An Interview with ISC’09 Keynote Speaker Andy von Bechtolsheim

By Nicole Hemsoth

June 21, 2009

When 1,500 leading members of the world’s high performance computing community convene June 23-26 at the 2009 International Supercomputing Conference, The opening keynote address will be presented by Andreas “Andy” von Bechtolsheim, the legendary co-founder of Sun Microsystems and founder and Chief Development Officer of Arista Networks. Von Bechtolsheim will discuss “The Evolution of Interconnects for High Performance Computing.”

ISC, which will be held in Hamburg for the first time in the 24-year history of the conference, has a well-established reputation for presenting well-founded, precise and up-to-date information in an environment that encourages informal conversations and sharing of ideas. And of all the thought-provoking sessions scheduled for ISC’09, none are likely to spark more discussion than the keynote addresses.

In his presentation, von Bechtolsheim will discuss trends in the high performance computation market, including the challenge of building large fabrics and the role of InfiniBand and 10 Gigabit Ethernet. He will also look at how to address the challenges of building, integrating, and using petascale systems including system power and cooling, system stability, and scalablity. Finally, he will look at the impact of solid state memory for HPC deployments and how it can address data bandwidth within the system to deliver improved overall performance through a more balanced system architecture.

Andy Bechtolsheim, Arista Networks Co-Founder Von Bechtolsheim was a co-founder and Chief System Architect at Sun Microsystems, responsible for next generation server, storage, and network architectures. From 1995-96, he was CEO and President of Granite Systems, a Gigabit Ethernet Switching startup company he founded that Cisco acquired in September 1996. From 1996 to 2003, he was VP Engineering and later General Manager for the Gigabit Systems Business Unit at Cisco System that developed the Catalyst 4000/4500 Gigabit Switch family, the highest volume modular switching platform in the industry.

Von Bechtolsheim earned a M.S. in Computer Engineering from Carnegie Mellon University in 1976. He was a doctoral student in Computer Science and Electrical Engineering at Stanford University from 1977-82. He has been honored with a Fulbright scholarship, a German National Merit Foundation scholarship, the Stanford Entrepreneur Company of the year award, the Smithsonian Leadership Award for Innovation, and is a member of the National Academy of Engineering.

The following interview with von Bechtolsheim by Christoph Poeppe from “Spektrum der Wissenschaft” (the German sister publication of Scientific American) was translated by Jon Bashor and Heike Walther.

Spektrum der Wissenschaft: What drives a person, who was apparently meant to pursue a scientific career, to take a path that leads him to such exceptional commercial success? What went wrong?

Bechtolsheim: I don’t see any fundamental conflict between science and commercial success, at least not where I work — in Silicon Valley. All in all, though, I have always been much less interested in academic research and much more interested in how to build better products that drive a commercial success.

Spektrum der Wissenschaft: But didn’t you start out as a physicist?

Bechtolsheim: Not really. In 1974, I did win the German Science Fair in Physics building a device that could precisely measure flows using ultrasound, and in high school I took advanced classes in physics and bio- chemistry, because these were the most interesting classes that were offered. But I was always much more interested in computers and computer science, which is really an engineering discipline. There have been very few major breakthroughs in mathematics and theory in the last twenty-five years that affected the field of computer science. All the new advances that we have seen were really based on better engineering.

Innovation in the computer field is very different than innovating in a traditional industry such as chemistry. At the moment, “Green Energy” is a big focus for venture capitalists. But to make ethanol at a lower cost, you need an unbelievably large amount of investment capital to build new facility, and this is difficult to come by these days.

In information technology, many of the most successful new companies were started with very modest capital. For example Google, which has become the most successful search company, was financed with just 30 million dollars of venture capital.

And Google has been branching out to offer all kinds of new services and applications.

Spektrum der Wissenschaft: I’m really only familiar with Google as a search engine…

Bechtolsheim: Besides the Google search engine, there is also Google Maps and Google Apps and Google Talk and the YouTube video portal – the possibilities stretch out from there. The end user just needs a browser and an Internet connection to use all these services. The computer work is done inside Google’s gigantic data centers, where with clever engineering and large scale, Google has achieved enormous cost advantages compared to conventional data centers.

Spektrum der Wissenschaft: How so?

Bechtolsheim: Google has built a reliable system environment out of a large number of simple, low-cost servers. Google builds its datacenters in locations that have low-cost power and cooling, and it manages these data center with very few people. It is estimated that the cost per CPU hour in a Google datacenter is between one-fifth to one-tenth of a traditional enterprise data center.

Spektrum der Wissenschaft: What’s your personal connection to Google?

