Speeding Up Computers

For over thirty years until 2004-5 the speed of processors followed the Moore’s Law and was exponentially increasing. Approximately every 18 months processors speed doubled. This rapid and steady progress was very economically satisfying. It was enough to wait several months for new processor generation without the need for redeveloping software. However shortly after the electronic technology hit energy consumption and heat dissipation walls in the middle of the new century first decade we witnessed the appearance of multi-core processors. Initially with two or four cores. Later up to eight. The growth of processor speed continued but the technology behind the increased performance was parallelism not the higher clock frequency. A more dramatic change took place when NVIDIA and AMD designed and manufactured highly parallel processing devices with many cores. Initially they were used mainly for graphics applications and called Graphics Processing Units (GPU). Later they have become useful for general purpose applications (GPGPU) beyond graphics and visualization [Kirk2010, Sanders2011].