IBM, Nvidia draw $325 million US Energy Department supercomputer contract

Shawn Knight

Posts: 15,292   +192
Staff member

ibm nividia energy dept nvidia supercomputers volta project coral fastforward 2 nvidia volta ibm power family

IBM and Nvidia have entered into a deal with the Department of Energy to build two huge supercomputers capable of performing calculations five to seven times faster than the most advanced systems currently in use in the US according to a report from CNET.

The $325 million contract will see one of the machines installed at Oak Ridge National Laboratory in Tennessee for civilian use. The other supercomputer will be used for nuclear weapons simulation at Lawrence Livermore National Laboratory in California.

ibm nividia energy dept nvidia supercomputers volta project coral fastforward 2 nvidia volta ibm power family

IBM is tasked with building the overall systems that will mate future components from the company’s own Power family with Volta accelerators from Nvidia. IBM has a storied history in the supercomputer market but graphics chip maker Nvidia is considered a newcomer.

The publication notes that each machine will be able to crank out more than 100 petaflops – a quadrillion calculations per second – at full speed. To put that into perspective, you’d need roughly 3 million modern-day notebooks to achieve similar performance.

The Department of Energy is also spending an extra $100 million on a project called FastForward 2. Said program is expected to make next generation supercomputers 20 to 40 times faster than existing high-end models. Collectively, this is all part of a project by the name of Coral.

The companies plan to begin installation in 2017, we’re told.

Permalink to story.

 
IBM as outsource company worst to get hired as contractor for them they pay less. What do they do well good question since they use to be no. 1 for outsourcing next to HP and ESD which is now HP.
 
What kind of problems require 1 quadrillion calculation per second. I'm sure there are some, I just don't know.
 
Back