Memory latency

This is achieved by using a first-level sum-addressed memory data cache [6]. Mixing the memory address adder with the word line decoder for the data cache largely eliminates the address adder's latency. This gives us a linear memory latency improvement.