10. Review

We have discussed the standard Von Neumann architecture and its variant the Harvard architecture.

Memory efficiency is an important feature in the form of one or more caches.

Then we discussed the use of parallel processing in the form of the pipeline and superpipeline. Another popular technique is superscalar which tries to do further parallel processing by dispatching instructions based on what kind they are.

All this effort is made to improve CPU performance.

As real examples, below are two published technical descriptions of CPUs that include the features we have discussed.

AMD Phenom II

Runs at 3.3 GHz as default, it has a 64KB Level 1 cache as well as 512KB Level 2 per core as well as a shared Level 3 6MB cache. Operates at 125 W thermal design power. Uses dual channel DDR3 memory. Uses 45nm gate technology. Uses deep pipelines and superscalar architecture.

Intel Gulftown (Core i7-980X)

Has 6 processing cores running at 3.3GHz with 32KB Level 1 cache, 256KB Level 2 cache per core and a shared 12MB shared cache. Uses 32nm gates and runs at 130 Watts. Uses triple channel DDR3 memory. Uses deep pipelines and superscalar architecture.