VLD1.P	64(R4), [V20.D2, V21.D2, V22.D2, V23.D2]

	SHA512ROUND(V0, V1, V2, V3, V4, V20, V24, V12, V13, V19, V16, V17)
	SHA512ROUND(V3, V0, V4, V2, V1, V21, V25, V13, V14, V12, V17, V18)
	SHA512ROUND(V2, V3, V1, V4, V0, V22, V26, V14, V15, V13, V18, V19)
	SHA512ROUND(V4, V2, V0, V1, V3, V23, V27, V15, V16, V14, V19, V12)
	SHA512ROUND(V1, V4, V3, V0, V2, V24, V28, V16, V17, V15, V12, V13)

    T21 v21, T22 v22, T23 v23, T24 v24) {
  return internal::ValueArray24<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
      T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24>(v1, v2,
      v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18,
      v19, v20, v21, v22, v23, v24);
}

template <typename T1, typename T2, typename T3, typename T4, typename T5,

    T21 v21, T22 v22, T23 v23, T24 v24) {
  return internal::ValueArray24<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
      T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24>(v1, v2,
      v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18,
      v19, v20, v21, v22, v23, v24);
}

template <typename T1, typename T2, typename T3, typename T4, typename T5,

	VZIP2	V22.D2, V20.D2, V5.D2
	VZIP1	V23.D2, V21.D2, V9.D2
	VZIP2	V23.D2, V21.D2, V13.D2
	VLD1.P	64(R2), [V20.B16, V21.B16, V22.B16, V23.B16]
	VZIP1	V26.D2, V24.D2, V2.D2
	VZIP2	V26.D2, V24.D2, V6.D2
	VZIP1	V27.D2, V25.D2, V10.D2
	VZIP2	V27.D2, V25.D2, V14.D2
	VLD1.P	64(R2), [V24.B16, V25.B16, V26.B16, V27.B16]
	VZIP1	V30.D2, V28.D2, V3.D2
	VZIP2	V30.D2, V28.D2, V7.D2
	VZIP1	V31.D2, V29.D2, V11.D2
	VZIP2	V31.D2, V29.D2, V15.D2

	VPDI  $0x4, V22, V22, V22 // flip the doublewords to big-endian order
	VPDI  $0x4, V23, V23, V23 // flip the doublewords to big-endian order
	VPDI  $0x4, V24, V24, V24 // flip the doublewords to big-endian order
	VSTM  V17, V24, 0(R7)     // 128-bytes into z
	ADD   $128, R7
	ADD   $128, R10           // i += 16
	SUB   $16, R3             // n -= 16
	BGE   UU1                 // if n >= 0 goto U1

	VSTEH	$7, V31, (R2)           // e7f020007809
	VSTEB	$15, V29, 4094(R12)     // e7d0cffef808
	VMSLG	V21, V22, V23, V24      // e78563007fb8
	VMSLEG  V21, V22, V23, V24      // e78563807fb8
	VMSLOG  V21, V22, V23, V24      // e78563407fb8
	VMSLEOG V21, V22, V23, V24      // e78563c07fb8
	VSUMGH	V1, V2, V3              // e73120001065
	VSUMGF	V16, V17, V18           // e72010002e65

	VEOR V14.B16, V14.B16, V14.B16
	VEOR V15.B16, V15.B16, V15.B16

	// Save seed state for adding back later.
	VMOV V4.B16, V20.B16
	VMOV V5.B16, V21.B16
	VMOV V6.B16, V22.B16
	VMOV V7.B16, V23.B16
	VMOV V8.B16, V24.B16
	VMOV V9.B16, V25.B16
	VMOV V10.B16, V26.B16
	VMOV V11.B16, V27.B16

	// 4 iterations. Each iteration is 8 quarter-rounds.
	MOVD $4, R0
loop:
	QR(V0, V4, V8, V12)
	QR(V1, V5, V9, V13)

      T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24) : v1_(v1),
      v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
      v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
      v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21),
      v22_(v22), v23_(v23), v24_(v24) {}

  template <typename T>
  operator ParamGenerator<T>() const {

      T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24) : v1_(v1),
      v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
      v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
      v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21),
      v22_(v22), v23_(v23), v24_(v24) {}

  template <typename T>
  operator ParamGenerator<T>() const {

	WFI                                        // 7f2003d5
	YIELD                                      // 3f2003d5
	//TODO FABD F0, F5, F11                    // abd4a07e
	//TODO VFABD V30.S2, V8.S2, V24.S2         // 18d5be2e
	//TODO VFABS V5.S4, V24.S4                 // b8f8a04e
	FABSS F2, F28                              // 5cc0201e
	FABSD F0, F14                              // 0ec0601e
	//TODO FACGE F25, F16, F0                  // 00ee797e