GPU Computing In China

In Beijing this week to attend Nvidia’s GPU Technology Conference, I took the opportunity to visit one of HP’s largest SL390s customers in China. With computer rooms spread across three floors of a large data center, I passed by row after row of HP racks, many filled with GPU-powered SL390s servers.

While this customer also had many racks of systems from local manufacturers such as Dawning and Levono, as far as I could tell the only GPU powered systems where HP’s SL390s. Designed from the ground up for GPUs, the SL390s incorporates Nvidia’s M2090, M2070, or M2070Q GPUs directly into the system chassis, no external GPU boxes, cables, or connectors required. In addition, every SL390s comes standard with 10G Ethernet and InfiniBand standard, courtesy of an integrated Mellanox Connect-X2 networking chip.

I was particularly impressed by the quality of the network cabling throughout this customer’s data center, especially the HP racks. Below is a close up of half a rack of SL390s 2U servers, each capable of supporting three Nvidia GPUs in addition to two Intel x86 CPUs.

In the picture below, you can see this customer is taking advantage of the SL390s’ Infiniband networking with the Voltaire (now part of Mellanox) IB switch.

With it’s front-cabled design, the rear of the SL390s racks are particularly clean, ensuring full air flow from the fans and allowing the best operating environment possible for the servers.

Of course, after a busy data center tour, I was especially grateful to my host for serving a delicious traditional Chinese lunch!

With it’s many computationally intensive industries including life sciences, oil & gas, manufacturing, and media & entertainment, it is encouraging to see so many Chinese companies quickly adopting GPU computing. The resulting power savings compared to doing the same work on non-accelerated systems is important in China as well as the rest of the world.

For those who couldn’t attend the GPU Technology Conference in person, I would highly recommend watching Nvidia’s CEO and co-founder Jen-Hsun Huang’s keynote. Jen-Hsun spent over 90 minutes on stage giving a wonderful keynote with dozens of examples of how GPU Computing has evolved from the first what would now be considered primitive GPUs of the mid 1990’s to modern day GPUs enabling everything from photo-realistic graphics to the world’s most powerful supercomputers on to future exascale systems which are expected to be possible before the end of the decade.


About Marc Hamilton

Marc Hamilton – Vice President, Solutions Architecture and Engineering, NVIDIA. At NVIDIA, the Visual Computing Company, Marc leads the worldwide Solutions Architecture and Engineering team, responsible for working with NVIDIA’s customers and partners to deliver the world’s best end to end solutions for professional visualization and design, high performance computing, and big data analytics. Prior to NVIDIA, Marc worked in the Hyperscale Business Unit within HP’s Enterprise Group where he led the HPC team for the Americas region. Marc spent 16 years at Sun Microsystems in HPC and other sales and marketing executive management roles. Marc also worked at TRW developing HPC applications for the US aerospace and defense industry. He has published a number of technical articles and is the author of the book, “Software Development, Building Reliable Systems”. Marc holds a BS degree in Math and Computer Science from UCLA, an MS degree in Electrical Engineering from USC, and is a graduate of the UCLA Executive Management program.
This entry was posted in Uncategorized. Bookmark the permalink.