Speeding up execution by overlapping the phases of individual instructions (fetching, decoding, executing) within a single processor.
Applying the same operation to a large set of data elements simultaneously, a technique common in graphics processing and multimedia.
Modern computing has reached a crossroads where single-processor speeds are nearing their physical limits. To continue advancing, the field has pivoted toward , focusing on three pillars: parallelism , scalability , and programmability . 1. The Power of Parallelism
Parallelism is the simultaneous execution of multiple tasks to reduce overall computation time. It is no longer a luxury for supercomputers but a necessity for everything from mobile phones to cloud servers. Architects exploit parallelism at several distinct levels: