Big Data - Hardware - Graphical Processing Unit

The graphical processing unit (GPU) has gotten considerable publicity as an unused computing resource that could reduce the run times of data mining and other analytical problems by parallelizing the computations. The GPU is already found in every desktop computer in the world. In the early 2000s, GPUs got into the computing game. Graphics processing has evolved considerably from early text-only displays of the first desktop computers. This quest for better graphics has been driven by industry needs for visualization tools. One example is engineers using three-dimensional (3D) computer-aided design (CAD) software to create prototypes of new designs prior to ever building them. An even bigger driver of GPU computing has been the consumer video game industry, which has seen price and performance trends similar to the rest of the consumer computing industry. The relentless drive to higher performance at lower cost has given the average user unheard-of performance both on the CPU and the GPU. Three-dimensional graphics processing must process millions or billions of 3D triangles in 3D scenes multiple times per second to create animation. Placing and coloring all of these triangles in their 3D environment requires a huge number of very similar calculations. Initially, 3D graphics were done using a fixed rendering pipeline, which took the 3D scene information and turned it into pixels that could be presented to the user in a video or on the screen. This fixed pipeline was implemented in hardware, with various parts of the GPU doing different pieces of the problem of turning triangles into pixels. In the early 2000s, this fixed pipeline was gradually replaced by generalized software shaders, which were miniprograms that performed the operations of the earlier fixed hardware pipeline. With these shaders, high-performance computing folks noticed that the floating-point coordinates and colors could look an awful lot like physics or chemistry problems if you looked at them just right. The more hardcore hacker types started creating graphics problems that looked a lot like nonsense except that the underlying calculations being done solved hard problems remarkably fast. The performance gains got noticed, and computing frameworks, which used the GPUs for doing nongraphics calculations, were developed. These calculations are the same type needed for data mining. GPUs are a green field. Historically the ability to develop code to run on the GPU was restrictive and costly. Those programming interfaces for developing software that takes advantage of GPUs have improved greatly in the last few years. Software has only started to take advantage of the GPU, and it will be several years before the computations needed for data mining are efficiently delegated to the GPU for execution. When that time comes, the speedup in many types of data mining problems will be reduced from hours to minutes and from minutes to seconds.