VADDUDM	V2, V5, V5	// The count will be fixed up afterwards.
	ADD	$32, R3
	BDNZ	cmploop

	VADDUDM	V4, V5, V5
	MFVSRD	V5, R18
	VSLDOI	$8, V5, V5, V5
	MFVSRD	V5, R21
	ADD	R21, R18, R18
	ANDCC	$31, R4, R4
	// Skip the tail processing if no bytes remaining.
	BEQ	tail_0

#ifdef GOPPC64_power10
	SRD	$3, R18, R18	// Fix the vector loop count before counting the tail on P10.

tail:	// Count the last 0 - 31 bytes.

	// Truncate not needed, load does sign/zero extend

	// ppc64x:-"MOVBZ\tR\\d+,\\sR\\d+"
	val16[0] = uint16(*u8)

	// AND not needed due to size
	// ppc64x:-"ANDCC"
	sval16[1] = 255 & int16(x8+y8)

	// ppc64x:-"ANDCC"
	val16[1] = 255 & uint16(*u8+z8)

}
func shiftidx(u8 *uint8, x16 *int16, u16 *uint16) {

	// ppc64x:-"MOVBZ\tR\\d+,\\sR\\d+"
	val16[0] = uint16(sval16[*u8>>2])

	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

func signextendAndMask8to64(a int8) (s, z uint64) {
	// ppc64x: "MOVB", "ANDCC\t[$]1015,"
	s = uint64(a) & 0x3F7
	// ppc64x: -"MOVB", "ANDCC\t[$]247,"
	z = uint64(uint8(a)) & 0x3F7
	return
}

// Verify zero-extended values are not sign-extended under a bit mask (#61297)
func zeroextendAndMask8to64(a int8, b int16) (x, y uint64) {
	// ppc64x: -"MOVB\t", -"ANDCC", "MOVBZ"
	x = uint64(a) & 0xFF

	ADDIS $1000, R3, R4             // 3c8303e8

	ANDCC $1, R3                    // 70630001
	ANDCC $1, R3, R4                // 70640001
	ANDCC $-1, R4                   // 3be0ffff7fe42039
	ANDCC $-1, R4, R5               // 3be0ffff7fe52039
	ANDCC $65535, R5                // 70a5ffff
	ANDCC $65535, R5, R6            // 70a6ffff
	ANDCC $65536, R6                // 74c60001
	ANDCC $65536, R6, R7            // 74c70001

	BLE	CR6, small	// <= 8
	BLT	CR7, xor32	// Case for 32 ≤ n < 64 bytes

	// Case for n ≥ 64 bytes
preloop64:
	SRD	$6, R6, R7	// Set up loop counter
	MOVD	R7, CTR
	MOVD	$16, R10
	MOVD	$32, R14
	MOVD	$48, R15
	ANDCC	$63, R6, R9	// Check for tailing bytes for later
	PCALIGN $16
	// Case for >= 64 bytes
	// Process 64 bytes per iteration
	// Load 4 vectors of a and b
	// XOR the corresponding vectors

	// ppc64x/power8:"FCMP","ISEL",-"SETBC\tCR0LT"
	b := !(x >= y)
	return b
}
func TestLogicalCompareZero(x *[64]uint64) {
	// ppc64x:"ANDCC",^"AND"
	b := x[0]&3
	if b!=0 {
		x[0] = b
	}
	// ppc64x:"ANDCC",^"AND"
	b = x[1]&x[2]
	if b!=0 {
		x[1] = b
	}
	// ppc64x:"ANDNCC",^"ANDN"
	b = x[1]&^x[2]
	if b!=0 {
		x[1] = b
	}
	// ppc64x:"ORCC",^"OR"

	"ADDCV",
	"ADDCVCC",
	"ADDME",
	"ADDMECC",
	"ADDMEVCC",
	"ADDMEV",
	"ADDE",
	"ADDECC",
	"ADDEVCC",
	"ADDEV",
	"ADDZE",
	"ADDZECC",
	"ADDZEVCC",
	"ADDZEV",
	"ADDEX",
	"AND",
	"ANDCC",
	"ANDN",
	"ANDNCC",
	"ANDISCC",
	"BC",
	"BCL",
	"BEQ",
	"BGE",
	"BGT",
	"BLE",
	"BLT",
	"BNE",
	"BVC",
	"BVS",
	"BDNZ",
	"BDZ",
	"CMP",
	"CMPU",
	"CMPEQB",
	"CNTLZW",

}

func CmpLogicalToZero(a, b, c uint32, d, e uint64) uint64 {

	// ppc64x:"ANDCC",-"CMPW"
	// wasm:"I64Eqz",-"I32Eqz",-"I64ExtendI32U",-"I32WrapI64"
	if a&63 == 0 {
		return 1
	}

	// ppc64x:"ANDCC",-"CMP"
	// wasm:"I64Eqz",-"I32Eqz",-"I64ExtendI32U",-"I32WrapI64"
	if d&255 == 0 {
		return 1
	}

	// ppc64x:"ANDCC",-"CMP"
	// wasm:"I64Eqz",-"I32Eqz",-"I64ExtendI32U",-"I32WrapI64"
	if d&e == 0 {

	// arm:"AND\t[$]63",-".*udiv",-"SRA"
	// arm64:"TST\t[$]63",-"UDIV",-"ASR",-"AND"
	// ppc64x:"ANDCC",-"RLDICL",-"SRAD",-"CMP"
	a := n1%64 == 0 // signed divisible

	// 386:"TESTL\t[$]63",-"DIVL",-"SHRL"
	// amd64:"TESTQ\t[$]63",-"DIVQ",-"SHRQ"
	// arm:"AND\t[$]63",-".*udiv",-"SRA"
	// arm64:"TST\t[$]63",-"UDIV",-"ASR",-"AND"
	// ppc64x:"ANDCC",-"RLDICL",-"SRAD",-"CMP"
	b := n2%64 != 0 // signed indivisible

	return a, b
}