Nvidia recently announced their latest push into the high performance computing arena with the second generation of their Tesla line of GPU computing platforms. This refresh of Tesla bumps compute density to 240 GPU cores with a theoretical peak of right around 1 Tflop. Alongside this, Nvidia will also release a new 1U server chassis to house the new units.
What does this mean for accelerated computing? Well, if you have already adopted Nvidia’s cuda programming environment into your application, you can now start grinning. The 1U Tesla server platform [named `S1070` as of this release] will pack 4-240 core graphics processors for a grand total of 960 cores. 4 Tflops of Tesla goodness in 1U at an estimated 700W.
For more info on Nvidia’s new Tesla gear, read the full article here.