Interview: Brent Gorda from Intel on the Keys to Startup Innovation for HPC

Print Friendly, PDF & Email

Brent Gorda from Intel sets and achieves some fairly ambitious goals. His contributions to HPC have transformed the community. In fact he’s more than just an influencer, he’s the epitome of the StartupHPC ideal – his collective contributions in academia, government research and industry, including the launch of multiple startups and numerous educational programs, have been integral to inspiring STEM and HPC leadership for decades.

With SC15 on the event horizon and a keynote at this year’s StartupHPC Summit, Gorda offered us his personal insights, inspirations and perspectives on the state of HPC.

StartupHPC: What led you to the HPC industry and what’s the journey been like going from the lab, being a startup founder and an acquisition into one of the world’s largest companies?

Brent Gorda (right), General Manager, Intel High Performance Data Division

Brent Gorda (right), General Manager, Intel High Performance Data Division

Brent Gorda: I became interested in HPC, or parallel computing straight out of university. In my final undergraduate year, I attended a talk by Dan Wilson from Myrias Research where he challenged the audience to think in parallel: “if we are limited in performance by one CPU, we can break that limit by using many”. His description of parallel computing was so inspirational I had to speak with him about it. During the post-talk conversation, Dan wrote a phone number on a beer label (which I still have) leading to my first job in the test group (rather “as” the test group).

Working in HPC has been incredibly rewarding and I’m thrilled to see the recent upswing in the market. By its nature, HPC is constantly pushing the capabilities of technology to solve the world’s hardest problems and I find that to be both exciting and challenging. To have had the opportunity to be involved in a successful HPC startup is a unique experience that I will treasure forever.

StartupHPC: Reflecting on your diverse career, what inspired you to startup Whamcloud? Any Plans for another?

Brent Gorda: Whamcloud was started out of passion and a need to keep a critical technology in play for HPC. The passion was strong enough for me to leave a high-level position and start completely from ground zero

In hindsight, it has been rewarding for me professionally and the HPC community has greatly benefited as a direct result. Having been involved in a number of startups, it’s beneficial if leaders have a deep passion for the team, the product and the customer. Many founders have a deep passion for the anticipated “cashing in”, which I believe to be a sign of misdirected focus and energy.

As for what’s next for me, I see so many new technologies like Intel’s 3D Xpoint that are emerging. I think the startup opportunities are accelerating right now and it’s an exciting time to be involved in HPC. My personal aspirations are to be involved in innovation and help accelerate change.

StartupHPC: Where do you think startup opportunity exists (industry/technology gaps, emerging trends, etc.) in the HPC market at this point in time?

Brent Gorda: HPC-like technologies are becoming more ‘mainstream’ and being used to address enterprise needs such as big data, IOT and machine learning. Add in some of the emerging hardware technologies and there is plenty of opportunity to create value.

StartupHPC: Is there a need for/why do you think there are not more HPC-inspired startups?

Brent Gorda: HPC brings together the brightest minds to advance scientific simulation and computing. It is often regarded as a small market with hard-to-please customers. Building products that achieve critical mass for this environment is not easy. Many times the customer will think of a different way (“better” in their opinion) to solve the problem. This can make it difficult to gain acceptance for your solution and make that sale. The other side of this is that teaming with smart people in HPC can help you build the best products ever and accelerate your ideas. Our community is not generally shy to give feedback!

StartupHPC: You’re a tremendous industry leader, launching initiatives like the Student Cluster Challenge in particular. What advice do/would you give STEM students interested in/pursuing careers or creating companies in HPC? And what advice do you have for longtime peers to encourage/create more startups from HPC for HPC?

Brent Gorda: That’s quite the compliment, thank you! Launching the Student Cluster Competition was a huge amount of work, but another incredibly satisfying activity I initiated in 2006 (first run in Reno at SC07). It has been very successful with competitions held in China, Europe, South Africa to mention some. For me, a big indication of success is the number of contestants who end up joining the HPC community. The event has succeeded, as I had hoped, in matching community needs for talent with bright young enthusiastic people. Having this positive impact on the community will, I hope, be enduring.

