by Andy Walsh

Last week was a milestone for the Tesla business. We announced that IBM has integrated our Tesla 20-series GPUs into its new iDataPlex servers. IBM released data that shows the big difference that Tesla GPUs make. Check out the data below, that shows a 4x or more increase in both single and double precision performance per rack for a little over 10% increase in power consumption:

IBM perf graph 
Data Courtesy of IBM

Really exciting times for the Tesla business. And for researchers worldwide. The global race for better science begins this summer as GPU-based hybrid systems replace traditional CPU clusters for large scale computational science. Dave Turek, vice president of Deep Computing at IBM explained their decision to embrace GPUs in an interview he gave HPCwire last week. The punchline is that this is all about meeting the growing customer demand:

"I think what's changed is that customers have been experimenting for a long time and now they're getting ready to buy. It wasn't the technology that drove us to do this. It was the maturation of the marketplace and the attitude toward using this technology. It's as simple as that."

Scott Denham, also from IBM’s Deep Computing group was kind enough to allow us to film a short interview with him about their adoption of Fermi-based Tesla GPUs and the impact on HPC. Enjoy.