RET

	PCALIGN	$16
cmp4:	// 4 - 7B
	CMP	R9,$4
	BLT	cmp2
	ANDCC	$3,R9,R9
	_LWBEX	(R0)(R5),R10
	_LWBEX	(R0)(R6),R11
	_LWBEX	(R9)(R5),R12
	_LWBEX	(R9)(R6),R14
	RLDIMI	$32,R10,$0,R12
	RLDIMI	$32,R11,$0,R14
	CMPU	R12,R14
	BR	cmp0

	PCALIGN $16
cmp2:	// 2 - 3B
	CMP	R9,$2
	BLT	cmp1
	ANDCC	$1,R9,R9
	_LHBEX	(R0)(R5),R10
	_LHBEX	(R0)(R6),R11
	_LHBEX	(R9)(R5),R12

	RET

cmp4:	// Length 4 - 7
	CMPU	R4,$4
	BLT	cmp2
	MOVD	$-4,R11
	ADD	$-4,R4,R4

	_LWBEX	(R0)(R3),R10
	_LWBEX	(R11)(R9),R11
	CMPB	R10,R5,R10
	CMPB	R11,R5,R11
	CNTLZW	R10,R10
	CNTLZW	R11,R11
	CMPU	R10,$32
	CMPU	R11,$32,CR1
	SRD	$3,R10,R3
	SRD	$3,R11,R11
	BNE	found

	ADD	R4,R11,R4
	MOVD	$-1,R3
	ISEL	CR1EQ,R3,R4,R3
	RET

cmp2:	// Length 2 - 3

	// s390x:`MOVHBR\t\(.*\), [R]`,`CMPW\t.*, [$]`
	return s == "xx"
}

func CompareString2(s string) bool {
	// amd64:`CMPL\t\(.*\), [$]`
	// arm64:`MOVWU\t\(.*\), [R]`,`CMPW\t.*, [R]`
	// ppc64le:`MOVWZ\t\(.*\), [R]`,`CMPW\t.*, [R]`
	// s390x:`MOVWBR\t\(.*\), [R]`,`CMPW\t.*, [$]`
	return s == "xxxx"
}

func CompareString3(s string) bool {
	// amd64:`CMPQ\t\(.*\), [A-Z]`
	// arm64:-`CMPW\t`

  // CHECK-DAG: [[CMP3:%.+]] = chlo.broadcast_compare %arg1, [[ZEROS2]] {comparison_direction = #chlo<comparison_direction LT>}
  // CHECK-DAG: [[CMP4:%.+]] = chlo.broadcast_compare [[CMP2]], [[CMP3]] {broadcast_dimensions = array<i64: 1>, comparison_direction = #chlo<comparison_direction NE>}
  // CHECK-DAG: [[AND:%.+]] = chlo.broadcast_and [[CMP1]], [[CMP4]]
  // CHECK-DAG: [[ONES:%.+]] = mhlo.constant dense<1>

			Goto("b1")),
		Bloc("b1",
			Valu("cmp1", OpLess64, c.config.Types.Bool, 0, nil, "arg1", "arg2"),
			If("cmp1", "b2", "b3")),
		Bloc("b2",
			Valu("cmp2", OpLess64, c.config.Types.Bool, 0, nil, "arg2", "arg3"),
			Goto("b3")),
		Bloc("b3",
			Valu("phi2", OpPhi, c.config.Types.Bool, 0, nil, "cmp1", "cmp2"),
			If("phi2", "b4", "b5")),
		Bloc("b4",
			Valu("cmp3", OpLess64, c.config.Types.Bool, 0, nil, "arg3", "arg1"),

func si1c(c <-chan int64) {
	// amd64:"CMPQ\t.+, [$]256"
	// s390x:"CLGIJ\t[$]12, R[0-9]+, [$]255"
	for x := <-c; x >= 0 && x < 256; x = <-c {
	}
}

func si2c(c <-chan int32) {
	// amd64:"CMPL\t.+, [$]256"
	// s390x:"CLIJ\t[$]12, R[0-9]+, [$]255"
	for x := <-c; x >= 0 && x < 256; x = <-c {
	}
}

func si3c(c <-chan int16) {
	// amd64:"CMPW\t.+, [$]256"
	// s390x:"CLIJ\t[$]12, R[0-9]+, [$]255"

var x32 [2]uint32
var x64 [2]uint64

func compMem1() int {
	// amd64:`CMPB\tcommand-line-arguments.x\+1\(SB\), [$]0`
	if x[1] {
		return 1
	}
	// amd64:`CMPB\tcommand-line-arguments.x8\+1\(SB\), [$]7`
	if x8[1] == 7 {
		return 1
	}
	// amd64:`CMPW\tcommand-line-arguments.x16\+2\(SB\), [$]7`
	if x16[1] == 7 {
		return 1
	}
	// amd64:`CMPL\tcommand-line-arguments.x32\+4\(SB\), [$]7`
	if x32[1] == 7 {
		return 1
	}

(CMP(Q|L|W|B)loadidx1 {sym} [off] ptr idx x mem) => (CMP(Q|L|W|B) (MOV(Q|L|W|B)loadidx1 {sym} [off] ptr idx mem) x)
(CMPQloadidx8 {sym} [off] ptr idx x mem) => (CMPQ (MOVQloadidx8 {sym} [off] ptr idx mem) x)
(CMPLloadidx4 {sym} [off] ptr idx x mem) => (CMPL (MOVLloadidx4 {sym} [off] ptr idx mem) x)
(CMPWloadidx2 {sym} [off] ptr idx x mem) => (CMPW (MOVWloadidx2 {sym} [off] ptr idx mem) x)

//   b in DI
//   count in BX
// Output:
//   result in AX
TEXT memeqbody<>(SB),NOSPLIT,$0-0
	CMPQ	BX, $8
	JB	small
	CMPQ	BX, $64
	JB	bigloop
#ifndef hasAVX2
	CMPB	internal∕cpu·X86+const_offsetX86HasAVX2(SB), $1
	JE	hugeloop_avx2

	// 64 bytes at a time using xmm registers
	PCALIGN $16
hugeloop:
	CMPQ	BX, $64
	JB	bigloop
	MOVOU	(SI), X0
	MOVOU	(DI), X1
	MOVOU	16(SI), X2
	MOVOU	16(DI), X3

	LDREXD	(R11), R12    // 9fcfbbe1
	STREX	R3, (R4), R5  // STREX  (R4), R3, R5 // 935f84e1
	STREXD	R8, (R9), g   // STREXD (R9), R8, g  // 98afa9e1

	CMPF    F8, F9        // c89ab4ee10faf1ee
	CMPD.CS F4, F5        // c45bb42e10faf12e
	CMPF.VS F7            // c07ab56e10faf16e
	CMPD    F6            // c06bb5ee10faf1ee

	MOVW	R4, F8        // 104b08ee
	MOVW	F4, R8        // 108b14ee