precomp[15] = t3 // 16

	p256PointAddAsm(&t0, &t0, p)
	p256PointAddAsm(&t1, &t1, p)
	p256PointAddAsm(&t2, &t2, p)
	precomp[2] = t0 // 3
	precomp[4] = t1 // 5
	precomp[8] = t2 // 9

	p256PointDoubleAsm(&t0, &t0)
	p256PointDoubleAsm(&t1, &t1)
	precomp[5] = t0 // 6
	precomp[9] = t1 // 10

	p256PointAddAsm(&t2, &t0, p)
	p256PointAddAsm(&t1, &t1, p)
	precomp[6] = t2  // 7

#undef y3out
#undef z3out
#define y2in(off) (off + 32)(b_ptr)
#define x3out(off) (off)(b_ptr)
#define y3out(off) (off + 32)(b_ptr)
#define z3out(off) (off + 64)(b_ptr)
// func p256PointAddAsm(res, in1, in2 *P256Point) int
TEXT ·p256PointAddAsm(SB),0,$392-32
	// See https://hyperelliptic.org/EFD/g1p/auto-shortw-jacobian-3.html#addition-add-2007-bl
	// Move input to stack in order to free registers
	MOVD	in1+8(FP), a_ptr

#define hsqr(off)  (32*17 + off)(SP)
#define rsqr(off)  (32*18 + off)(SP)
#define hcub(off)  (32*19 + off)(SP)
#define rptr       (32*20)(SP)
#define points_eq  (32*20+8)(SP)

//func p256PointAddAsm(res, in1, in2 *P256Point) int
TEXT ·p256PointAddAsm(SB),0,$680-32
	// See https://hyperelliptic.org/EFD/g1p/auto-shortw-jacobian-3.html#addition-add-2007-bl
	// Move input to stack in order to free registers
	MOVQ res+0(FP), AX

	// X=R ; Y-  ; MUL; U1=T // Y3 = R*Y3
	// X=S1; Y=T2; MUL; T-   // T2 = S1*T2
	// SUB(T<U1-T); Y3:=T    // Y3 = Y3-T2 << store-out Y3 result reg
	*/
// p256PointAddAsm(res, in1, in2 *p256Point)
TEXT ·p256PointAddAsm(SB), NOSPLIT, $16-32
	MOVD res+0(FP), P3ptr
	MOVD in1+8(FP), P1ptr
	MOVD $p256mul<>+0x00(SB), CPOOL
	MOVD $16, R16
	MOVD $32, R17
	MOVD $48, R18
	MOVD $64, R19

#undef X0
#undef X1
#undef Y0
#undef Y1
#undef T0
#undef T1
#undef PL
#undef PH
#undef Z3L
#undef Z3H
#undef ZER
#undef SEL1
#undef CAR1
#undef CAR2

// func p256PointAddAsm(res, in1, in2 *P256Point) int
#define P3ptr  R1
#define P1ptr  R2
#define P2ptr  R3
#define CPOOL  R4
#define ISZERO R5
#define TRUE   R6

// Temporaries in REGs
#define T1L   V16