A dichotomy exists among the various approaches to achieving performance in the workstation and server markets. Some implementations aggressively pursue high clock rate goals, often by settling for less instruction-level parallelism. In contrast, IBM workstations have been well-known for supporting high degrees of concurrency. POWER2 (Performance Optimized With Enhanced RISC) based RISC System/6000 workstations further widen the gap between the instruction-level parallelism and clock rate approaches. This paper describes the architecture and performance aspects of these new systems. >