VPGATHERDQ Y0, 664(X14*8), Y6           // c4a2fd9034f598020000
	VPGATHERQQ X2, (BP)(X7*2), X1           // c4e2e9914c7d00
	VPGATHERQQ Y2, (BP)(Y7*2), Y1           // c4e2ed914c7d00
	VPGATHERQQ X12, (R13)(X14*2), X11       // c40299915c7500
	VPGATHERQQ Y12, (R13)(Y14*2), Y11       // c4029d915c7500
	VPGATHERQQ X2, (BP)(X7*2), X1           // c4e2e9914c7d00
	VPGATHERQQ Y2, (BP)(Y7*2), Y1           // c4e2ed914c7d00

	{"$~15", "$-16"},
	{"(((8)&0xf)*4)(SP)", "32(SP)"},
	{"(((8-14)&0xf)*4)(SP)", "40(SP)"},
	{"(6+8)(AX)", "14(AX)"},
	{"(8*4)(BP)", "32(BP)"},
	{"(AX)", "(AX)"},
	{"(AX)(CX*8)", "(AX)(CX*8)"},
	{"(BP)(CX*4)", "(BP)(CX*4)"},
	{"(BP)(DX*4)", "(BP)(DX*4)"},
	{"(BP)(R8*4)", "(BP)(R8*4)"},
	{"(BX)", "(BX)"},
	{"(DI)", "(DI)"},
	{"(DI)(BX*1)", "(DI)(BX*1)"},
	{"(DX)", "(DX)"},
	{"(R9)", "(R9)"},

	VPGATHERQQ (BP)(X2*2), K1, X2   // ERROR "index and destination registers should be distinct"
	VPGATHERQQ (BP)(Y15*2), K1, Y15 // ERROR "index and destination registers should be distinct"
	VPGATHERQQ (BP)(Z20*2), K1, Z20 // ERROR "index and destination registers should be distinct"
	VPGATHERDQ (BP)(X2*2), K1, X2   // ERROR "index and destination registers should be distinct"

	for _, v := range s {
		if len(v) > maxInt-n {
			panic("bytes: Join output length overflow")
		}
		n += len(v)
	}

	b := bytealg.MakeNoZero(n)[:n:n]
	bp := copy(b, s[0])
	for _, v := range s[1:] {
		bp += copy(b[bp:], sep)
		bp += copy(b[bp:], v)
	}
	return b
}

// HasPrefix reports whether the byte slice s begins with prefix.
func HasPrefix(s, prefix []byte) bool {

</p>

<pre>
get_tls(CX)
MOVQ	g(CX), AX     // Move g into AX.
MOVQ	g_m(AX), BX   // Move g.m into BX.
</pre>

<p>
Register <code>BP</code> is callee-save.
The assembler automatically inserts <code>BP</code> save/restore when frame size is larger than zero.
Using <code>BP</code> as a general purpose register is allowed,
however it can interfere with sampling-based profiling.
</p>

<h3 id="arm">ARM</h3>

<p>

			"#define KEYROUND(xmm, load, off, r1, r2, index) \\",
			"\tMOVBLZX	(BP)(DX*4),	R8 \\",
			"\tload((off+1), r2) \\",
			"\tMOVB	R8,		(off*4)(R12) \\",
			"\tPINSRW	$index, (BP)(R8*4), xmm",
			"#define LOAD(off, reg) \\",
			"\tMOVBLZX	(off*4)(R12),	reg \\",
			"\tADDB	reg,		DX",
			"KEYROUND(X0, LOAD, 8, AX, BX, 0)",
		),