default:
				return intelReg[a]
			}
		}
	}
	return strings.ToLower(arg.String())
}

var intelOp = map[Op]string{
	JAE:       "jnb",
	JA:        "jnbe",
	JGE:       "jnl",
	JNE:       "jnz",
	JG:        "jnle",
	JE:        "jz",
	SETAE:     "setnb",
	SETA:      "setnbe",
	SETGE:     "setnl",
	SETNE:     "setnz",
	SETG:      "setnle",
	SETE:      "setz",
	CMOVAE:    "cmovnb",

	instructions["JNAE"] = x86.AJCS /* alternate */
	instructions["JNB"] = x86.AJCC  /* alternate */
	instructions["JNBE"] = x86.AJHI /* alternate */
	instructions["JNC"] = x86.AJCC  /* alternate */
	instructions["JNE"] = x86.AJNE  /* not equal (ZF = 0) */
	instructions["JNG"] = x86.AJLE  /* alternate */
	instructions["JNGE"] = x86.AJLT /* alternate */
	instructions["JNL"] = x86.AJGE  /* alternate */

	MOVL	DX, BP
	SUBL	BX, DX // DX = blen-alen
	JLE	2(PC)
	MOVL	BX, BP // BP = min(alen, blen)
	CMPL	SI, DI
	JEQ	allsame
	CMPL	BP, $4
	JB	small
#ifdef GO386_softfloat
	JMP	mediumloop
#endif
largeloop:
	CMPL	BP, $16
	JB	mediumloop
	MOVOU	(SI), X0
	MOVOU	(DI), X1
	PCMPEQB X0, X1
	PMOVMSKB X1, BX
	XORL	$0xffff, BX	// convert EQ to NE
	JNE	diff16	// branch if at least one byte is not equal
	ADDL	$16, SI

	// return fr * 2**exponent
ldexp:
	ADDL    $0x3FF, BX // add bias
	JLE     denormal
	CMPL    BX, $0x7FF
	JGE     overflow
lastStep:
	SHLQ    $52, BX
	MOVQ    BX, X1
	MULSD   X1, X0
	MOVSD   X0, ret+8(FP)
	RET
notFinite:
	// test bits for -Inf
	MOVQ    $NegInf, AX
	CMPQ    AX, BX
	JNE     notNegInf
	// -Inf, return 0
underflow: // return 0
	MOVQ    $0, ret+8(FP)
	RET

	// Directly return DX, we don't need to accumulate
	// since we have <16 bytes.
	POPCNTL DX, DX
	MOVQ	DX, (R8)
	RET

avx2:
#ifndef hasAVX2
	CMPB   internal∕cpu·X86+const_offsetX86HasAVX2(SB), $1
	JNE sse
#endif
	MOVD AX, X0
	LEAQ -64(SI)(BX*1), R11
	LEAQ (SI)(BX*1), R13
	VPBROADCASTB  X0, Y1
	PCALIGN $32
avx2_loop:
	VMOVDQU (DI), Y2
	VMOVDQU 32(DI), Y4
	VPCMPEQB Y1, Y2, Y3

	MOVOU	(SI), X0
	MOVOU	(DI), X1
	PCMPEQB X0, X1
	PMOVMSKB X1, AX
	XORQ	$0xffff, AX
	JNE	diff16

	MOVOU	16(SI), X0
	MOVOU	16(DI), X1
	PCMPEQB X0, X1
	PMOVMSKB X1, AX
	XORQ	$0xffff, AX
	JNE	diff32

	MOVOU	32(SI), X0
	MOVOU	32(DI), X1
	PCMPEQB X0, X1
	PMOVMSKB X1, AX
	XORQ	$0xffff, AX
	JNE	diff48

	MOVOU	48(SI), X0
	MOVOU	48(DI), X1
	PCMPEQB X0, X1
	PMOVMSKB X1, AX

	MOVL	trap+0(FP), AX	// syscall entry
	// copy args down
	LEAL	a1+4(FP), SI
	LEAL	sysargs-144(SP), DI
	CLD
	MOVSL
	MOVSL
	MOVSL
	INT	$64
	MOVL	AX, r1+16(FP)
	MOVL	$0, r2+20(FP)
	CMPL	AX, $-1
	JNE	ok3

	LEAL	errbuf-128(SP), AX
	MOVL	AX, sysargs-144(SP)
	MOVL	$128, sysargs1-140(SP)
	MOVL	$SYS_ERRSTR, AX
	INT	$64
	CALL	runtime·exitsyscall(SB)
	MOVL	sysargs-144(SP), AX
	MOVL	AX, errbuf-148(SP)

	MOVL	(0*4)(BX), AX		// fn
	CALL	AX

	MOVL	AX, (4*4)(BX)		// r1
	MOVL	DX, (5*4)(BX)		// r2

	// Standard libc functions return -1 on error and set errno.
	CMPL	AX, $-1
	JNE	ok
	CMPL	DX, $-1
	JNE	ok

	// Get error code from libc.
	CALL	libc_errno(SB)
	MOVL	(AX), AX
	MOVW	AX, (6*4)(BX)		// err

ok:
	MOVL	$0, AX			// no error (it's ignored anyway)
	MOVL	BP, SP
	POPL	BP

	// test bits for special cases
	MOVQ    p+0(FP), BX
	MOVQ    $~(1<<63), AX
	ANDQ    AX, BX // p = |p|
	MOVQ    q+8(FP), CX
	ANDQ    AX, CX // q = |q|
	MOVQ    $PosInf, AX
	CMPQ    AX, BX
	JLE     isInfOrNaN
	CMPQ    AX, CX
	JLE     isInfOrNaN
	// hypot = max * sqrt(1 + (min/max)**2)
	MOVQ    BX, X0
	MOVQ    CX, X1
	ORQ     CX, BX
	JEQ     isZero
	MOVAPD  X0, X2
	MAXSD   X1, X0
	MINSD   X2, X1
	DIVSD   X0, X1

func ptrEqual() {
	// amd64:-"JEQ",-"JNE"
	// ppc64x:-"BEQ",-"BNE"
	// s390x:-"BEQ",-"BNE"
	copy(x[:], x[:])
}

func ptrOneOffset() {
	// amd64:-"JEQ",-"JNE"
	// ppc64x:-"BEQ",-"BNE"
	// s390x:-"BEQ",-"BNE"
	copy(x[1:], x[:])
}

func ptrBothOffset() {
	// amd64:-"JEQ",-"JNE"
	// ppc64x:-"BEQ",-"BNE"
	// s390x:-"BEQ",-"BNE"
	copy(x[1:], x[2:])
}