Leveraging Resource for Innovation

In order to succeed, small businesses and entrepreneurs need to innovate in ways that set them above the rest of the market. Producing a product that merely tweaks existing designs will not be enough. To gain market share, new business ventures need to be providing goods or services that are something completely new. Both the Council on Competitiveness and the US DOE/Office of Science call simulation and computational science the "third leg" of innovation, with recommendations for the use of High Performance Computing (HPC) to make simulation more productive and cost effective. The Council on Competitiveness notes that there are three factors that tend to keep businesses from using HPC aggressively: Lack of talent/expertise, Lack of easy-to-use, scalable, production ready software, and cost/ROI. For start ups and small to mid-sized businesses ( SMBs), these hurdles can be especially high. Large HPC clusters are very expensive, programmers with HPC experience do not hire cheap and software licensing costs can add to the cost burden. While there are several programs that attempt to make these resources more available to businesses, we still have significant room for improvement. The US DOE's INCITE program, for example will benefit 9 different corporations in 2009 with grants of processing time on some of the nations largest supercomputers, but less than 10% of the grants given in 2009 will benefit non-academics. Although businesses as diverse as Intel and Dreamworks have been previous grant recipients, for small businesses still on the low end of the HPC learning curve, the INCITE program does not fill the HPC gap.
In many cases, government granting agencies actually make it difficult for academic grant recipients to assist for-profit ventures. For example, the NSF Grant Policy Manual (gpm05_131) states:

544 Principles Relating to the Use of NSF-Supported Research Instrumentation and Facilities

The following principles on use of NSF-supported instrumentation and facilities were adopted by the National Science Board:

The National Science Foundation seeks the maximum productive use of the Nation's scientific instrumentation and research expertise. Ensuring that the highest quality instrumentation, facilities, and services are available to scientific users, both academic and industrial, is a key requirement, as are harmonious relations and cooperation between industry and universities. Private research and testing laboratories, as well as university, government, and industrial laboratories, have a contribution to make.

The National Science Board recognizes that there may be circumstances where NSF grantees use NSF-supported research instrumentation to provide services in commerce for a fee, to an extent that such practice, (1) detracts from the performance of their obligation under the grant, and/or (2) may have a material and deleterious effect on the success of private companies engaged in the provision of equivalent services. It is contrary to the NSF's intent for grantees to use NSF-supported research instrumentation or facilities to provide services for a fee in competition with private companies in a manner that is prohibited by OMB Circular A-110.

Grantees should implement the above principles and related grant conditions in a reasonable manner. Grantees are expected to provide fair and adequate consideration of any complaints about use of instrumentation and facilities.

The "gray" nature of what is allowed and what is not, along with the concerns of time and resources needed to provide documentation and explanation if complaints are made has caused most universities and research institutions to avoid selling low cost services to businesses. While regulations like this were put in place to prevent grant recipients from forming enterprises that unfairly compete with unfunded business ventures, in times of economic trouble where outside the box thinking is needed, we need regulations that are not only clearly stated to permit some assistance to businesses, but actually encourage the huge pools of talent and resources currently locked in our universities to grow the HPC and simulation capabilities in the business ecosystem.

It is important to state that I am not trying to argue for the diversion of research funds away from pure science, or the pursuit of "big questions" ( climate shifts, the origin of the universe, cancer research, etc...), but I do believe that all of us could benefit just as much or more by allowing universities that purchase technology with grants to sell the leftover cycles to businesses. It does not matter if the cycles are actual computer cycles to run simulations, or leftover "cycles" of people time to provide consulting or training; fully utilizing the resources our tax dollars pay for to assist in business innovations seems only prudent in today's competitive market. I believe that this is important enough that I would be fully in agreement of preference for grants to recipients who can show a plan and an infrastructure to provide such services. Are you an entrepreneur or SMB owner? What could you do with access to low cost, high quality HPC resources, consulting services and/or training?

switching roles: journalist mode

I am getting settled in here at SC08 in Austin Texas. This is one of the conferences where I get to attend as Press, a fun hat switch for me. I like approaching new technologies from the perspective of outsider looking in, questioning how we can stretch the discipline and grow it in many directions. When I have my journalist hat on, it is how to grow it in the direction of Manufacturing IT.
Instead of looking as a consumer of technology, I get to brainstorm on behalf of many technology consumers, trying to think not just what my family or my company needs.. but what anyone could possibly do with some of the new ideas being introduced. I get to dream, think big and travel way outside of the box. Most of my "official" posts will appear over on the Intech High Performance Computing Blog, but I will try to keep posting here as well, including some pictures as the week goes on.

