SHA512ROUND1(23, 0x76f988da831153b5, R9, R10, R11, R12, R13, R14, R15, R8)
	SHA512ROUND1(24, 0x983e5152ee66dfab, R8, R9, R10, R11, R12, R13, R14, R15)
	SHA512ROUND1(25, 0xa831c66d2db43210, R15, R8, R9, R10, R11, R12, R13, R14)
	SHA512ROUND1(26, 0xb00327c898fb213f, R14, R15, R8, R9, R10, R11, R12, R13)
	SHA512ROUND1(27, 0xbf597fc7beef0ee4, R13, R14, R15, R8, R9, R10, R11, R12)

	AMANDV		R14, (R13), R12 // acb96238
	AMORW		R14, (R13), R12 // ac396338
	AMORV		R14, (R13), R12 // acb96338
	AMXORW		R14, (R13), R12 // ac396438
	AMXORV		R14, (R13), R12 // acb96438
	AMMAXW		R14, (R13), R12 // ac396538
	AMMAXV		R14, (R13), R12 // acb96538
	AMMINW		R14, (R13), R12 // ac396638
	AMMINV		R14, (R13), R12 // acb96638
	AMMAXWU		R14, (R13), R12 // ac396738
	AMMAXVU		R14, (R13), R12 // acb96738

	MOVQ   R12, R11
	SHRQ   $0x33, R11
	MOVQ   R14, R13
	SHRQ   $0x33, R13
	ANDQ   AX, SI
	IMUL3Q $0x13, R13, R13
	ADDQ   R13, SI
	ANDQ   AX, R8
	ADDQ   BX, R8
	ANDQ   AX, R10
	ADDQ   DI, R10
	ANDQ   AX, R12
	ADDQ   R9, R12
	ANDQ   AX, R14
	ADDQ   R11, R14

	// Store output
	MOVQ out+0(FP), AX
	MOVQ SI, (AX)
	MOVQ R8, 8(AX)
	MOVQ R10, 16(AX)
	MOVQ R12, 24(AX)

	MOVW	R4, 12(R13)
	MOVW	R5, 16(R13)
	MOVW	R6, 20(R13)
	MOVW	R7, 24(R13)
	MOVW	R8, 28(R13)
	MOVW	g, 32(R13)
	MOVW	R11, 36(R13)

	// Skip floating point registers on goarmsoftfp != 0.
	MOVB    runtime·goarmsoftfp(SB), R11
	CMP	$0, R11
	BNE     skipfpsave
	MOVD	F8, (40+8*0)(R13)
	MOVD	F9, (40+8*1)(R13)
	MOVD	F10, (40+8*2)(R13)
	MOVD	F11, (40+8*3)(R13)
	MOVD	F12, (40+8*4)(R13)

				{1, 49151}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R15
			},
		},
	},
	{
		name:   "CMPW",
		argLen: 2,
		asm:    x86.ACMPW,
		reg: regInfo{
			inputs: []inputInfo{
				{0, 49151}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R15
				{1, 49151}, // AX CX DX BX SP BP SI DI R8 R9 R10 R11 R12 R13 R15
			},
		},
	},
	{

	SHA256ROUND0(5, 0x59f111f1, R11, R12, R13, R14, R15, R8, R9, R10)
	SHA256ROUND0(6, 0x923f82a4, R10, R11, R12, R13, R14, R15, R8, R9)
	SHA256ROUND0(7, 0xab1c5ed5, R9, R10, R11, R12, R13, R14, R15, R8)
	SHA256ROUND0(8, 0xd807aa98, R8, R9, R10, R11, R12, R13, R14, R15)
	SHA256ROUND0(9, 0x12835b01, R15, R8, R9, R10, R11, R12, R13, R14)
	SHA256ROUND0(10, 0x243185be, R14, R15, R8, R9, R10, R11, R12, R13)

	AND $0x7ffffffffffff, R1, R11
	AND $0x7ffffffffffff, R2, R12
	AND $0x7ffffffffffff, R3, R13
	AND $0x7ffffffffffff, R4, R14

	ADD R0>>51, R11, R11
	ADD R1>>51, R12, R12
	ADD R2>>51, R13, R13
	ADD R3>>51, R14, R14
	// R4>>51 * 19 + R10 -> R10
	LSR $51, R4, R21
	MOVD $19, R22
	MADD R22, R10, R21, R10

	STP (R10, R11), 0(R20)
	STP (R12, R13), 16(R20)
	MOVD R14, 32(R20)

	MOVD	0(R1), R3 // argc
	ADD	$8, R1, R4 // argv

	// Did the TLS pointer get set? If so, don't change it (e.g musl).
	CMP	R13, $0
	BNE	tls_and_argcv_in_reg

	MOVD	$runtime·m0+m_tls(SB), R13 // TLS
	ADD	$0x7000, R13

tls_and_argcv_in_reg:
	BR	main(SB)

TEXT main(SB),NOSPLIT,$-8
	MOVD	$runtime·rt0_go(SB), R12
	MOVD	R12, CTR

	MOVW	$1, R13
	TBZ	$1, R8, ks192
	TBNZ	$2, R8, ks256
	LDPW	(R9), (R4, R5)
	LDPW	8(R9), (R6, R7)
	STPW.P	(R4, R5), 8(R10)
	STPW.P	(R6, R7), 8(R10)
	MOVW	$0x1b, R14
ks128Loop:
		VMOV	R7, V2.S[0]
		WORD	$0x4E030042       // TBL V3.B16, [V2.B16], V2.B16
		AESE	V0.B16, V2.B16    // Use AES to compute the SBOX
		EORW	R13, R4
		LSLW	$1, R13           // Compute next Rcon

	MOVV	(g_sched+gobuf_sp)(g), R3

	// Now on a scheduling stack (a pthread-created stack).
g0:
	// Save room for two of our pointers.
	ADDV	$-16, R3
	MOVV	R13, 0(R3)	// save old g on stack
	MOVV	(g_stack+stack_hi)(R13), R13
	SUBVU	R12, R13
	MOVV	R13, 8(R3)	// save depth in old g stack (can't just save SP, as stack might be copied during a callback)
	JAL	(R25)

	// Restore g, stack pointer. R4 is return value.