dataTLS:
	ADD	$14*16, pTbl
	VLD1.P	(pTbl), [T1.B16, T2.B16]
	VEOR	B0.B16, B0.B16, B0.B16

	MOVD	(aut), H0
	VMOV	H0, B0.D[0]
	MOVW	8(aut), H0
	VMOV	H0, B0.S[2]
	MOVB	12(aut), H0
	VMOV	H0, B0.B[12]

	MOVD	$0, autLen
	B	dataMul

octetsLoop:
		CMP	$128, autLen
		BLT	startSinglesLoop
		SUB	$128, autLen

		VLD1.P	32(aut), [B0.B16, B1.B16]

		VLD1.P	32(pTbl), [T1.B16, T2.B16]

	VESLG  $2, T_3, T_4  \
	VAG    T_3, T_4, T_4 \
	VAG    T_2, h2, h2   \
	VAG    T_4, h0, h0   \
	VESRLG $26, h2, T_0  \
	VESRLG $26, h0, T_1  \
	VN     MOD26, h2, h2 \
	VN     MOD26, h0, h0 \
	VAG    T_0, h3, h3   \
	VAG    T_1, h1, h1   \
	VESRLG $26, h3, T_2  \
	VN     MOD26, h3, h3 \
	VAG    T_2, h4, h4

// EXPAND splits the 128-bit little-endian values in0 and in1

		} else {
			b[u] -= delta
		}
	}
	return zig, nil
}

func (d *decoder) reconstructProgressiveImage() error {
	// The h0, mxx, by and bx variables have the same meaning as in the
	// processSOS method.
	h0 := d.comp[0].h
	mxx := (d.width + 8*h0 - 1) / (8 * h0)
	for i := 0; i < d.nComp; i++ {
		if d.progCoeffs[i] == nil {
			continue
		}
		v := 8 * d.comp[0].v / d.comp[i].v

	if len(p) > 0 && C._goboringcrypto_SHA1_Update(h.noescapeCtx(), unsafe.Pointer(&*addr(p)), C.size_t(len(p))) == 0 {
		panic("boringcrypto: SHA1_Update failed")
	}
	return len(p), nil
}

func (h0 *sha1Hash) sum(dst []byte) []byte {
	h := *h0 // make copy so future Write+Sum is valid
	if C._goboringcrypto_SHA1_Final((*C.uint8_t)(noescape(unsafe.Pointer(&h.out[0]))), h.noescapeCtx()) == 0 {
		panic("boringcrypto: SHA1_Final failed")
	}

//
// a = H0
// b = H1
// c = H2
// d = H3
// e = H4
// f = H5
// g = H6
// h = H7
//
// for t = 0 to 63 {
//    T1 = h + BIGSIGMA1(e) + Ch(e,f,g) + Kt + Wt
//    T2 = BIGSIGMA0(a) + Maj(a,b,c)
//    h = g
//    g = f
//    f = e
//    e = d + T1
//    d = c
//    c = b
//    b = a
//    a = T1 + T2
// }
//
// H0 = a + H0
// H1 = b + H1
// H2 = c + H2

//
// a = H0
// b = H1
// c = H2
// d = H3
// e = H4
// f = H5
// g = H6
// h = H7
//
// for t = 0 to 63 {
//    T1 = h + BIGSIGMA1(e) + Ch(e,f,g) + Kt + Wt
//    T2 = BIGSIGMA0(a) + Maj(a,b,c)
//    h = g
//    g = f
//    f = e
//    e = d + T1
//    d = c
//    c = b
//    b = a
//    a = T1 + T2
// }
//
// H0 = a + H0
// H1 = b + H1
// H2 = c + H2

//
// a = H0
// b = H1
// c = H2
// d = H3
// e = H4
// f = H5
// g = H6
// h = H7
//
// for t = 0 to 79 {
//    T1 = h + BIGSIGMA1(e) + Ch(e,f,g) + Kt + Wt
//    T2 = BIGSIGMA0(a) + Maj(a,b,c)
//    h = g
//    g = f
//    f = e
//    e = d + T1
//    d = c
//    c = b
//    b = a
//    a = T1 + T2
// }
//
// H0 = a + H0
// H1 = b + H1
// H2 = c + H2

//
// a = H0
// b = H1
// c = H2
// d = H3
// e = H4
// f = H5
// g = H6
// h = H7
//
// for t = 0 to 63 {
//    T1 = h + BIGSIGMA1(e) + Ch(e,f,g) + Kt + Wt
//    T2 = BIGSIGMA0(a) + Maj(a,b,c)
//    h = g
//    g = f
//    f = e
//    e = d + T1
//    d = c
//    c = b
//    b = a
//    a = T1 + T2
// }
//
// H0 = a + H0
// H1 = b + H1
// H2 = c + H2

  const char task1_name[] = "/job:localhost/replica:0/task:1/device:CPU:0";
  const char task2_name[] = "/job:localhost/replica:0/task:2/device:CPU:0";

  // Create one variable per task.
  TFE_TensorHandle* h0 = TestVariable(ctx, 1.0, task1_name);
  TFE_TensorHandle* h1 = TestVariable(ctx, 2.0, task2_name);
  TFE_TensorHandle* h2 = TestVariable(ctx, 3.0, task0_name);

		}
		if rno <= uint16(WZR) {
			op = "MOVWU" + suffix
		} else if rno >= uint16(B0) && rno <= uint16(B31) {
			op = "FMOVB" + suffix
			args[0] = fmt.Sprintf("F%d", rno&31)
		} else if rno >= uint16(H0) && rno <= uint16(H31) {
			op = "FMOVH" + suffix
			args[0] = fmt.Sprintf("F%d", rno&31)
		} else if rno >= uint16(S0) && rno <= uint16(S31) {
			op = "FMOVS" + suffix
			args[0] = fmt.Sprintf("F%d", rno&31)