Feeds

BIG IRON cluster warriors put pedal to HOT METAL

We node you like those accelerators

Boost IT visibility and business value

HPC blog We saw an interesting assortment of gear from the teams competing in the SC13 Standard (Big Iron) Track of the Student Cluster Competition.

As you can see on the handy table below, the teams had a lot in common on the surface. They all ran some variety of Intel Xeon as their base CPU, most were using Mellanox Infiniband interconnects, and every team had some sort of accelerator or co-processor.

But there are plenty of differences as well. The Australian iVEC and the Chinese NUDT teams, for example, took a 'more nodes is more better' approach, sporting eight nodes each. The Aussies also went for an interesting node strategy, with all 8 of their accelerators located on only two nodes.

You'll also notice that almost all of the teams went in for accelerators in a big way, festooning their clusters with either 8 NVIDIA Keplers or 8 Intel Phi co-processors. Team Longhorn (Texas) took what they call a more balanced approach with only four GPU accelerators.

Using this many accelerators can be somewhat risky. While they can add a whole lot more processing power, they can only do so on applications that are optimised to use them efficiently.

The problem for competitors is that SCC rules do not allow teams to power down or physically remove any component after the start of the bout. So if you have an app (or apps) that don't efficiently use your accelerators, you can't do much beyond idling them to reduce the drain on your power. And even at idle, having a bunch of these things can really take a chunk out of your 26 amp/115 volt power allowance.

In upcoming stories and video blogs, we'll take a closer look at each team, their gear, and their rationale for making the decisions they made. We'll also give you an idea of how the competition proceeded day by day and, of course, cover the first Pro-Am Celebrity Cluster Challenge. (Which sucked up a huge amount of my time, since I was the organiser and ultimate arbiter of the Pro-Am. It was fun, but fraught with challenges. Stay tuned for more.) ®

The essential guide to IT transformation

More from The Register

next story
The Return of BSOD: Does ANYONE trust Microsoft patches?
Sysadmins, you're either fighting fires or seen as incompetents now
Microsoft: Azure isn't ready for biz-critical apps … yet
Microsoft will move its own IT to the cloud to avoid $200m server bill
Oracle reveals 32-core, 10 BEEELLION-transistor SPARC M7
New chip scales to 1024 cores, 8192 threads 64 TB RAM, at speeds over 3.6GHz
Docker kicks KVM's butt in IBM tests
Big Blue finds containers are speedy, but may not have much room to improve
US regulators OK sale of IBM's x86 server biz to Lenovo
Now all that remains is for gov't offices to ban the boxes
Flash could be CHEAPER than SAS DISK? Come off it, NetApp
Stats analysis reckons we'll hit that point in just three years
Object storage bods Exablox: RAID is dead, baby. RAID is dead
Bring your own disks to its object appliances
prev story

Whitepapers

5 things you didn’t know about cloud backup
IT departments are embracing cloud backup, but there’s a lot you need to know before choosing a service provider. Learn all the critical things you need to know.
Implementing global e-invoicing with guaranteed legal certainty
Explaining the role local tax compliance plays in successful supply chain management and e-business and how leading global brands are addressing this.
Build a business case: developing custom apps
Learn how to maximize the value of custom applications by accelerating and simplifying their development.
Rethinking backup and recovery in the modern data center
Combining intelligence, operational analytics, and automation to enable efficient, data-driven IT organizations using the HP ABR approach.
Next gen security for virtualised datacentres
Legacy security solutions are inefficient due to the architectural differences between physical and virtual environments.