// schedule 64 input dwords, by doing 12 rounds of 4 each
	MOVQ $4, frame_SRND(SP)

loop1:
	VPADDQ  (BP), Y4, Y0
	VMOVDQU Y0, frame_YFER(SP)

	MY_VPALIGNR(Y0, Y7, Y6, 8)

	VPADDQ Y4, Y0, Y0

	MY_VPALIGNR(Y1, Y5, Y4, 8)

	VPSRLQ $1, Y1, Y2
	VPSLLQ $(64-1), Y1, Y3
	VPOR   Y2, Y3, Y3

	VPSRLQ $7, Y1, Y8

	MOVQ  AX, DI
	RORXQ $41, DX, R13
	RORXQ $18, DX, R14