|
According to associate NVIDIA made an important declaration about the planned increase and the development of cluster segment for super-calculations on base Of tesla. NVIDIA and Mental images start new Reality server version 3.0. Similar products existed earlier and they enjoyed a specific demand, in the third version this will be for the first time the mixed platform, which uses hardware and program advantages.
Computer for calculations collects large revolutions. It is natural that video processors are wonderfully suitable for similar tasks. For example, Lucas film release picture, where the most rapid processors for rendering one sequence require 23 hours. Accordingly, for compiling 90 minute picture 354 years will be required :-) , and here thousand graphic core NVIDIA processors manage the same task in all for several weeks.
But here there are not many people, which intend to render three-dimensional scenes for motion picture . It is expected that in 2010, there will be the 4G networks, and respectively the rates of Internet- connections, will be claimed by mobile users , home or office. Thus, using NVIDIA calculation monster, it is possible to send the necessary information from the telephone or notebook, and then to accept the conversely finished object. Or this can occur directly on Web site, where the user will select the necessary components .

It was shown during the demonstration the possibilities of iray and Reality server 3.0, as different effects were superimposed on the scene of photo-realistic office, and rendering passed in real Moreover, moreover delay did not exceed 10-15 seconds. Rolf Kherken noted that the fastest working stations will manage such tasks approximately in 200 minutes.
Reality server 3.0 - is the combination from servers, based on NVIDIA tesla and program component. Rays tracing is now accelerated by video processors.

Platform is based on Tesla s1070 1U compact servers, the basic difference - in the total quantity of video processors (GPU), and in accordance with this, the possibility of simultaneous work with a number of users. Related Products :
|