In The Register’s first HPC channel webcast of 2011, Dan Olds of GCG talks to Rich Brueckner of InsideHPC to take the measure of current HPC performance and talk about the factors that got us to this point.

Some interesting data in the slides, including performance metrics of the #1, #100 and #500 systems on the Top500 list over the past ten years. When you chart out the results, you see a pattern where we just sort of plodded along until somewhere around 2006 – and then performance ramps up in hockey-stick fashion.

At this point, every server vendor+dog has some sort of integrated CPU-GPU system that they’re offering to their high performance computing (HPC) and other number-crunching customers.

The recently introduced Tetra from Appro has upped the ante by offering more GPUs than any other manufacturer in a small 1U package.

This solution is super-dense with up to 40 server units per rack, each with dual CPUs and quad NVIDIA Tesla M2050 GPUs. Appro offers this server in both Intel and AMD flavors – giving customers a choice between the 4- or 6-core Intel Xeon or the 8- or 12-core AMD Magny Cours processors.

In the video below, John Lee explains how customers decide between AMD and Intel: it comes down to the characteristics of the workload. The AMD procs can be configured with more memory and, of course, have more cores to address it.

Bring up the topic of overclocking to the major system manufacturers and you’ll hear a variety of responses ranging from, ‘We’re plenty fast enough now – take a look at our _____ (insert benchmark here) that’s enabled by our use of ___________ (insert techno feature here)” to “It’s just not safe, son, and our customers need to be protected from this kind of thing.”

Even with their liquid-cooled systems, the big vendors they still push back against any notion of using that greatly enhanced cooling capability to wring more performance out of the system.

Read More

The SCC marathon of 2010 is over. The final results were turned in Wednesday evening, the data examined by expert judges, and winners were announced at the SC10 awards luncheon on Thursday afternoon. Unfortunately, I had to travel to another meeting and couldn’t attend (or film) the ceremony – which leaves quite a hole in my wall-to-wall coverage.

I’m sorry I missed it. These kids worked hard, and I would have really enjoyed seeing how it all ended and seeing the winning teams get their proper recognition. Next year, I’ll definitely be there. That said, on to the results…

First, the highest LINPACK score was turned in by the University of Texas, Austin. They were the first team in SCC history to achieve teraflop performance with their 1.07 LINPACK result. While they’re the first team in the SCC Teraflop Club, they aren’t alone. Louisiana State and National Tsing Hua University (NTHU) from Taiwan also topped the TFlop barrier.

Read More

One of the most interesting aspects of the Student Cluster Competition is the way it stresses the humans even more than the hardware. The students have 46 hours to run as many of the benchmarks as possible, so they man their equipment around the clock in order to maximize their throughput.

The accompanying videos were taken early Wednesday morning, about 15 hours before their final results were due. This was the second and last night of the competition. Team moods were mostly quiet and subdued. The fatigue was showing on everyone, including me, as I asked the same questions over and over again. Despite my barely coherent interview technique, we learn more about the teams, their strategies, and the nature of the challenge.

‘Read More’ to see additional video coverage.

Back in the day, when I first started attending industry trade shows, I was always surprised by the range and number of smallish players competing head-on against much larger, industry-leading companies. (Incidentally, at the first trade show I attended, the wheel edged out fire as the most disruptive technology on the floor.)

I used to wonder how these firms survived what must be withering competition from the corporate titans who seem to have all the cards stacked in their favor. The big guys have name recognition and lots of different products, with big sales forces and even larger marketing budgets.

What I learned over the years is that bigger ain’t necessarily better – at least not in every situation. The smaller companies survive and even thrive by catering to important niches in the market and delivering the right product faster than the big guys.

At SC10, I’m exploring this topic by spending some time with Appro, a second-tier HPC vendor that competes day in and day out with IBM, HP, Cray, SGI, and Dell in the HPC arena.

Read More

Sunday at the Student Cluster Competition (SCC) at SC10 in New Orleans… we stop in at the Stony Brook booth to meet the team and see what they’ve brought with them. They introduced us to their cluster, dubbed the “Bear-O-Dactyl.” It’s a combination bear and pterodactyl – but I’ll let them explain the reasoning behind the nickname.

Texas is at SC10 with what looks to me to be the most elaborate booth of the SCC competition – at least in terms of artwork and theme. Their booth is festooned (or at least half-festooned) with “TACC to the Future” posters that play off the “Back to the Future” movies from the 1980s. TACC (Texas Advanced Computing Center) is the lead technical (and probably spiritual) advisor to the Longhorn team.

Read More

This is the first SCC for the Florida A&M (FAMU) team, and it couldn’t have started much worse. They found out that due to a delivery snafu, their system won’t arrive in time for them to participate in the competition. However, there’s more to the story – and it illustrates the true spirit of the SCC.

The Colorado team is no stranger to these competitions. They’ve been here before – four times before. While they’ve seen some success, winning the LINPACK crown (no actual crown), they’ve never won the big trophy (there isn’t any trophy of any size).