TEXT ·testBGE(SB),NOSPLIT,$0-17
	MOV	a+0(FP), X5
	MOV	b+8(FP), X6
	MOV	$1, X7
	BGE	X5, X6, b
	MOV	$0, X7
b:
	MOV	X7, r+16(FP)
	RET

// func testBGEU(a, b int64) (r bool)
TEXT ·testBGEU(SB),NOSPLIT,$0-17
	MOV	a+0(FP), X5
	MOV	b+8(FP), X6
	MOV	$1, X7
	BGEU	X5, X6, b
	MOV	$0, X7
b:
	MOV	X7, r+16(FP)
	RET

// func testBGEZ(a int64) (r bool)
TEXT ·testBGEZ(SB),NOSPLIT,$0-9

	MULHU	X5, X6, X7				// b3335302
	MULHSU	X5, X6, X7				// b3235302
	MULW	X5, X6, X7				// bb035302
	DIV	X5, X6, X7				// b3435302
	DIVU	X5, X6, X7				// b3535302
	REM	X5, X6, X7				// b3635302
	REMU	X5, X6, X7				// b3735302
	DIVW	X5, X6, X7				// bb435302
	DIVUW	X5, X6, X7				// bb535302
	REMW	X5, X6, X7				// bb635302
	REMUW	X5, X6, X7				// bb735302

	// 8.2: Load-Reserved/Store-Conditional

	VGF2P8MULB X0, X7, K7, X7                          // 62f2450fcff8
	VGF2P8MULB 15(R8), X7, K7, X7                      // 62d2450fcfb80f000000
	VGF2P8MULB (BP), X7, K7, X7                        // 62f2450fcf7d00
	VGF2P8MULB X15, X9, K7, X7                         // 62d2350fcfff
	VGF2P8MULB X12, X9, K7, X7                         // 62d2350fcffc
	VGF2P8MULB X0, X9, K7, X7                          // 62f2350fcff8

	VPMADD52LUQ X3, X7, K1, X7                         // 62f2c509b4fb
	VPMADD52LUQ 15(R8), X7, K1, X7                     // 62d2c509b4b80f000000
	VPMADD52LUQ (BP), X7, K1, X7                       // 62f2c509b47d00
	VPMADD52LUQ X5, X14, K1, X7                        // 62f28d09b4fd
	VPMADD52LUQ X31, X14, K1, X7                       // 62928d09b4ff
	VPMADD52LUQ X3, X14, K1, X7                        // 62f28d09b4fb

//     Maj(x, y, z) = (x AND y) XOR (x AND z) XOR (y AND z)
#define SHA512T2(a, b, c) \
	ROR	$28, a, X6; \
	ROR	$34, a, X7; \
	XOR	X7, X6; \
	ROR	$39, a, X8; \
	XOR	X8, X6; \
	AND	a, b, X7; \
	AND	a, c, X8; \
	XOR	X8, X7; \
	AND	b, c, X9; \
	XOR	X9, X7; \
	ADD	X7, X6

// Calculate T1 and T2, then e = d + T1 and a = T1 + T2.
// The values for e and a are stored in d and h, ready for rotation.

TEXT ·archMax(SB),NOSPLIT,$0
	MOVD	x+0(FP), F0
	MOVD	y+8(FP), F1
	FCLASSD	F0, X5
	FCLASSD	F1, X6

	// +Inf special cases
	MOV	$PosInf, X7
	BEQ	X7, X5, isMaxX
	BEQ	X7, X6, isMaxY

	// NaN special cases
	MOV	$NaN, X7
	BEQ	X7, X5, isMaxX
	BEQ	X7, X6, isMaxY

	// normal case
	FMAXD	F0, F1, F0
	MOVD	F0, ret+16(FP)
	RET

isMaxX: // return x
	MOVD	F0, ret+16(FP)
	RET

		x4 := s[1]
		x5 := s[7]
		x6 := s[5]
		x7 := s[3]

		// Stage 1.
		x8 := w7 * (x4 + x5)
		x4 = x8 + w1mw7*x4
		x5 = x8 - w1pw7*x5
		x8 = w3 * (x6 + x7)
		x6 = x8 - w3mw5*x6
		x7 = x8 - w3pw5*x7

		// Stage 2.
		x8 = x0 + x1
		x0 -= x1
		x1 = w6 * (x3 + x2)
		x2 = x1 - w2pw6*x2
		x3 = x1 + w2mw6*x3
		x1 = x4 + x6
		x4 -= x6
		x6 = x5 + x7
		x5 -= x7

		// Stage 3.

	VPSHLDVD X0, X7, K7, X7                            // 62f2450f71f8
	VPSHLDVD 7(AX), X7, K7, X7                         // 62f2450f71b807000000
	VPSHLDVD (DI), X7, K7, X7                          // 62f2450f713f
	VPSHLDVD X15, X9, K7, X7                           // 62d2350f71ff
	VPSHLDVD X12, X9, K7, X7                           // 62d2350f71fc
	VPSHLDVD X0, X9, K7, X7                            // 62f2350f71f8

	VPMULTISHIFTQB X14, X20, K5, X7                    // 62d2dd0583fe
	VPMULTISHIFTQB 17(SP)(BP*1), X20, K5, X7           // 62f2dd0583bc2c11000000
	VPMULTISHIFTQB -7(CX)(DX*8), X20, K5, X7           // 62f2dd0583bcd1f9ffffff
	VPMULTISHIFTQB X9, X7, K5, X7                      // 62d2c50d83f9
	VPMULTISHIFTQB X7, X7, K5, X7                      // 62f2c50d83ff

	VPOPCNTW (DI), K3, Z16                             // 62e2fd4b5407
	VPSHUFBITQMB X24, X7, K6, K0                       // 6292450e8fc0
	VPSHUFBITQMB X7, X7, K6, K0                        // 62f2450e8fc7
	VPSHUFBITQMB X0, X7, K6, K0                        // 62f2450e8fc0
	VPSHUFBITQMB (R8), X7, K6, K0                      // 62d2450e8f00
	VPSHUFBITQMB 15(DX)(BX*2), X7, K6, K0              // 62f2450e8f845a0f000000