Since the LILI-128 cipher is a clock-controlled keystream generator, the speed of the keystream data is degraded in a clock-synchronized hardware logic design. Basically, the clock-controlled $LFSR_d$ in the LILI-128 cipher requires a system clock that is 1 ~4 times higher. Therefore, if the same clock is selected, the system throughput of the data rate will be lowered. Accordingly, this paper proposes a 4-bit parallel $LFSR_d$, where each register bit includes four variable data routines for feed feedback of shifting within the $LFSR_d$ . Furthermore, the timing of the propose design is simulated using a $Max^+$plus II from the ALTERA Co., the logic circuit is implemented for an FPGA device (EPF10K20RC240-3), and the throughput stability is analyzed up to a late of 50 Mbps with a 50MHz system clock. (That is higher than the 73 late at 45 Mbps, plus the maximum delay routine in the proposed design was below 20ns.) Finally, we translate/simulate our FPGA/VHDL design to the Lucent ASIC device( LV160C, 0.13 $mu extrm{m}$ CMOS & 1.5v technology), and it could achieve a throughput of about 500 Mbps with a 0.13$mu extrm{m}$ semiconductor for the maximum path delay below 1.8ns.