Stunning growth in the performance of graphics processors along with stiffer competition is prompting developers of the technology to look beyond the traditional drive for photo realism in search of broader applications in parallel computing.
Designers say they want to apply massively parallel graphics operations to new computing architectures. The result, according to at least one technologist at this year’s International Electron Devices Meeting here, could be a CPU-GPU co-processing, or “heterogeneous” computing, that could turn PCs into supercomputers. John Chen, vice president of technology and foundry operations at Nvidia Corp., argued here Monday (Dec. 7) that such multithreaded, high-throughput computers could help, for instance, in decentralizing medical imaging, a step that could bring doctor and patient closer.
