Guest blog written by Ed Turkel, HP, Manager, Product and Segment Marketing Hyperscale Business Unit
The insatiable demand for performance in High- Performance Computing (HPC) has some experts musing that the industry might see its first exascale system later this decade. Getting there will involve much more than scaling up existing technologies. It will come with enormous challenges and requirements across manageability, efficiency and RAS.
HP’s Project Voyager presents a significant opportunity to enable ever larger HPC systems gain performance by adding brains to the mix. The intelligence is due to HP ProActive Insight Architecture and is built into every one of the new HP ProLiant Gen8 servers maximizing performance and RAS with minimal effort.
When managing 1,000’s or 10,000’s of servers the benefits of this intelligence is significant. The new levels of automation and intelligence work across a variety of vectors to solve critical HPC challenges including unrelenting demands to increase performance, energy and space efficiencies and increased time to innovation to name a few. To continue to drive greater performance that accelerates HPC innovation automating critical system management tasks is required for system administration productivity. If you want to minimize the number of servers needed to reach your desired performance level – and who doesn’t -- each server needs to be purpose-built for performance and scale. And finally, you need to maximize the energy and space efficiency of systems at scale -- that is, more HPC performance in a given power, cooling and space footprint.
The Purdue University Carter cluster provides great insight into the advantages of HP ProLiant Gen8 systems for HPC. It was recently ranked 54th on the latest international TOP500 list of the world's most powerful supercomputers and is based on 648 HP ProLiant SL230s Gen8 servers with Mellanox FDR InfiniBand. The system is twice as powerful than Purdue’s last 4 TOP500 systems, uses one half the power, and a quarter of the space.
All-in-all, HPC and other extreme scale out workloads have a lot to gain by taking a look at the brains built into HP Project Voyager and ProLiant Gen8 servers. Whether achieving greater levels of performance or new energy and space efficiencies, HPC needs more than scale to meet increasing demands up to exascale -- it needs brains. It needs the innovations that build intelligence into the server.
What are your thoughts on how innovation is helping meet the daily demands of ever increasing demands of performance?