned muladd;
 	unsigned nearbyint;
 	unsigned pow;
 	unsigned powi;
diff --git a/src/compiler/sema_builtins.c b/src/compiler/sema_builtins.c
index 530b99ed7..25ac6927f 100644
--- a/src/compiler/sema_builtins.c
+++ b/src/compiler/sema_builtins.c
@@ -463,6 +463,13 @@ bool sema_expr_analyse_builtin_call(SemaContext *context, Expr *expr)
 			if (!sema_check_builtin_args_match(args, arg_count)) return false;
 			rtype = args[0]->type;
 			break;
+		case BUILTIN_MULADD:
+			if (!sema_check_builtin_args(args,
+										 (BuiltinArg[]) { BA_FLOAT, BA_FLOAT, BA_FLOAT },
+										 arg_count)) return false;
+			if (!sema_check_builtin_args_match(args, arg_count)) return false;
+			rtype = args[0]->type;
+			break;
 		case BUILTIN_VOLATILE_LOAD:
 		{
 			if (!sema_check_builtin_args(args, (BuiltinArg[]) { BA_POINTER }, 1)) return false;
@@ -553,6 +560,7 @@ static inline unsigned builtin_expected_args(BuiltinFunction func)
 		case BUILTIN_FMA:
 		case BUILTIN_FSHL:
 		case BUILTIN_FSHR:
+		case BUILTIN_MULADD:
 		case BUILTIN_PREFETCH:
 			return 3;
 		case BUILTIN_MEMSET:
diff --git a/src/compiler/symtab.c b/src/compiler/symtab.c
index 21d4d0c27..9edd3b949 100644
--- a/src/compiler/symtab.c
+++ b/src/compiler/symtab.c
@@ -200,6 +200,7 @@ void symtab_init(uint32_t capacity)
 	builtin_list[BUILTIN_MEMCOPY] = KW_DEF("memcpy");
 	builtin_list[BUILTIN_MEMSET] = KW_DEF("memset");
 	builtin_list[BUILTIN_MEMMOVE] = KW_DEF("memmove");
+	builtin_list[BUILTIN_MULADD] = KW_DEF("muladd");
 	builtin_list[BUILTIN_NEARBYINT] = KW_DEF("nearbyint");
 	builtin_list[BUILTIN_POPCOUNT] = KW_DEF("popcount");
 	builtin_list[BUILTIN_POW] = KW_DEF("pow");
diff --git a/test/test_suite/builtins/simple_builtins.c3t b/test/test_suite/builtins/simple_builtins.c3t
index b22cbbaeb..6d586c9e0 100644
--- a/test/test_suite/builtins/simple_builtins.c3t
+++ b/test/test_suite/builtins/simple_builtins.c3t
@@ -6,6 +6,7 @@ fn int foo(double b)
     double d = $$ceil(b);
     double e = $$max(1.0, d);
     double f = $$fma(d, 2.0, 3.0);
+    double m = $$muladd(d, 2.0, 3.0);
     int xeb = 13;
 
     int[3] abcd;
@@ -16,13 +17,14 @@ fn int foo(double b)
     return 1;
 }
 
-// #expect: foo.ll
+/* #expect: foo.ll
 
 define i32 @foo_foo(double %0) #0 {
 entry:
   %d = alloca double, align 8
   %e = alloca double, align 8
   %f = alloca double, align 8
+  %m = alloca double, align 8
   %xeb = alloca i32, align 4
   %abcd = alloca [3 x i32], align 4
   %sy = alloca i32, align 4
@@ -34,24 +36,27 @@ entry:
   %4 = load double, double* %d, align 8
   %5 = call double @llvm.fma.f64(double %4, double 2.000000e+00, double 3.000000e+00)
   store double %5, double* %f, align 8
+  %6 = load double, double* %d, align 8
+  %7 = call double @llvm.fmuladd.f64(double %6, double 2.000000e+00, double 3.000000e+00)
+  store double %7, double* %m, align 8
   store i32 13, i32* %xeb, align 4
-  %6 = getelementptr inbounds [3 x i32], [3 x i32]* %abcd, i64 0, i64 0
-  store i32 0, i32* %6, align 4
-  %7 = getelementptr inbounds [3 x i32], [3 x i32]* %abcd, i64 0, i64 1
-  store i32 0, i32* %7, align 4
-  %8 = getelementptr inbounds [3 x i32], [3 x i32]* %abcd, i64 0, i64 2
+  %8 = getelementptr inbounds [3 x i32], [3 x i32]* %abcd, i64 0, i64 0
   store i32 0, i32* %8, align 4
-  %9 = load volatile i32, i32* %xeb, align 4
-  store i32 %9, i32* %sy, align 4
-  %10 = load i32, i32* %sy, align 4
-  %add = add i32 %10, 1
-  store volatile i32 %add, i32* %xeb, align 4
-  %11 = getelementptr inbounds [3 x i32], [3 x i32]* %abcd, i64 0, i64 2
+  %9 = getelementptr inbounds [3 x i32], [3 x i32]* %abcd, i64 0, i64 1
+  store i32 0, i32* %9, align 4
+  %10 = getelementptr inbounds [3 x i32], [3 x i32]* %abcd, i64 0, i64 2
+  store i32 0, i32* %10, align 4
+  %11 = load volatile i32, i32* %xeb, align 4
+  store i32 %11, i32* %sy, align 4
   %12 = load i32, i32* %sy, align 4
-  %add1 = add i32 %12, 2
-  store volatile i32 %add1, i32* %11, align 4
+  %add = add i32 %12, 1
+  store volatile i32 %add, i32* %xeb, align 4
   %13 = getelementptr inbounds [3 x i32], [3 x i32]* %abcd, i64 0, i64 2
-  %14 = load volatile i32, i32* %13, align 4
-  store i32 %14, i32* %sy, align 4
+  %14 = load i32, i32* %sy, align 4
+  %add1 = add i32 %14, 2
+  store volatile i32 %add1, i32* %13, align 4
+  %15 = getelementptr inbounds [3 x i32], [3 x i32]* %abcd, i64 0, i64 2
+  %16 = load volatile i32, i32* %15, align 4
+  store i32 %16, i32* %sy, align 4
   ret i32 1
 }
diff --git a/test/test_suite2/builtins/simple_builtins.c3t b/test/test_suite2/builtins/simple_builtins.c3t
index 0f0278cb1..3cb43eba7 100644
--- a/test/test_suite2/builtins/simple_builtins.c3t
+++ b/test/test_suite2/builtins/simple_builtins.c3t
@@ -6,6 +6,7 @@ fn int foo(double b)
     double d = $$ceil(b);
     double e = $$max(1.0, d);
     double f = $$fma(d, 2.0, 3.0);
+    double m = $$muladd(d, 2.0, 3.0);
     int xeb = 13;
 
     int[3] abcd;
@@ -23,6 +24,7 @@ entry:
   %d = alloca double, align 8
   %e = alloca double, align 8
   %f = alloca double, align 8
+  %m = alloca double, align 8
   %xeb = alloca i32, align 4
   %abcd = alloca [3 x i32], align 4
   %sy = alloca i32, align 4
@@ -34,24 +36,27 @@ entry:
   %4 = load double, ptr %d, align 8
   %5 = call double @llvm.fma.f64(double %4, double 2.000000e+00, double 3.000000e+00)
   store double %5, ptr %f, align 8
+  %6 = load double, ptr %d, align 8
+  %7 = call double @llvm.fmuladd.f64(double %6, double 2.000000e+00, double 3.000000e+00)
+  store double %7, ptr %m, align 8
   store i32 13, ptr %xeb, align 4
-  %6 = getelementptr inbounds [3 x i32], ptr %abcd, i64 0, i64 0
-  store i32 0, ptr %6, align 4
-  %7 = getelementptr inbounds [3 x i32], ptr %abcd, i64 0, i64 1
-  store i32 0, ptr %7, align 4
-  %8 = getelementptr inbounds [3 x i32], ptr %abcd, i64 0, i64 2
+  %8 = getelementptr inbounds [3 x i32], ptr %abcd, i64 0, i64 0
   store i32 0, ptr %8, align 4
-  %9 = load volatile i32, ptr %xeb, align 4
-  store i32 %9, ptr %sy, align 4
-  %10 = load i32, ptr %sy, align 4
-  %add = add i32 %10, 1
-  store volatile i32 %add, ptr %xeb, align 4
-  %11 = getelementptr inbounds [3 x i32], ptr %abcd, i64 0, i64 2
+  %9 = getelementptr inbounds [3 x i32], ptr %abcd, i64 0, i64 1
+  store i32 0, ptr %9, align 4
+  %10 = getelementptr inbounds [3 x i32], ptr %abcd, i64 0, i64 2
+  store i32 0, ptr %10, align 4
+  %11 = load volatile i32, ptr %xeb, align 4
+  store i32 %11, ptr %sy, align 4
   %12 = load i32, ptr %sy, align 4
-  %add1 = add i32 %12, 2
-  store volatile i32 %add1, ptr %11, align 4
+  %add = add i32 %12, 1
+  store volatile i32 %add, ptr %xeb, align 4
   %13 = getelementptr inbounds [3 x i32], ptr %abcd, i64 0, i64 2
-  %14 = load volatile i32, ptr %13, align 4
-  store i32 %14, ptr %sy, align 4
+  %14 = load i32, ptr %sy, align 4
+  %add1 = add i32 %14, 2
+  store volatile i32 %add1, ptr %13, align 4
+  %15 = getelementptr inbounds [3 x i32], ptr %abcd, i64 0, i64 2
+  %16 = load volatile i32, ptr %15, align 4
+  store i32 %16, ptr %sy, align 4
   ret i32 1
 }