T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21) {
  return internal::ValueArray21<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
      T12, T13, T14, T15, T16, T17, T18, T19, T20, T21>(v1, v2, v3, v4, v5, v6,
      v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21);
}

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

	LVX	(R4+off48),V19	// load next from buffer
	OR	$0,R2,R2

	VPMSUMD	V20,const1,V12	// vpmsumd with constants
	LVX	(R4+off64),V20	// load next from buffer
	OR	$0,R2,R2

	VPMSUMD	V21,const1,V13	// vpmsumd with constants
	LVX	(R4+off80),V21	// load next from buffer
	OR	$0,R2,R2

	VPMSUMD	V22,const1,V14	// vpmsumd with constants
	LVX	(R4+off96),V22	// load next from buffer
	OR	$0,R2,R2

	VSPLTISW $7, V30
	PCALIGN $16
loop_vsx:
	VADDUWM V0, V4, V0
	VADDUWM V1, V5, V1
	VADDUWM V2, V6, V2
	VADDUWM V3, V7, V3

	VPERMXOR V12, V0, V21, V12
	VPERMXOR V13, V1, V21, V13
	VPERMXOR V14, V2, V21, V14
	VPERMXOR V15, V3, V21, V15

	VADDUWM V8, V12, V8
	VADDUWM V9, V13, V9
	VADDUWM V10, V14, V10
	VADDUWM V11, V15, V11

	VXOR V4, V8, V4
	VXOR V5, V9, V5
	VXOR V6, V10, V6

	SHA512ROUND1(V5, V6, V7, V0, V1, V2, V3, V4, V19, V20, V21, V13, V18, R_x0b0)
	SHA512ROUND1(V4, V5, V6, V7, V0, V1, V2, V3, V20, V21, V22, V14, V19, R_x0c0)
	SHA512ROUND1(V3, V4, V5, V6, V7, V0, V1, V2, V21, V22, V23, V15, V20, R_x0d0)
	SHA512ROUND1(V2, V3, V4, V5, V6, V7, V0, V1, V22, V23, V8, V16, V21, R_x0e0)
	SHA512ROUND1(V1, V2, V3, V4, V5, V6, V7, V0, V23, V8, V9, V17, V22, R_x0f0)
	ADD	$0x100, TBL

	SHA256ROUND0(V5, V6, V7, V0, V1, V2, V3, V4, V19, R_x0d0)
	VPERMLE(V20, V20, LEMASK, V20)
	SHA256ROUND0(V4, V5, V6, V7, V0, V1, V2, V3, V20, R_x0e0)
	VSLDOI	$4, V20, V20, V21
	SHA256ROUND0(V3, V4, V5, V6, V7, V0, V1, V2, V21, R_x0f0)
	VSLDOI	$4, V21, V21, V22
	SHA256ROUND0(V2, V3, V4, V5, V6, V7, V0, V1, V22, R_x100)
	VSLDOI	$4, V22, V22, V23
	SHA256ROUND1(V1, V2, V3, V4, V5, V6, V7, V0, V23, V8, V9, V17, V22, R_x110)

      T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
      T18 v18, T19 v19, T20 v20, T21 v21) : 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) {}

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

	VREV64	V17.B16, V17.B16
	VREV64	V18.B16, V18.B16
	VREV64	V19.B16, V19.B16

	MOVD	R3, R4
	// load first 4 round consts in V20-V23
	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)

	VCIPHERLAST     V19, V23, V19; \
	VCIPHERLAST     V20, V23, V20; \
	VCIPHERLAST     V21, V23, V21; \
	VCIPHERLAST     V22, V23, V22; \
	XXLXOR          V1, V15, V1; \
	XXLXOR          V2, V16, V2; \
	XXLXOR          V3, V17, V3; \
	XXLXOR          V4, V18, V4; \
	XXLXOR          V5, V19, V5; \
	XXLXOR          V6, V20, V6; \
	XXLXOR          V7, V21, V7; \
	XXLXOR          V8, V22, V8

// Finish encryption on 4 streams and

	MOVD	p_len+32(FP), R3                           // message length
	VLD1	(R0), [V0.S4, V1.S4]                       // load h(a,b,c,d,e,f,g,h)
	VLD1.P	64(R2), [V16.S4, V17.S4, V18.S4, V19.S4]
	VLD1.P	64(R2), [V20.S4, V21.S4, V22.S4, V23.S4]
	VLD1.P	64(R2), [V24.S4, V25.S4, V26.S4, V27.S4]
	VLD1	(R2), [V28.S4, V29.S4, V30.S4, V31.S4]     //load 64*4bytes K constant(K0-K63)

blockloop:

#define Z1H    V5
#define X2L    V6
#define X2H    V7
#define Y2L    V8
#define Y2H    V9
#define Z2L    V10
#define Z2H    V11

#define ONE   V18
#define IDX   V19
#define SEL1  V20
#define SEL2  V21
TEXT ·p256Select(SB), NOSPLIT, $0
	MOVD   res+0(FP), P3ptr
	MOVD   table+8(FP), P1ptr
	VLREPB idx+(16+7)(FP), IDX
	VREPIB $1, ONE
	VREPIB $1, SEL2
	MOVD   $1, COUNT

	VZERO X1H
	VZERO X1L