The semiconductor industry is changing, and the computer graphics industry is changing along with it. Gone are the days when processor speeds doubled every year and software programs enjoyed the benefit of improved application performance for free.
The dominant computing paradigm is shifting from sequential to parallel processing. In the past, single-threaded applications enjoyed the benefit of ever increasing processor clock speeds. But this is no longer true. To harness the full potential of new hardware, software needs to be parallel.
This is why we have developed Meridian rendering software. Parallel to its core, Meridian is a massively concurrent architecture designed to harness the full theoretical capacity of computers with hundreds of processor cores. Based on a new patented rendering method that uses interval arithmetic, Meridian analytically renders geometry like NURBS directly into perfectly anti-aliased pixels. Since there is no longer any need to tesselate geometry into dense polygon meshses of tiny pixel-sized polygons, each tile in an image can be processed using only a very small and constant amount of memory regardless of the scene size and complexity.
This makes the Meridian rendering method highly parallel and an ideal software configuration for modern many-core computers. As dozens or hundreds of processor cores are added to a rendering machine, users can expect that Meridian will keep the cores busy even on the most demanding scenes.
