14 DAY TRIAL // At this year's Fusion Developer Summit, AMD has released additional information about the architecture used for the next generation graphics cards based on the Southern Islands core, which were designed in order to deliver improved GPGPU performance.
These details were divulged by Eric Demers, AMD Graphics chief technical officer, during a keynote presentation held during FDS 2011.
While the AMD representative didn't mention any specific products, he did state that all these changes will be introduced inside all of the company's GPU products (stand alone or integrated) over the next few years.
According to Demers, the most important modification that will be brought by the next-gen graphics architecture will be the support for C, C++, and other high-level programming languages.
In order for this functionality to be implemented, AMD had to redesign the processing units found inside the GPU to transform them into "scalar coprocessors" similar to those found in the vector supercomputers of the 1980s.
This means that the new processing units will have to include elements found inside multiple instruction, multiple data (MIMD); multiple instruction, single data; and simultaneous multi-threading (SMT) designs, claims Tech Report.
In addition to this change, AMD's next generation graphics architecture will also feature support for x86-64 memory addressing, which , according to Demers, should eliminate the "glitching" users experiment when textures have to be loaded from the system's memory.
More details about the modifications that AMD wants to introduce in its upcoming graphics cores are available in the presentation slides included in this article.
Yesterday, Rick Bergman, senior vice president and general manager of AMD products group, suggested that AMD's Southern Islands GPUs will arrive later this year.
“In a couple of days you are going to hear about our exciting new graphics architecture that will be coming out later this year and will be utilized by our future APUs."
