This refers to the internal data processing pathways of a CPU. The CPU processes data through different stages, like Fetch, Decode, Execute, and Store. By Pipelining, it becomes possible to perform these stages independently and on different bits of data. This allows for more efficient use of CPU cycles.
Modern CPUs have increased the number of stages in the pipeline, and also use advanced techniques like branch prediction, which is used to estimate the possible outcome of an if/else condition, to ensure that every stage of the pipeline is utilised. The performance of the CPU is greatly influenced by the way it implements pipelining.