diff --git a/src/gen.cpp b/src/gen.cpp
index e1caf68..26a5439 100644
--- a/src/gen.cpp
+++ b/src/gen.cpp
@@ -346,6 +346,41 @@ struct Code : public mcl::Generator {
 		ret(Void);
 		endFunc();
 	}
+	void gen_mcl_fpDbl_add()
+	{
+		const int bu = bit + unit;
+		const int b2 = bit * 2;
+		const int b2u = b2 + unit;
+		resetGlobalIdx();
+		Operand pz(IntPtr, bit);
+		Operand px(IntPtr, b2);
+		Operand py(IntPtr, b2);
+		Operand pp(IntPtr, bit);
+		std::string name = "mcl_fpDbl_add" + cybozu::itoa(bit);
+		Function f(name, Void, pz, px, py, pp);
+		beginFunc(f);
+		Operand x = load(px);
+		Operand y = load(py);
+		x = zext(x, b2u);
+		y = zext(y, b2u);
+		Operand t = add(x, y); // x + y = [H:L]
+		Operand L = trunc(t, bit);
+		store(L, pz);
+
+		Operand H = lshr(t, bit);
+		H = trunc(H, bu);
+		Operand p = load(pp);
+		p = zext(p, bu);
+		Operand Hp = sub(H, p);
+		t = lshr(Hp, bit);
+		t = trunc(t, 1);
+		t = select(t, H, Hp);
+		t = trunc(t, bit);
+		pz = getelementptr(pz, makeImm(32, 1));
+		store(t, pz);
+		ret(Void);
+		endFunc();
+	}
 	void gen_all()
 	{
 		gen_mcl_fp_addsubNC(true);
@@ -355,6 +390,7 @@ struct Code : public mcl::Generator {
 	{
 		gen_mcl_fp_add();
 		gen_mcl_fp_sub();
+		gen_mcl_fpDbl_add();
 	}
 	void setBit(uint32_t bit)
 	{
diff --git a/src/short.txt b/src/short.txt
index 5e913a7..5f2099d 100644
--- a/src/short.txt
+++ b/src/short.txt
@@ -1,30 +1,6 @@
 @define bu = bit + unit
 @define b2 = bit * 2
 @define b2u = b2 + unit
-define void @mcl_fpDbl_add$(bit)(i$(bit)* %pz, i$(b2)* %px, i$(b2)* %py, i$(bit)* %pp) {
-entry:
-	%x = load i$(b2)* %px
-	%y = load i$(b2)* %py
-	%x1 = zext i$(b2) %x to i$(b2u)
-	%y1 = zext i$(b2) %y to i$(b2u)
-	%t0 = add i$(b2u) %x1, %y1 ; x + y = [H:L]
-	%t1 = trunc i$(b2u) %t0 to i$(bit) ; t1 = L
-	store i$(bit) %t1, i$(bit)* %pz ; store L
-
-	%t2 = lshr i$(b2u) %t0, $(bit)
-	%H = trunc i$(b2u) %t2 to i$(bu)
-
-	%p = load i$(bit)* %pp
-	%p1 = zext i$(bit) %p to i$(bu)
-	%Hp = sub i$(bu) %H, %p1 ; H - p
-	%t3 = lshr i$(bu) %Hp, $(bit)
-	%t4 = trunc i$(bu) %t3 to i1
-	%t5 = select i1 %t4, i$(bu) %H, i$(bu) %Hp
-	%t6 = trunc i$(bu) %t5 to i$(bit)
-	%pz1 = getelementptr i$(bit)* %pz, i32 1
-	store i$(bit) %t6, i$(bit)* %pz1
-	ret void
-}
 
 define void @mcl_fpDbl_sub$(bit)(i$(bit)* %pz, i$(b2)* %px, i$(b2)* %py, i$(bit)* %pp) {
 	%x = load i$(b2)* %px