2. Parallel Programming for Multicore and Multi-CPU Machines

Moores’s Law states CPUs double in frequency every two years. This was possible by increasing the transistor density. As time has gone on this has proven to be more difficult because with more transistors more heat is produced. As this problem became more apparent Computer Architects turned to another solution, a cheaper yet effective solution. Which was the implementation of multiple cores within a CPU. IBM was the first to release a Multi-core CPU back in 2001. Intel’s attempt to release processors of this caliber came in 2002 with the Intel Pentium 4. This processor didn’t implement parallel computing more or less just hyperthreading. This allowed it to switch between programs really fast so it gave the appearance of doing multiple processes at once. Since then more and more computers have been implementing multicore CPUs, some even have multiple CPUs. Because of this computer scientists needed to find a way to utilize these extra cores, thus was the birth of parallel computing.

In this book, we will be discussing multicore and multi CPU machines and how they work, how to use them, and how to optimize your code to use them effectively.