Parallelizing Programs for the Multicore Era

Modern processors are very advanced but still suffer the bottleneck of slow memory access speeds. Processors now have the ability to run multiple tasks in parallel. This means that the processor runs multiple tasks concurrently, rather than sequentially, thus reducing the effect of memory access times. Utilizing this ability is called parallelization. There are many standards in place that allow programmers to implement this technique in their programs. This can improve the speed of the program, as well as reduce memory usage. The techniques of parallelization, also called multithreading, are investigated in the context of matrix multiplication, an important mathematical operation. Also studied are the methods by which a processor stores and retrieves memory to the cache, which exists within the processor itself, and methods by which cache usage can be optimized. Parallelization was found to have a significant improvement on program running time, achieving up to 5 times speed boosts in certain cases.