BYTE    $0xc4; BYTE $0x42; BYTE $0x7d; BYTE $0x5a; BYTE $0x70; BYTE $0x10 // broadcasti128 16(r8), ymm14
	BYTE    $0xc4; BYTE $0x42; BYTE $0x7d; BYTE $0x5a; BYTE $0x60; BYTE $0x20 // broadcasti128 32(r8), ymm12
	BYTE    $0xc4; BYTE $0xc2; BYTE $0x7d; BYTE $0x5a; BYTE $0x60; BYTE $0x30 // broadcasti128 48(r8), ymm4
	VPADDD  ·avx2InitMask<>(SB), DD0, DD0

	// Special optimization, for very short buffers
	CMPQ inl, $192

// D (double)    = 64 bit float
// S (single)    = 32 bit float

// copied from ../../amd64/reg.go
var regNamesAMD64 = []string{
	"AX",
	"CX",
	"DX",
	"BX",
	"SP",
	"BP",
	"SI",
	"DI",
	"R8",
	"R9",
	"R10",
	"R11",
	"R12",
	"R13",
	"g", // a.k.a. R14
	"R15",
	"X0",
	"X1",
	"X2",
	"X3",
	"X4",
	"X5",
	"X6",
	"X7",
	"X8",
	"X9",
	"X10",
	"X11",
	"X12",

type R0 struct {
	*R1
	*R2
	*R3
	*R4
}

type R1 struct {
	*R5
	*R6
	*R7
	*R8
}

type R2 R1
type R3 R1
type R4 R1

type R5 struct {
	*R9
	*R10
	*R11
	*R12
}

type R6 R5
type R7 R5
type R8 R5

type R9 struct {
	*R13
	*R14
	*R15
	*R16
}

type R10 R9
type R11 R9