HPCwire: The Big Ugly

Since 1986 - Covering the Fastest Computers
in the World and the People Who Run Them

Translation Disclaimer

Subscribe | Sign In

Visit additional Tabor Communication Publications

Tabor Communications
Corporate Video

Blog: HPC Matters

HPC Matters is a joint blog consisting of contributors from the Tabor Communications team on their observations and insights into HPC matters.

HPC Matters | Main Blog Index

June 25, 2008

The Big Ugly

Back at the Newport HPCC conference in 2007, I broached the sensitive topic of whether the current ultra-scale procurement programs were good for the HPC industry by driving innovation into the market; or were these programs, in fact, draining resources, margin and long term opportunity out of the market?

This remained a hot topic in 2008, with Stephen Wheat of Intel driving a discussion about the business and R&D model for HPC innovation and the countervailing trends and behavior of the large funding agencies. If you haven’t read John West’s April 25th article in HPCwire entitled “HPC Innovation in the Era of “Good Enough,” I would urge you to do so. Using Wheat’s presentation and comments by others, including Dan Reed, he captures the issues and makes some persuasive arguments about what needs to be done.

But getting back to the issue at hand, I would like you to ruminate on the following: If you were to look at the history of NSF Tier 2 bids, the number of submissions has declined substantially. As a case in point, in the recent round, the NSF received only five bids. If you were to look carefully, both IBM and HP were conspicuously absent. In comparison, in the first round, IBM alone submitted eight bids! I would ask why?

Deals such as TACC do little to help the issue and unfortunately drive the business fundamentals out of the HPC business. While the Agency folks may think these ultra-scale systems are buoyed through cross-subsidization, I can tell you first hand they are not. The HPC divisions of even the largest suppliers, such as IBM and HP, must stand on their own merits. This means a strong business model and robust margins. The dirty little secret is that despite the high ticket price of these systems, there is typically not enough money to cover the direct and indirect costs of building a petascale system (remember software!), let alone margin to invest in the next innovation cycle. And in my humble opinion, the problem is getting worse, not better.

Now clearly this is not a simple issue as the broader stakeholders are all party to this dance. For the vendor community, the lure of being "chosen" is huge, further reinforcing this dynamic. As a long-term industry observer, the graveyard of suppliers who have fallen victim to the "halo" of participation is deep, with LNXI being the most recent casualty. If the big guys can’t sustain it, who can?

Posted by Debra Goldfarb - June 24, 2008 @ 9:00 PM, Pacific Daylight Time

Discussion

There are 4 discussion items posted.

changes in the HPC market
Submitted by John.McCalpin on May 31, 2008 @ 2:04 PM EDT

Most of the changes in the HPC market over the last 20 years have involved shifting to lower price points -- providing improved turnaround (due to reduced resource sharing) and improved price/performance (due to volume-related economies of scale). These have been very effective at enabling scientific & engineering computing at many levels, but with generally detrimental effects on the viability of the extreme high end. It seems to me that we have completed the "commoditization" of the scientific and technical computing markets -- cores are cheap enough that there is no need to share, and volumes of commodity processors are so high that there would be no cost benefit from making more of them. For a few years, low-end and mid-range HPC users have been able to afford more cores than they could effectively use for a single job, but have been able to exploit modest levels of task parallelism to keep their systems busy with useful work. Unfortunately, the "power wall" has forced almost all processor designs to rapidly shift toward greater and greater parallelism, with relatively steady single-thread performance. Improvements in cost/performance and power/performance can typically be obtained only by going further down this path, or by designing and configuring systems to be much more specifically tuned to a particular workload. It will be challenging to balance the requirements of creating customized and semi-customized systems for specific application areas without destroying the economies of scale that make commodity products inexpensive.

(Administrators note: This post was contributed by John McCalpin.)

Post #1

Innovation
Submitted by Matt on Jun 2, 2008 @ 10:44 AM EDT

Innovation isn't dead, it just doesn't thrive on winner-take-all risk-averse giant procurements. So if the big procurements don't act as technology spurs in the big organizations, and they aren't profitable on their own, why would BigCo bid on them? And commodity scaling isn't the barrier here: small teams can take advantage of the only true commodities in computing: DRAMs and silicon wafers. For more on this, check out http://www.bigncomputing.org/Big_N_Computing/Big_N_Computing/Entries/2008/5/8_Small_Teams_and_BIG_THINGS.html (sorry for the long URL)

(Administrators note: This post was contributed by Matt Reilly.)

Post #2

Innovation vs innovation
Submitted by Theo on Jun 2, 2008 @ 8:28 PM EDT

Clearly there is a big difference between the forces behind Blue Gene to enable innovation in domains that are deep research with time horizons of 5-10 years versus forces behind a small team startup that tries to leverage some new insight for economic gain.

Commercial entities will find it hard to invest in infrastructure for long term horizons (>10year time). They may invest in core competency but they can't invest in infrastructure without getting their stockholders up in arms. Talking about arms, the procurement of strategic weapon systems is also done up front: the science and engineering will follow, so defense related procurements have funded quite a bit of the engineering prowess of the US.

So big procurements do generate loads of innovation and much of aerospace is dependent on them. However, the trickle down effect of these procurements has proven to be negative. And I think that is the connection with your observation that commercially successful innovation comes from the bottom up. Did I understand you correctly?

(Administrators note: This post was contributed by Theodore Omtzigt.)

Post #3

Innovation and $
Submitted by Matt on Jun 5, 2008 @ 9:38 AM EDT

Theodore:
I was suggesting that innovation comes from "the bottom up," but I'd rather not think of myself at "the bottom."

The point that I was trying to make is that real innovation resulting from giant procurements is actually pretty rare.

As for strategic weapons systems, the computing market has gotten so large relative to the capital spent on the defense imperatives, that commercial
(specifically desktop and mobile) markets drive innovation in the largest companies.

And I've got stats, graphs, and 8x10 color glossies with a paragraph on the back of each one to elaborate on this. ;)

Matt Reilly
Chief Engineer
SiCortex, Inc

Post #4

Join the Discussion

Become a Registered User Today!

RSS Feeds

Feeds by Topic

Feeds by Industry

Feeds by Content Type

Subscribe to All Content

All

Debra Goldfarb

More Debra Goldfarb

Feature Articles

CERN, Google Drive Future of Global Science Initiatives

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).
Read more...

Saddling Phi for TACC’s Stampede

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...
Read more...

"No Exascale for You!" An Interview with Berkeley Lab's Horst Simon

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.
Read more...

Read more HPCwire features...

Short Takes

Running Computational Fluid Dynamics in the Cloud

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.
Read more...

Computing the Physics of Bubbles

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.
Read more...

Internet2 Awards Program Seeks Innovative Applications

May 10, 2013 | Program provides cash awards up to $10,000 for the best open-source end-user applications deployed on 100G network.
Read more...

Floating Funding to Exascale Island

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...
Read more...

HPC and the True Cost of Cloud

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.
Read more...

Read more headlines...

Best Practices in Big Data Storage

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.

Progress in Parallel: the Bull Parallel Programming Center

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.

View the White Paper Library

SGI DMF ZeroWatt Disk Solution

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.

Cray CS300-AC Cluster Supercomputer Air Cooling Technology Video

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.