VPDPWSSDS 17(SP)(BP*8), X15, K2, X30               // 6262050a53b4ec11000000
	VPDPWSSDS 17(SP)(BP*4), X15, K2, X30               // 6262050a53b4ac11000000
	VPDPWSSDS X13, X30, K2, X30                        // 62420d0253f5
	VPDPWSSDS X6, X30, K2, X30                         // 62620d0253f6
	VPDPWSSDS X12, X30, K2, X30                        // 62420d0253f4
	VPDPWSSDS 17(SP)(BP*8), X30, K2, X30               // 62620d0253b4ec11000000

	V4FMADDSS 7(AX), [X0-X3], K5, X30                  // 62627f0d9bb007000000 or 62627f2d9bb007000000 or 62627f4d9bb007000000
	V4FMADDSS (DI), [X0-X3], K5, X30                   // 62627f0d9b37 or 62627f2d9b37 or 62627f4d9b37
	V4FMADDSS 7(AX), [X10-X13], K5, X30                // 62622f0d9bb007000000 or 62622f2d9bb007000000 or 62622f4d9bb007000000
	V4FMADDSS (DI), [X10-X13], K5, X30                 // 62622f0d9b37 or 62622f2d9b37 or 62622f4d9b37

// func addMulVVW1024(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW1024(SB),$0-32
	MOV	$16, X30
	JMP	addMulVVWx(SB)

// func addMulVVW1536(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW1536(SB),$0-32
	MOV	$24, X30
	JMP	addMulVVWx(SB)

// func addMulVVW2048(z, x *uint, y uint) (c uint)
TEXT ·addMulVVW2048(SB),$0-32
	MOV	$32, X30
	JMP	addMulVVWx(SB)

TEXT addMulVVWx(SB),NOFRAME|NOSPLIT,$0
	MOV	z+0(FP), X5

	VRCP28SS X6, X15, K2, X30                          // 6262050acbf6
	VRCP28SS X12, X15, K2, X30                         // 6242050acbf4
	VRCP28SS X13, X30, K2, X30                         // 62420d02cbf5
	VRCP28SS X6, X30, K2, X30                          // 62620d02cbf6
	VRCP28SS X12, X30, K2, X30                         // 62420d02cbf4
	VRCP28SS X13, X11, K2, X3                          // 62d2250acbdd

// is computed with correct precision.
func truncate() {
	const (
		x30 = 1 << 30
		x60 = 1 << 60

		staticF32 = float32(x30) + 1 - x30
		staticF64 = float64(x60) + 1 - x60
		staticC64 = complex64(x30) + 1 - x30
		staticC128 = complex128(x60) + 1 - x60
	)
	dynamicF32 := float32(x30)
	dynamicF32 += 1
	dynamicF32 -= x30

	dynamicF64 := float64(x60)
	dynamicF64 += 1
	dynamicF64 -= x60

	VPERMB X0, X2, K1, X30                             // 62626d098df0
	VPERMB 7(SI)(DI*4), X2, K1, X30                    // 62626d098db4be07000000
	VPERMB -7(DI)(R8*2), X2, K1, X30                   // 62226d098db447f9ffffff
	VPERMB X26, X9, K1, X30                            // 620235098df2
	VPERMB X19, X9, K1, X30                            // 622235098df3

package main

func main() {
	var x00, x01, x02, x03, x04, x05, x06, x07, x08, x09 int
	var x10, x11, x12, x13, x14, x15, x16, x17, x18, x19 int
	var x20, x21, x22, x23, x24, x25, x26, x27, x28, x29 int
	var x30, x31, x32 int

	_ = x00
	_ = x01
	_ = x02
	_ = x03
	_ = x04
	_ = x05
	_ = x06
	_ = x07
	_ = x08
	_ = x09

	_ = x10
	_ = x11
	_ = x12
	_ = x13
	_ = x14
	_ = x15
	_ = x16

	//	_111111 = _111 + _111000
	//	x12     = _111111 << 6 + _111111
	//	x24     = x12 << 12 + x12
	//	x30     = x24 << 6 + _111111
	//	x31     = 2*x30 + 1
	//	x32     = 2*x31 + 1
	//	x63     = x32 << 31 + x31
	//	x126    = x63 << 63 + x63
	//	x252    = x126 << 126 + x126
	//	x255    = x252 << 3 + _111
	//	i397    = ((x255 << 33 + x32) << 94 + x30) << 2
	//	return    1 + i397
	//

	var z = new(P384Element).Set(e)

	VAESENC X14, X16, X30                              // 62427d00dcf6 or 6242fd00dcf6
	VAESENC X19, X16, X30                              // 62227d00dcf3 or 6222fd00dcf3
	VAESENC X8, X16, X30                               // 62427d00dcf0 or 6242fd00dcf0
	VAESENC 99(R15)(R15*8), X16, X30                   // 62027d00dcb4ff63000000 or 6202fd00dcb4ff63000000

 * Calling into the gc tool chain, where all registers are caller save.
 * Called from standard ARM EABI, where x19-x29 are callee-save, so they
 * must be saved explicitly, along with x30 (LR).
 */
.globl EXT(crosscall1)
EXT(crosscall1):
	.cfi_startproc
	stp x29, x30, [sp, #-96]!
	.cfi_def_cfa_offset 96
	.cfi_offset 29, -96
	.cfi_offset 30, -88
	mov x29, sp
	.cfi_def_cfa_register 29
	stp x19, x20, [sp, #80]
	.cfi_offset 19, -16