Performance Improvement of a VLIW ARchitecture without Pipeline-Stall during Instruction Cache Miss