Multi-Threading Versus Code Parallelism

Multi-threading is about programming a specific number of threads to deal with specific operations. Parallel programming is detecting the number of processors and splitting the work of a particular operation across those known CPUs (processor times cores). Splitting work in applications across multiple threads does introduce a form of concurrency into applications; however, multi-threading doesn’t natively offer computational performance improvement (in most cases) and comes at a high price in code complexity when built using current programming paradigms.

The parallel programming goal is to make sure applications take advantage of multiple cores for improving computational performance and continue to improve performance as more cores are added over time. This doesn’t simply occur by making applications multi-threaded. In fact, badly implemented threading code can hurt performance. True code parallelism offers the same benefits as multi-threading offers, but in a way that aims to improve computational performance of applications, whilst safeguarding against common concurrency defects and issues.