One area where AMD has been able to hold the upper hand on is the business of promoting their graphics processing units as general-purpose processors. First was the Folding@Home program last September. More recently, during the R600 launch, they spent a lot of time looking at applications for their GPU other than just graphics and added features in the GPU to aid future GPGPU projects.

Maybe not having to compete with Intel made them feel rosy, but now NVIDIA is announcing their progression into the world of high-performance computing (HPC): Tesla.

HPC is not your average piece of software, coding for multi-threaded processors is much more difficult than serial processing, which means that only tasks that are multi-threaded in nature find the bottleneck in hardware. NVIDIA's Tesla wants to be the answer for those problems. HPC applications require much more processing power because they access immense amounts of data simultaneously; for expamle: simulation of neural circuits, computational microscopes, seismic and reservoir simulation, cell phone design... There are other applications for HCP in researching electromagnetics, energy, biomedical, pharmaceutical, industrial, financial and military operations.

What does it have to do with you? Not much at the moment, but this is the beginning of a transition from clock speed to multi-core power. Consumer level programs will gradually take advantage of more than a couple of cores, and prices for solutions like Tesla will go down (right now they range from $1500 for a GPU card to $12,000 if you want a S870 server).

Link: PCPerspective.

TrackBack URI for this entry

Comments (0)

Subscribe to this comment's feed

Show/Hide comments

Show/Hide comment form