IBM Decks Out iDataPlex Hyperscale Servers with Nvidia GPUs

Paul Lilly

Taking aim at organizations interested in running scientific and commercial applications on graphics processors, IBM this week updated its iDataPlex servers to now includes GPUs from Nvidia. According to David Turek, vice president of deep computing at IBM's Systems and Technology Group, Nvidia's M2050 GPUs will be faster at executing certain tasks than traditional CPUs.

Some possible clients could include scientific labs run by governments or oil and gas exploration companies where it's common for software to be developed from the ground up, Turek explained. In doing so, these types of organizations can optimize their programs to run faster on GPUs.

Back in 2008, IBM's iDataPlex servers were marketed as a foundation for cloud computing, says Charles King, principal analyst at Pund-IT. By adding GPUs to the mix, IBM's iDataPlex line sets itself even further apart from the competition.

"This is evidence that IBM is seeing that architecture as being more flexible than a foundation for cloud computing," King said.

So why go with Nvidia over ATI? The expanded use of CUDA, says Turek, who also pointed out the increasing popularity of OpenCL, which is backed by industry heavyweights AMD, Apple, Intel, and Nvidia.

Image Credit: IBM

Around the web

by CPMStar (Sponsored) Free to play