Executive Director and nVidia CEO Jen-Sun Huang (Jen-Hsun Huang) argues that performance GPU computing will grow to 570 times over the next six years (I wonder why exactly 570 and not, for example, 600?), With this for the same time the performance of processors will increase only once in three. Huang also said that the high performance chip cards will stimulate the emergence of various applications, such as interactive ray tracing, CGI modeling, energy exploration and other applications that emulate the real world.

Huang noted that the CPU and graphics chip that could potentially merge into a single entity, but in the coming years, the phenomenon becomes widespread, primarily because of problems with software development. Huang said that in developing the final product involved the three main groups of developers, including programmers, OEM / ODM-designers and creators of the chips. Since they are independent of each other, the interaction between them is difficult, making it difficult to integrate new and rapidly developing architectures and development in the final device. On the other hand, due to this separation of functions provided by their fairly rapid development in their own speed, as well as flexibility and variability of end products.

Not long ago, nVidia announced that it intends in 2015 to produce chips for video on the 11nm process technology that each chip will be up to 5000 universal processors, and that the GPU will be able to provide the computing power in 20Tflop. Very optimistic projections, which somehow want to believe.