As I’ve noted before, when you’re not playing action games, the killer GPU in your PC is basically a case heater. For the most part, it uselessly sucks power and radiates heat as you perform mundane computing tasks: web browsing, word processing, spreadsheet calculations, MP3 playback. GPUs are the most underutilized resource in PCs.

Finally, that’s changing. AMD now bundles its ATI Stream parallel-processing software in the latest ATI Catalyst graphics drivers. As users download and install these free drivers, they automatically prep their systems to run ATI Stream programs that leverage the GPU as a massively parallel processor. Before, users had to download ATI Stream separately. AMD is following Nvidia, which began bundling its CUDA parallel-processing software with display drivers in 2007.

Although ATI Stream and CUDA are for programmers, anyone can use the application software written for these platforms. When you install and run an ATI Stream or CUDA application, it automatically executes on the x86 CPU and on the GPU, which does the heavy lifting. Most people won’t notice anything different—except better performance.

Bundling these platforms with drivers helps solve the classic chicken-or-egg problem. Software developers hesitate to write programs for platforms lacking a large installed base, and users hesitate to adopt new platforms for which little software is available. AMD says its new drivers potentially expand the ATI Stream installed base to about two million PCs. Nvidia says 107 million systems can run CUDA, although the actual installed base is smaller.

Developers are heeding the call. The ATI Avivo Video Converter, free from AMD, transcodes digital video among several different formats. A transcoding job that requires three hours, 23 minutes on an Intel Core 2 Duo processor at 3.0GHz takes just 12 minutes with Avivo. Avivo is a preview of a new commercial product called PowerDirector 7 from CyberLink. For Nvidia GPUs, Elemental Technologies has a similar video transcoder called Badaboom. More apps are coming or are already here.

Parallel processing has been around since the dawn of computing but typically has been limited to expensive, specialized systems. Some experts doubted the technology would ever go mainstream. Nvidia and AMD are proving them wrong—using the processor you already have.

Tom Halfhill was formerly a senior editor for Byte magazine and is now an analyst for Microprocessor Report.