Visit additional Tabor Communication Publications
December 07, 2007
IDC painted a rosy picture for the HPC server market in the third quarter of 2007. According to the analyst firm, revenue for the HPC market grew 8.8 percent from Q2 and 18 percent compared to the same period last year. The total HPC server revenue in Q3 worked out to $3.0 billion, which represents an astounding 22.9 percent of the $13.1 billion worldwide server revenue in Q3, a 0.5 percent increase from Q2 and a 6.3 percent increase from Q3 2006.
Another way to look at it is that non-HPC server revenue declined slightly in the third quarter and only managed anemic year over year growth. Without HPC to buoy it up, enterprise server sales are flagging. It's pretty clear that virtualization is asserting downward pressure on the overall server market, allowing users to get more application mileage out of less hardware. But at this point, high performance computing has resisted the effects of virtualization. The typical HPC application demands more compute cycles than are typically available in a single server, so sharing that server would defeat the purpose. Today, HPC application virtualization only makes sense over an entire server grid, but that model doesn't explicitly constrain server purchases.
Embedded in the surging HPC revenue are the processor sales. In 2006, HPC systems accounted for 26 percent of all processors sold in the server market. Intel's latest push to position its new "Harpertown" and "Wolfdale" Penryn server chips as HPC processors and AMD's struggles to feed the demand for its new quad-core Opterons are two indications of how important the HPC space has become to the chipmakers. While both Intel and AMD are busily adding virtualization support in their hardware, at some level they probably wish virtualization would just go away, since it reduces demand for their server chips. In HPC though, they see the opportunity to feed a beast with an insatiable appetite for compute cycles.
IDC attributes much of the HPC revenue growth to lower entry prices. The fastest growing segment is at the low end -- systems priced under $50,000. With more powerful processors available, workgroups, and even individual engineers and researchers can purchase HPC systems for as little as $10,000 dollars. IDC projects the sub-$50,000 segment to have an 11.4 percent CAGR through 2011.
The other big driver IDC identified is the growing trend of replacing physical R&D with computer simulation and modeling. Even if HPC system prices were stagnant, there is a cost benefit of using computing to replace physical experimentation, since the latter is almost always labor and capital intensive.
Looking at the yearly HPC revenue, IDC reported server revenue topped $10 billion in 2006. According to them, adding in other elements of the ecosystem like storage and services pushes the total to $16.3 billion. Even this figure may be conservative. Tabor Research, using an end-user research approach, estimates HPC server revenue represents only 43 percent of the total spending, even excluding things like staff, facilities and power/cooling costs. Using IDC's $10 billion server figure, the Tabor Research model would yield something closer to $23.5 billion in total spending. Now you're talking real money.
Of course, all of this growth is riding on the popularity of cluster computing systems, which in the third quarter represented 68 percent of all HPC server revenue. With no competing architecture on the horizon offering comparable price/performance for the majority of applications, clusters are destined to maintain their dominance in HPC for the foreseeable future.
Which brings us to something of a paradox. If clusters are such a growth industry, why aren't there more publicly traded cluster computing system vendors? The big vendors, like IBM, HP, Dell and Sun Microsystems, are public, but the HPC server business is just a slice of a much larger set of offerings. All of the HPC-only cluster system vendors, such as Appro, Linux Networx, Penguin Computing, etc., are privately held.
Supermicro Inc., a company that sells a range of all-purpose x86 servers, some of which make it into the HPC market, launched its IPO in March 2007. Although the company is profitable, the current stock is sitting at about $8.50/share (the initial offering was $8.00/share). With double-digit HPC cluster growth, one might have predicted better results. Time will tell.
The closest thing we have to a pure HPC cluster company that's publicly owned is SGI, although their globally shared memory Itanium-based Altix line technically disqualifies them. Plus, the company offers storage and visualization products and also generates significant revenue from services. In any case, as I reported last week, SGI's year-old cluster business is just now attempting to reach escape velocity and turn a profitable quarter. Their stock is currently just above $17/share, following an April high of $30.
Rackable Systems is another publicly traded cluster vendor, but, like Supermicro, it targets a much wider audience than HPC. The company also offers a number of storage products. Despite some innovative engineering in power and cooling, Rackable is finding it tough going in the cluster marketplace after a banner year in 2006. The company's stock is in no better shape than SGI's -- worse actually. It's now hovering at around $10/share, down from a January high of over $30.
Theoretically, the tier two players who specialize in cluster systems should be able to compete effectively against the tier one vendors, by either adding value, undercutting prices, or both. But the commodity nature of cluster computing cuts both ways. It makes entering the market easy, but establishing long-term differentiation hard. Almost everyone is building these machines from the same commodity parts: x86 processors, standard memory chips, Linux (or Windows) software, and Ethernet or InfiniBand gear. Higher value features that address usability, ease of deployment, and manageability are only now starting to be perceived as equally important as raw performance. With thin profit margins on the hardware, vendors are using these higher value features as their "secret sauce."
The tough competition won't dissuade vendors, though. With IDC projecting $15 billion in yearly HPC server revenue by 2011, there will be plenty of companies satisfied to get a slice of the market. The fun is just beginning.
As always, comments about HPCwire are welcomed and encouraged. Write to me, Michael Feldman, at email@example.com.
Posted by Michael Feldman - December 06, 2007 @ 9:00 PM, Pacific Standard Time
Michael Feldman is the editor of HPCwire.
No Recent Blog Comments
In a recent solicitation, the NSF laid out needs for furthering its scientific and engineering infrastructure with new tools to go beyond top performance, Having already delivered systems like Stampede and Blue Waters, they're turning an eye to solving data-intensive challenges. We spoke with the agency's Irene Qualters and Barry Schneider about..
Large-scale, worldwide scientific initiatives rely on some cloud-based system to both coordinate efforts and manage computational efforts at peak times that cannot be contained within the combined in-house HPC resources. Last week at Google I/O, Brookhaven National Lab’s Sergey Panitkin discussed the role of the Google Compute Engine in providing computational support to ATLAS, a detector of high-energy particles at the Large Hadron Collider (LHC).
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...
May 23, 2013 |
he study of climate change is one of those scientific problems where it is almost essential to model the entire Earth to attain accurate results and make worthwhile predictions. In an attempt to make climate science more accessible to smaller research facilities, NASA introduced what they call ‘Climate in a Box,’ a system they note acts as a desktop supercomputer.
May 22, 2013 |
At some point in the not-too-distant future, building powerful, miniature computing systems will be considered a hobby for high schoolers, just as robotics or even Lego-building are today. That could be made possible through recent advancements made with the Raspberry Pi computers.
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.
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.