by Jennifer R
The graphics processing unit is a big subject in the video game industry and among video game players. High-end GPUs have great processing power to handle the more complex calculations required for graphic-heavy games. The authors suggest harnessing this processing power for database servers. They point out that “a GPU can deliver hundreds of billions of operations per second—some GPUs more than a teraflop, or a trillion operations per second—while requiring only slightly more electrical power and cooling than a CPU. For the same levels of power and cooling, a GPU can deliver 20 to 30 times as much total computational power. That’s a lot less power per calculation”. Why use a graphics processing unit rather than add more computer processing units (CPU)? The authors say for certain CPUs, you can get a GPU for an equivalent price that has several times the CPU’s computational power. There is also the difference in their architecture. CPUs are serial in design, meaning they perform instructions in sequence, whereas GPUs are parallel, being able to handle multiple instructions at once. This is ideal for databases, where a massive amount data can be requested and accessed simultaneously.
With the costs of GPUs decreasing as their processing capabilities increase, I think this is a viable, cost-efficient way to allocate more processing powers for databases. With GPUs, you would be storing data caches on the video memory. As I understand it, accessing data from main memory is much faster than trying to access it from the hard drive. For additional computational speed, this is a good thing. However, I would be concerned about the volatile nature of main memory. As my brother described it, if something happens to cause a power outage, anything stored on main memory goes poof! Secondary storage mediums such as the hard disk retain stored information, so long as you were not writing to the disk when the power outage occurred.
Source: Blas, A.D. & Kaldewey, T (2009). Data Monster: Why Graphics Processors Will Transform Database Processing . IEEE Spectrum. Retrieved April 29, 2012 from http://spectrum.ieee.org/computing/software/data-monster