Visit additional Tabor Communication Publications
HPC Matters is a joint blog consisting of contributors from the Tabor Communications team on their observations and insights into HPC matters.
November 27, 2008
It happens every year at the Supercomputing Conference and Expo, whether it's seafood in Seattle, pierogies in Pittsburgh, or anchos in Austin. Some days it's a feast, others a famine.
For those of you who, like me, used SC08 as an opportunity to synch up with important contacts, you know the drill. Or rather, you know the two different drills. In drill number one, you eat like a hobbit. After a 7:30 breakfast meeting you have a second breakfast meeting at 8:30. Then it's elevenses with your first lunch, followed by a proper lunch meeting that lasts a full hour. Fill up then, because you won't eat again until the meet-and-greet at 4:30, after which you have back-to-back dinners and at least two catered receptions to attend. You roll back to the hotel, and if you're lucky, you're at the Doubletree, so you can eat a fresh chocolate chip and walnut cookie in bed. (Really, those cookies are very good.)
Drill number two doesn't work that way. For whatever reason, your meetings line up differently the next day, and somehow you keep missing meals. If you're in the Doubletree, that delicious cookie is the only food you sniff all day. The Ritz will fetch you anything you want of course, but at other hotels, you might just go to bed hungry, dreaming of the two brunches you'll have tomorrow.
The food is incidental, of course, and what really matters is what you learn in the time you're there. Just in time for American Thanksgiving, here's a quick feast-or-famine guide to what you could gorge on in the HPC space, versus areas in which the cupboard is bare.
Interconnect Options: Feast
At my second dinner Monday evening, one vendor asked me if I thought InfiniBand could continue to survive with Mellanox as the only chip provider. Of course it could survive, I replied, because our data shows InfiniBand adoption continuing to grow in all interconnect segments. But it was likely that this growth would attract new entrants, and we'd eventually have more than one player. The next day, QLogic announced its own chips, lending a new option to a growing space.
Other sectors of the interconnect market are heating up as well. 10 Gigabit Ethernet has gone mainstream, with a wealth of vendors seeking to differentiate their offerings. The most interesting of these may be Myricom, which has remodeled its business plan to derive a majority of its revenue from 10GbE.
But neither of these developments is where I see the real action heating up over the next two years. The big trend to watch is now optics. Users in many verticals are expressing early interest in optical interconnects because of their potential in lowering latency. Avago, Finisar, Lightfleet, Luxtera, and Obsidian all had presences at SC08, and Tabor Research predicts we'll continue to see more from them over the next two years.
Utility Computing: Feast
Tabor Research surveys have shown a high degree of user interest in cloud computing. In this space, Arista made a splash by landing Silicon Valley visionary Andy Bechtolsheim, who is now championing that company's cloud strategy. In addition, a little-known company called Darkstrand made a big impression with its commercialization of access to supercomputing resources over the National LambdaRail network. Even demos in The MathWorks' booth showcased Amazon EC2. Cloud computing plays into the strengths of IBM and HP, the HPC server leaders, as well as Platform Computing, the leading middleware provider. Tabor Research expects these three companies to capitalize on the opportunity.
Server Vendors' Data Management Strategies: Famine
The evolution of pNFS is the most important story in HPC that no one is paying enough attention to. In beta now, pNFS will likely become a standard option in the first half of 2009 and go mainstream by the end of the year. With data management as one of users' most significant challenges, the server vendor community has yet to fully embrace the landscape shift that will start to take place between now and SC09.
The storage vendors are largely ready. Panasas has been publicly pushing pNFS since early 2007, and companies ranging from BlueArc to NetApp are well-positioned to gain from standardization in the file system arena. On the server side, the readiness is not as apparent, especially considering the big server vendors all sell storage products.
The lone strategic bright spot seems to be on Sun, the same company that popularized InfiniBand as a storage interconnect. Sun's Open Storage strategy is at times confusing but indubitably bold and is the most significant area of Sun differentiation in HPC. Plus, Sun owns Lustre now; I just wish they would do more with it. IBM has GPFS and will continue to ride that train successfully, at least for the near term. HP, Dell, and Cray all need to work on their data management solutions.
Accelerator Options: Feast
The biggest new product news here is NVIDIA's launch of its "personal supercomputer" based on its Tesla GPGPU. This exciting product has unfortunately been labeled with an albatross of a designator -- the industry is rife with failed, oxymoronic personal supercomputers -- but NVIDIA may have the clout and momentum to gain in academic and research markets while the marketing department figures out how to reposition for the industrial sector. The ecosystem around Tesla also continues to grow. NVIDIA partners (notably Acceleware) are pre-packaging Tesla-based solutions, and The Portland Group has been working on its own compilers for GPGPU. Intel's Nehalem CPU is an important consideration in this space, but in this case I think Intel's competitive heat is on AMD more than it is on NVIDIA.
