x1, x6, x11, x12 := quarterRound(s.p1, s.p6, s.p11, fcr12)
		x2, x7, x8, x13 := quarterRound(s.p2, s.p7, fcr8, s.p13)
		x3, x4, x9, x14 := quarterRound(s.p3, fcr4, s.p9, s.p14)

		// The remaining 18 rounds.
		for i := 0; i < 9; i++ {
			// Column round.
			x0, x4, x8, x12 = quarterRound(x0, x4, x8, x12)
			x1, x5, x9, x13 = quarterRound(x1, x5, x9, x13)

	var x5 uint64
	var x6 uint64
	x5, x6 = bits.Sub64(arg1[2], arg2[2], uint64(fiatScalarUint1(x4)))
	var x7 uint64
	var x8 uint64
	x7, x8 = bits.Sub64(arg1[3], arg2[3], uint64(fiatScalarUint1(x6)))
	var x9 uint64
	fiatScalarCmovznzU64(&x9, fiatScalarUint1(x8), uint64(0x0), 0xffffffffffffffff)
	var x10 uint64
	var x11 uint64
	x10, x11 = bits.Add64(x1, (x9 & 0x5812631a5cf5d3ed), uint64(0x0))
	var x12 uint64

	var x5 uint64
	var x6 uint64
	x5, x6 = bits.Sub64(arg1[2], arg2[2], uint64(p256Uint1(x4)))
	var x7 uint64
	var x8 uint64
	x7, x8 = bits.Sub64(arg1[3], arg2[3], uint64(p256Uint1(x6)))
	var x9 uint64
	p256CmovznzU64(&x9, p256Uint1(x8), uint64(0x0), 0xffffffffffffffff)
	var x10 uint64
	var x11 uint64
	x10, x11 = bits.Add64(x1, x9, uint64(0x0))
	var x12 uint64
	var x13 uint64

	var x5 uint64
	var x6 uint64
	x5, x6 = bits.Sub64(arg1[2], arg2[2], uint64(p224Uint1(x4)))
	var x7 uint64
	var x8 uint64
	x7, x8 = bits.Sub64(arg1[3], arg2[3], uint64(p224Uint1(x6)))
	var x9 uint64
	p224CmovznzU64(&x9, p224Uint1(x8), uint64(0x0), 0xffffffffffffffff)
	var x10 uint64
	var x11 uint64
	x10, x11 = bits.Add64(x1, uint64((p224Uint1(x9) & 0x1)), uint64(0x0))
	var x12 uint64

	MOVOU	96(SI), X6
	MOVOU	112(SI), X7
	MOVOU	128(SI), X8
	MOVOU	144(SI), X9
	MOVOU	160(SI), X10
	MOVOU	176(SI), X11
	MOVOU	192(SI), X12
	MOVOU	208(SI), X13
	MOVOU	224(SI), X14
	MOVOU	240(SI), X15
	MOVOU	X0, (DI)
	MOVOU	X1, 16(DI)
	MOVOU	X2, 32(DI)
	MOVOU	X3, 48(DI)
	MOVOU	X4, 64(DI)
	MOVOU	X5, 80(DI)
	MOVOU	X6, 96(DI)
	MOVOU	X7, 112(DI)
	MOVOU	X8, 128(DI)
	MOVOU	X9, 144(DI)
	MOVOU	X10, 160(DI)

// license that can be found in the LICENSE file.

package codegen

var x [2]bool
var x8 [2]uint8
var x16 [2]uint16
var x32 [2]uint32
var x64 [2]uint64

func compMem1() int {
	// amd64:`CMPB\tcommand-line-arguments.x\+1\(SB\), [$]0`
	if x[1] {
		return 1
	}
	// amd64:`CMPB\tcommand-line-arguments.x8\+1\(SB\), [$]7`
	if x8[1] == 7 {
		return 1
	}
	// amd64:`CMPW\tcommand-line-arguments.x16\+2\(SB\), [$]7`

	MINU	X30, X5					// b3d2e20b
	ORN	X6, X7, X8				// 33e46340
	ORN	X6, X7					// b3e36340
	SEXTB	X16, X17				// 93184860
	SEXTH	X17, X18				// 13995860
	XNOR	X18, X19, X20				// 33ca2941
	XNOR	X18, X19				// b3c92941
	ZEXTH	X19, X20				// 3bca0908

	// 1.3: Bitwise Rotation (Zbb)
	ROL	X8, X9, X10				// 33958460 or b30f8040b3dff4013395840033e5af00
	ROL	X8, X9					// b3948460 or b30f8040b3dff401b3948400b3e49f00

	// Now on a scheduling stack (a pthread-created stack).
g0:
	// Save room for two of our pointers.
	SUB	$16, X2
	MOV	X9, 0(X2)	// save old g on stack
	MOV	(g_stack+stack_hi)(X9), X9
	SUB	X8, X9, X8
	MOV	X8, 8(X2)	// save depth in old g stack (can't just save SP, as stack might be copied during a callback)

	JALR	RA, (X5)

	// Restore g, stack pointer. X10 is return value.
	MOV	0(X2), g
	CALL	runtime·save_g(SB)

	MOVOU yout(16*0), X2
	MOVOU yout(16*1), X3
	MOVOU zout(16*0), X4
	MOVOU zout(16*1), X5

	MOVL BX, X6
	MOVL CX, X7

	PXOR X8, X8
	PCMPEQL X9, X9

	PSHUFD $0, X6, X6
	PSHUFD $0, X7, X7

	PCMPEQL X8, X6
	PCMPEQL X8, X7

	MOVOU X6, X15
	PANDN X9, X15

	MOVOU x1in(16*0), X9
	MOVOU x1in(16*1), X10
	MOVOU y1in(16*0), X11
	MOVOU y1in(16*1), X12

// Note: called with the C calling convention.
TEXT runtime·mstart_stub(SB),NOSPLIT,$200
	// X10 points to the m.
	// We are already on m's g0 stack.

	// Save callee-save registers (X8, X9, X18..X27, F8, F9, F18..F27)
	MOV	X8, (1*8)(X2)
	MOV	X9, (2*8)(X2)
	MOV	X18, (3*8)(X2)
	MOV	X19, (4*8)(X2)
	MOV	X20, (5*8)(X2)
	MOV	X21, (6*8)(X2)
	MOV	X22, (7*8)(X2)
	MOV	X23, (8*8)(X2)
	MOV	X24, (9*8)(X2)