Bechtolsheim: My friend David Cheriton, who is a professor at Stanford, introduced me to Sergey Brin and Larry Page. Their idea to sort search results by relevance, which is calculated by the number of links between websites, convinced me right away. It does not matter what the content of a website is, the only thing that counts is how many and how many relevant websites are linked to this website. This approach is immune against tricks some sites use to artificially raise their hits, such as embedding the same word many, many times in a way that is invisible to the user.

And the business model of linking relevant search results to relevant sponsored links was a stroke of genius that had not occurred to anyone else.

Spektrum der Wissenschaft: In your new company, Arista, you are focused mainly on building network switches. What pushed you in this direction?

Bechtolsheim: All large web companies are building large data centers for what is now called “cloud computing.” This concept used to be called grid computing, computing clusters or server farms. There is extensive data transfer among the servers in these cloud compute clusters. The end result of this computational work, such as a list of search results, doesn’t contain much data, but to calculate the relevance of a website, the page rank, you have to look through large amounts of data.

The demand for bandwidth rises in proportion to the speed of the servers and the number of servers in such a cloud. With 10,000 servers that require 1 gigabit per second per server, the cloud network has to move 10 terabits per second. Of critical importance is that the switches allocate bandwidth fairly to all servers and connect them with very low latency.

Spektrum der Wissenschaft: Do you build your own switch silicon for your systems?

Bechtolsheim: No. In contrast to 10 years ago, today there are very good switch chips and network processors that there is no need to develop your own silicon, which is extremely expensive to do.

Spektrum der Wissenschaft: What do you bring to the table?

Bechtolsheim: We develop the network software. A switch needs to respond to a large number of protocols to operate correctly. We have developed a very modular and robust network operating system that we call EOS, which has separate processes for each task in the networking stack. If a process fails or gets updated, it does not affect the operation of the switch and the system continues without interruption. As a result our system is very stable. Further, EOS runs on top of a standard Linux kernel.

This means we can run any other program on the same switch, including customer specific solutions.

Spektrum der Wissenschaft: How many computers can one switch handle?

Bechtolsheim: Customers usually configure 20 to 40 computers per rack. Our rack-top switches have up to 48 ports, 40 of which connect to the computers in the rack and the rest connect to our core switch, which has hundreds of ports. This allows us to support very large clusters with 10,000s of servers.

The computers are so fast nowadays that in many cases the network bandwidth has become the limiting factor. With our switches we offer customer a great way to increase overall system performance.

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industry updates delivered to you every week!

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion XL — were added to the benchmark suite as MLPerf continues Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing power it brings to artificial intelligence.  Nvidia's DGX Read more…

Call for Participation in Workshop on Potential NSF CISE Quantum Initiative

March 26, 2024

Editor’s Note: Next month there will be a workshop to discuss what a quantum initiative led by NSF’s Computer, Information Science and Engineering (CISE) directorate could entail. The details are posted below in a Ca Read more…

Waseda U. Researchers Reports New Quantum Algorithm for Speeding Optimization

March 25, 2024

Optimization problems cover a wide range of applications and are often cited as good candidates for quantum computing. However, the execution time for constrained combinatorial optimization applications on quantum device Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at the network layer threatens to make bigger and brawnier pro Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HBM3E memory as well as the the ability to train 1 trillion pa Read more…

MLPerf Inference 4.0 Results Showcase GenAI; Nvidia Still Dominates

March 28, 2024

There were no startling surprises in the latest MLPerf Inference benchmark (4.0) results released yesterday. Two new workloads — Llama 2 and Stable Diffusion Read more…

Q&A with Nvidia’s Chief of DGX Systems on the DGX-GB200 Rack-scale System

March 27, 2024

Pictures of Nvidia's new flagship mega-server, the DGX GB200, on the GTC show floor got favorable reactions on social media for the sheer amount of computing po Read more…

NVLink: Faster Interconnects and Switches to Help Relieve Data Bottlenecks

March 25, 2024

Nvidia’s new Blackwell architecture may have stolen the show this week at the GPU Technology Conference in San Jose, California. But an emerging bottleneck at Read more…

Who is David Blackwell?

March 22, 2024

During GTC24, co-founder and president of NVIDIA Jensen Huang unveiled the Blackwell GPU. This GPU itself is heavily optimized for AI work, boasting 192GB of HB Read more…

Nvidia Looks to Accelerate GenAI Adoption with NIM

March 19, 2024

Today at the GPU Technology Conference, Nvidia launched a new offering aimed at helping customers quickly deploy their generative AI applications in a secure, s Read more…

The Generative AI Future Is Now, Nvidia’s Huang Says

March 19, 2024

We are in the early days of a transformative shift in how business gets done thanks to the advent of generative AI, according to Nvidia CEO and cofounder Jensen Read more…

Nvidia’s New Blackwell GPU Can Train AI Models with Trillions of Parameters

March 18, 2024

Nvidia's latest and fastest GPU, codenamed Blackwell, is here and will underpin the company's AI plans this year. The chip offers performance improvements from Read more…

Nvidia Showcases Quantum Cloud, Expanding Quantum Portfolio at GTC24

March 18, 2024

Nvidia’s barrage of quantum news at GTC24 this week includes new products, signature collaborations, and a new Nvidia Quantum Cloud for quantum developers. Wh Read more…

Alibaba Shuts Down its Quantum Computing Effort

November 30, 2023

In case you missed it, China’s e-commerce giant Alibaba has shut down its quantum computing research effort. It’s not entirely clear what drove the change. Read more…

Nvidia H100: Are 550,000 GPUs Enough for This Year?

