This paper provides a fast and efficient approach to analyze and compare systems implemented with through-silicon via (TSV) and monolithic inter-tier via (MIV) 3D integration technologies based on compact models for cycle-per-instruction, memory throughput, and multi-level interconnect networks. Additionally, the impact of via diameter and capacitance on the overall system throughput has been quantified. It is demonstrated that for the same die area and thermal constraint, an MIV-based processor offers over 25% improvement in computational throughput as compared with its 2D counterpart.