Is it worthwhile?

• Same type of analysis for either CPI contributed by cache, or time to resolve a miss, or average memory access time

• For example computation of average memory access time

  Av. mem. access time = hit rate in L1 * hit time in L1 +
  
  miss rate in L1 * miss penalty

• Obviously if hit time in L1 is same for direct-mapped and 2-way no advantage for pseudo.

Previous slide

Next slide

Back to first slide

View graphic version