But while everyone has been looking at GPGPU, an older accelerator category has slipped back into the mix: FPGA. Before you scoff, consider that FPGAs excel in areas in which GPGPUs have been weak, especially in scalable, high-bandwidth, text- or integer-based applications. This time around, programmability isn't at issue, because vendors like Convey, BlueArc, and XtremeData have integrated FPGAs into their offerings.
Convey enjoyed a terrific launch, thanks in part to the combined experience of Bruce Toal and Steve Wallach. (An interview with Wallach hit The New York Times during the show.) The FPGA-powered design is a major departure in computing architectures, and Convey could quickly become an interesting second-tier player alongside SiCortex, another differentiated architecture player with a strong SC08 showing.
Bring on the Next Course
Overall the energy of the show was strong, setting another attendance record (although that is as much a factor of the diversity of vendors than anything else). Still, amid these famine-oriented economic conditions, the SC08 floor was a refreshing island of plenty. In our most recent surveys, users indicated their budgets are still expected to grow over the next two years, and judging by the smorgasbord of activity in Austin, we could be ready to feast again in Portland in 2009.
Posted by Addison Snell - November 26, 2008 @ 9:00 PM, Pacific Standard Time
Addison Snell is the CEO of Intersect360 Research and a veteran of the high performance computing industry. During his tenure, he has established Intersect360 Research as a premier source of market information, analysis and consulting.
No Recent Blog Comments
The Xeon Phi coprocessor might be the new kid on the high performance block, but out of all first-rate kickers of the Intel tires, the Texas Advanced Computing Center (TACC) got the first real jab with its new top ten Stampede system.We talk with the center's Karl Schultz about the challenges of programming for Phi--but more specifically, the optimization...
Although Horst Simon was named Deputy Director of Lawrence Berkeley National Laboratory, he maintains his strong ties to the scientific computing community as an editor of the TOP500 list and as an invited speaker at conferences.
Supercomputing veteran, Bo Ewald, has been neck-deep in bleeding edge system development since his twelve-year stint at Cray Research back in the mid-1980s, which was followed by his tenure at large organizations like SGI and startups, including Scale Eight Corporation and Linux Networx. He has put his weight behind quantum company....
May 16, 2013 |
When it comes to cloud, long distances mean unacceptably high latencies. Researchers from the University of Bonn in Germany examined those latency issues of doing CFD modeling in the cloud by utilizing a common CFD and its utilization in HPC instance types including both CPU and GPU cores of Amazon EC2.
May 15, 2013 |
Supercomputers at the Department of Energy’s National Energy Research Scientific Computing Center (NERSC) have worked on important computational problems such as collapse of the atomic state, the optimization of chemical catalysts, and now modeling popping bubbles.
May 10, 2013 |
Program provides cash awards up to $10,000 for the best open-source end-user applications deployed on 100G network.
May 09, 2013 |
The Japanese government has revealed its plans to best its previous K Computer efforts with what they hope will be the first exascale system...
May 08, 2013 |
For engineers looking to leverage high-performance computing, the accessibility of a cloud-based approach is a powerful draw, but there are costs that may not be readily apparent.
05/10/2013 | Cleversafe, Cray, DDN, NetApp, & Panasas | From Wall Street to Hollywood, drug discovery to homeland security, companies and organizations of all sizes and stripes are coming face to face with the challenges – and opportunities – afforded by Big Data. Before anyone can utilize these extraordinary data repositories, however, they must first harness and manage their data stores, and do so utilizing technologies that underscore affordability, security, and scalability.
04/15/2013 | Bull | “50% of HPC users say their largest jobs scale to 120 cores or less.” How about yours? Are your codes ready to take advantage of today’s and tomorrow’s ultra-parallel HPC systems? Download this White Paper by Analysts Intersect360 Research to see what Bull and Intel’s Center for Excellence in Parallel Programming can do for your codes.
In this demonstration of SGI DMF ZeroWatt disk solution, Dr. Eng Lim Goh, SGI CTO, discusses a function of SGI DMF software to reduce costs and power consumption in an exascale (Big Data) storage datacenter.
The Cray CS300-AC cluster supercomputer offers energy efficient, air-cooled design based on modular, industry-standard platforms featuring the latest processor and network technologies and a wide range of datacenter cooling requirements.