August 17, 2023

The GPU Squeeze continues to place a premium on Nvidia H100 GPUs. In a recent Financial Times article, Nvidia reports that it expects to ship 550,000 of its lat Read more…

Shutterstock 1285747942

AMD’s Horsepower-packed MI300X GPU Beats Nvidia’s Upcoming H200

December 7, 2023

AMD and Nvidia are locked in an AI performance battle – much like the gaming GPU performance clash the companies have waged for decades. AMD has claimed it Read more…

DoD Takes a Long View of Quantum Computing

December 19, 2023

Given the large sums tied to expensive weapon systems – think $100-million-plus per F-35 fighter – it’s easy to forget the U.S. Department of Defense is a Read more…

Synopsys Eats Ansys: Does HPC Get Indigestion?

February 8, 2024

Recently, it was announced that Synopsys is buying HPC tool developer Ansys. Started in Pittsburgh, Pa., in 1970 as Swanson Analysis Systems, Inc. (SASI) by John Swanson (and eventually renamed), Ansys serves the CAE (Computer Aided Engineering)/multiphysics engineering simulation market. Read more…

Choosing the Right GPU for LLM Inference and Training

December 11, 2023

Accelerating the training and inference processes of deep learning models is crucial for unleashing their true potential and NVIDIA GPUs have emerged as a game- Read more…

Intel’s Server and PC Chip Development Will Blur After 2025

January 15, 2024

Intel's dealing with much more than chip rivals breathing down its neck; it is simultaneously integrating a bevy of new technologies such as chiplets, artificia Read more…

Baidu Exits Quantum, Closely Following Alibaba’s Earlier Move

January 5, 2024

Reuters reported this week that Baidu, China’s giant e-commerce and services provider, is exiting the quantum computing development arena. Reuters reported � Read more…

Leading Solution Providers

Contributors

Comparing NVIDIA A100 and NVIDIA L40S: Which GPU is Ideal for AI and Graphics-Intensive Workloads?

October 30, 2023

With long lead times for the NVIDIA H100 and A100 GPUs, many organizations are looking at the new NVIDIA L40S GPU, which it’s a new GPU optimized for AI and g Read more…

Shutterstock 1179408610

Google Addresses the Mysteries of Its Hypercomputer 

December 28, 2023

When Google launched its Hypercomputer earlier this month (December 2023), the first reaction was, "Say what?" It turns out that the Hypercomputer is Google's t Read more…

AMD MI3000A

How AMD May Get Across the CUDA Moat

October 5, 2023

When discussing GenAI, the term "GPU" almost always enters the conversation and the topic often moves toward performance and access. Interestingly, the word "GPU" is assumed to mean "Nvidia" products. (As an aside, the popular Nvidia hardware used in GenAI are not technically... Read more…

Shutterstock 1606064203

Meta’s Zuckerberg Puts Its AI Future in the Hands of 600,000 GPUs

January 25, 2024

In under two minutes, Meta's CEO, Mark Zuckerberg, laid out the company's AI plans, which included a plan to build an artificial intelligence system with the eq Read more…

Google Introduces ‘Hypercomputer’ to Its AI Infrastructure

December 11, 2023

Google ran out of monikers to describe its new AI system released on December 7. Supercomputer perhaps wasn't an apt description, so it settled on Hypercomputer Read more…

China Is All In on a RISC-V Future

January 8, 2024

The state of RISC-V in China was discussed in a recent report released by the Jamestown Foundation, a Washington, D.C.-based think tank. The report, entitled "E Read more…

Intel Won’t Have a Xeon Max Chip with New Emerald Rapids CPU

December 14, 2023

As expected, Intel officially announced its 5th generation Xeon server chips codenamed Emerald Rapids at an event in New York City, where the focus was really o Read more…

IBM Quantum Summit: Two New QPUs, Upgraded Qiskit, 10-year Roadmap and More

December 4, 2023

IBM kicks off its annual Quantum Summit today and will announce a broad range of advances including its much-anticipated 1121-qubit Condor QPU, a smaller 133-qu Read more…

  • arrow
  • Click Here for More Headlines
  • arrow
HPCwire