For STEM students, the best advice I can give is to “be interested” first and foremost. As a natural outcome of that, become knowledgeable, find something you love to do and do it with gusto and enthusiasm! There are great opportunities in HPC as the market expands and by being interested, you open doors. By being knowledgeable and enthusiastic, you create opportunities to walk through those doors. Being at SC15 is a tremendous opportunity to meet people that can help you incubate a great career in HPC. For me, the talk that Dan Wilson gave in 1986 at the University of Alberta inspired me to go speak with him. Take advantage of the craziness that is SC: attend the technical sessions and interact with the speakers and topics that catch your eye.

StartupHPC: What do you think mandates like the recent NSCI or mergers (like EMC/Dell) mean for the HPC industry/community? And startups?

Brent Gorda: NSCI brings a heightened (and global) awareness to HPC as well as funding and aggressive targets for advancements in HPC. This will focus activity in the community, which means opportunity for those working on the right technologies. For the STEM community, understand that this is a longer-term project that will span many years. There is time to forge relationships and gain the knowledge necessary to participate.

The EMC/Dell announcement is about big business and HPC where the two entities have complimentary specialties. The growing HPC focus at DELL and the existing HPC talent we work with at EMC should create some interesting products as well as another opportunity for STEM talent to find employment, chase their interests and contribute to the HPC community.

There will also be openings for startups as there are things that startups are able to do more quickly and efficiently than large companies. I’m interested to see what pieces of technology may be acquired to compliment this merger.

StartupHPC: What are your top (3) picks in HPC invention/innovation?

Brent Gorda: I would start with these three:

#1. 3D XPoint from Intel with OPA: real storage/networking real close to the compute engine
#2. The convergence of Big Data and HPC
#3. Cloud infrastructure for HPC – making cloud work for HPC

Clearly these are hardware & infrastructure picks, which is where I focus my attention. A big topic these days is the applicability of HPC infrastructure to Big Data problems (the coming together of HPC and Big Data). I see significant upside in this particular area in the next few years.

Above the infrastructure layer, there exists a need to make these systems approachable for new users. There is renewed emphasis on the democratization of HPC or “blue collar computing” as Stan Ahalt from Ohio State called it 10 years ago, and a startup with serious design ability could really make a difference. As an example, when I started the cluster challenge in 2007, I reached out to a large American bicycle manufacturer and learned that they’d love to be able to use HPC for CFD (Computational Fluid Dynamics) in design, but they were unable to approach the complexity of the HPC infrastructure. This need has only increased over time.

StartupHPC: Based on your extensive expertise and experience, do you think we can encourage more HPC inspired startups?

Brent Gorda: Every government lab should have a “Future Technologies Group” staffed with experienced scientists with a mandate to interact with industry and help guide (not fund) ideas. I held this position at LBNL and was able to help many emerging technologies go in meaningful directions. Jeff Vetter has the job at ORNL. I think he sets a standard by which the other labs should aspire and help incubate new technologies through feedback. At Whamcloud, I really appreciated the interactions with these incredibly smart people. I had the advantage of being peers with many of them off and on for the past decade or more, but these people are approachable.

Another thought is to rally together an Angel fund for HPC startups. I’m so impressed with the work ethics of this community and think that Angels would get good returns for moderate investments in our community. The time is right for many of those ideas to succeed and Angels can provide the startup necessary to get them launched and into the market test phase.

Let me give you one final thought for the benefit of the STEM community. Successful startups are the ones who have knowledge and a strong focus on a problem to solve. Knowledge comes from experience and there is no better way to accumulate that than working at one of the HPC facilities in the US or abroad. In fact I participated in this directly, moving to the US to work at a DOE lab. These opportunities still exist but the Labs often find it difficult to compete with Facebook and Google. If you are able to attend SC15 in Austin, you absolutely must visit the booths on the show floor and engage with the staff and seek out opportunities that exist. Again, SC is where the industry meets research and amazing things happen: go make it happen for you!

Register now for the StartupHPC Summit at SC15