Computing Power leading to a decline in thought power?

I recently attended an Industry workshop for High Performance Computing at Purdue University. This was industry's chance to give feedback to the university types on what we need, what we don't and what we can't use, no matter how cool. It was a chance for us to get a peek into some of the up and coming research and development in HPC, so we can mull over the possibilities in the next couple of years. It was also a chance for us to network and chat with each other. I had a chance to talk to people supporting HPC activities at about a half dozen major companies and the consensus was immediately clear.

People are relying on the brute computing power of today's computers, rather than thinking carefully. We are apparently not the only ones to notice. Yesterday, Tim Walker posted a thoughtful piece on the usefulness of thinking as well. Don't misunderstand- I am not a luddite. There is no doubt that today's hardest questions require incredible computing power to answer. The power of the computer on my desktop and in my lap right now mean that I can process music and video in ways not possible with paper and pencil. But to think that because you have seemingly endless computing power at your fingers tips you do not have to take time to think before typing is not only wrong, it is wasteful.

Let me give you an example. This is a completely made up example based off of real life stories. The names, products, companies are all changed.. but the thinking ( or lack thereof) remains the same. The impact to the bottom line of the companies has been experienced more than once.

John Smith works for WingsRUs -an up and coming wing company. They are a aiming at becoming the leading supplier of wings to the global wing market, so they want a brand new design that is more efficient, more economical to make and helps the craft it is attached to use less fuel. All great goals. WingsRUs is technologically savvy and uses the latest modeling and simulation methods to test all of their designs before building prototypes. They have a large supercomputer cluster they use to run simulations for fast turn around of ideas. These are all great practices. John has the germ of an idea for a wing that will revolutionize flight. He is sure it will be the next breakthrough product. But there are 3 variables that vary by 4 cm each which could impact his design and he is not quite sure which combinations will give the best results. John can easily submit groups or batches of simulations to the supercomputer cluster at his company, so he decides to parametrize those variables, varying each by a millimeter at a time and running the simulation for each possible combination to find the best solution. Although each simulation runs in 4 hours, this results in thousands of simulations and the analysis of each- adding weeks to the overall design cycle and consuming large amounts of the company computing resources. John consumes so many cycles, that other engineers have trouble getting their jobs through the queue and their managers put together a project with real ROI to expand the cluster and add more computing power to the company resrouces. It has computable ROI because they are in a situation where their engineers ability to work is impacted and with the additional compute power, they can achieve more. However, the same end result could have been achieved without any cash outlay at all, if our friendly engineer John had taken the time to think carefully, done a reasonable design of experiment and run fewer simulations in the first place.

The insanity of this is that every single corporate IT person I have talked to in the last year has at least one of these stories. Sometimes many. Sometimes it is caught before new compute power is purchased, but not always.

You can make the same arguments for something like storage. Yes, storage is fairly cheap right now.. But just because it is cheap, should you fill it up with junk emails and multiple versions of 5 year old files that you never would have kept back when storage was expensive? Wouldn't it make more sense to take time to think for a moment, make reasonable choices and save the money on extra storage?

There are concerns other than costs savings involved as well. An unused brain does not reinforce neural connections and soon becomes "flabby" like an unused muscle. The end result is not just a loss of money for companies, but a loss of the national brain trust and the ability to think an innovate.

I love my computer- sometimes in ways that are probably not healthy. Some people would say I am addicted to the connectivity it provides for me. But I refuse to go one step further and let it think for me. What about you?

Geeky Fun

If I had not been driving my daughters to the airport yesterday, I would have been over at Purdue, watching the Rube Goldberg Contest. There may be those who see this as a silly waste of time and talent, trying to find the hardest way to accomplish a task, but it is also a great lesson in team building and creative thinking. I will, however be over at Purdue Monday night and all day Tuesday for the Industry Workshop on HPC- both to give them some insight into the world of manufacturing, as well as fodder for the HPCMFG blog.

What are your favorite Geeky ways to kick back, have fun and expand your mind?