Behind the scenes at Labs
cancel
Showing results for 
Search instead for 
Did you mean: 

An Oral History of The Machine—Chapter One: Origin Stories

Curt_Hopkins

HP_Discover2014_14119.gif

Chapter One: Origin Stories

By Curt Hopkins, Managing Editor, Hewlett Packard Labs

The Machine is a computing architecture so radically different than any which has come before that it will affect everything we do in the future. Hewlett Packard Labs has spent the last five years developing the memory-driven computing, photonics, and fabric that has gone into The Machine and which have made the impossible inevitable.

We spoke to several dozen researchers – programmers, architects, open source advocates, optical scientists, and others – to construct a nine-part oral history of the years-long process of innovating the most fundamental change in computing in 70 years.

These men and women are not only scientists, they are also compelling story tellers with an exciting history to relate. If you’re interested in how differently we will be gathering, storing, processing, retrieving, and applying information in the near future, or you just enjoy good stories about science and discovery, read on. 

If you would like to read other entries in the series, click here

STAN WILLIAMS

HPE Senior Fellow, Director of Foundational Technologies at Hewlett Packard Labs. Worked on precursor technologies that eventually became part of The Machine.

Water to the well

In 1994, Dave Packard was still involved in the company. He had had a discussion with then-director of Labs, Joel Birnbaum. He told Birnbaum that part of Hewlett Packard’s research portfolio should be devoted to more fundamental research. Molecular scale research would be a good idea.

That started the search for a founder of a more fundamental research group within Labs, and I was chosen. I got to meet Dave Packard as part of that process. The discussion we had addressed two issues. Larger companies, he thought, should be doing work with a longer strategic intent, looking ahead to problems 15 years out or more.

The other side of coin was, he was being altruistic. For years HP had been able to take research from others, academics and so on, and make products based on their work.

Now, it was time for HP to step up and participate in this community of discoverers and “return some water to the well.”

I was very inspired by Dave’s vision, but realized that going forward we would to stress the strategic, focusing on something that would make a difference for HP. It had to be a very large target, so that as HP evolved, what you were doing would still have relevance to the company.

Misty water-colored Memristors

I spent a long time “climbing the big tree,” trying to figure out where technology was headed. Moore’s Law as a technology momentum vector was driving the tech industry and would affect HP no matter what direction we were going. I also realized there had to be an end to it. Matter is finitely grained. We weren’t going to be making subatomic transistors.

In 1998, I decided what was needed was to focus research on what would happen after Moore’s Law. We would do excellent science along the way but we’d also have a good patent portfolio and be ready to make the transition better than others.

I knew we weren’t going to make a run directly at transistors. Transistors were reaching the end of their value. So we looked at memory. How could we make it small, high density, and low power? Could we replace such things as hard disks with directly addressable memory?

We investigated various possibilities and wound up asking ourselves, could we use molecules as building blocks for memory? So we researched molecular electronics, a sort of precursor to the Memristor. What we discovered was that molecules were too fragile. You could turn them on and off maybe 20 times before the molecules would burn up.

Bright lights, big city

When 2000 rolled around, I was reading papers on the 95th anniversary of Einstein’s naming of the photon. People had learned more in the last five years than the previous 90, so I went to management and said there should be a research project on photonics for interconnects.

The previous year, HP had split into HP and Agilent and everyone working on photonics had gone with Agilent, so no one at Labs was thinking about photonic interconnects.

We were told no. Photonics would never be important to computers and we’ll never do it. We waited a year and we did some more research and put together a much better story and presented the proposal a second time. Still a definite no-go.

So, I hired S.Y. Wong from Agilent and snuck him back into the company as a transistor guy. I got Ray Beausoleil transferred into my group. Together, we started what was effectively a skunkworks program in photonics.

By then, a bunch of photonic interconnects companies had sprung up around Silicon Valley that went bankrupt in 2000. Entire photonics labs were for sale on eBay, still in their boxes, for 10 cents on the dollar! Million-dollar pieces of equipment for $100K. I talked to Labs’ comptroller and got $100,000 from him for the project.

Abraham Lempel, my immediate boss, had heard me being chastised by management for my photonics plan and he kept an eye on me, looking in once in a while and asking, “Anything you want to tell me?” I’d respond, “Nothing you want to hear.”

So we ran our photonics labs in parallel with molecular electronics for a couple of years until we started to get significant results. Then I went back to management and said, “Look, this is what we have been able to do.” They agreed to think about it. When I went to Tech Con, the company’s annual internal technical conference, and over several drinks convinced a key executive to agree to fund a program in photonics.

KIRK BRESNIKER

HPE Fellow, Chief Architect for The Machine. With a background in HP Business Critical Systems, he was appointed to a leading role by Martin Fink.

Granite Bay

My thread started back in 2003 when I was the lead architect for entry platforms in what we called Business Critical Systems. BCS was all of the computing that wasn’t HP ProLiant servers, a union of HP-UX and VMS mission-critical enterprise systems and the HP NonStop fault-tolerant platforms. To this we eventually added the scale-up x86 platforms of the HP DL785, HP DL980 and HP Superdome X.

We were a couple of years into the HP/Compaq merger and we were starting to develop what would become a breakout product, the BladeSystem c-Class.  I had led the pre-merger development around blade systems in HP, having come to the ideas in one massive lightning bolt idea in late 1998. I was making some limited traction, but it wasn’t until I took my first trip to Houston a month after the merger and hooked up with the ProLiant teams and Mark Potter that everything started to click.  We were still three years away from shipping c-Class and about four more after that from fulfilling the vision I had back in 1998, but it was looking good.

It was looking so good, in fact, that I started to worry about how we were designing the rest of the BCS product line above my entry systems. Things like this start small. When something bugs you, sometimes you wonder if it’s just you, or if other people will think you’re trying to tell them how to design their products rather than take care of your own business.  But sometimes you find out that you’re not the only one who has concerns.  A handful of engineers I worked with started to get together at a friend’s house in Granite Bay (near Roseville, California). It was a 1960’s-style house on Folsom Lake, with a sunken living room in that part of the Sierra foothills where the suburbs were just turning to country.

We talked freely, riffing on some of the issues we saw coming up on us. One in particular was that we were building the follow-on to our flagship Superdome the same way we had built everything else, as though the market were the same, as though things were going to stay the same way as we headed into the first decade of the 21st century as they had been throughout the 20th. We knew that we needed to radically reorient how we were making our systems.

What my little cabal and I saw was that there was the possibility of combining the characteristics of HP BladeSystem and the BCS product lines to create something new.

We wanted systems that were more scalable, more flexible, cost less, and were easier for us to develop.  The first name we had for this idea was Boundaryless Computing, which I re-christened Blades Unbound after Shelley’s lyric drama Prometheus Unbound.

This all came together in 2005 at Tech Con in Phoenix.  Three of us wrote up Blades Unbound and it was accepted for the poster session. Tech Con’s poster session is like a science fair on steroids.  You have a poster you cram as much of your work on as you can and then the smartest people you’ll ever meet come by and put you through your paces. For a nascent idea like ours it was amazing. We made more progress standing in front of that poster in two days than we had in the prior two years. We were emboldened to continue refining our ideas. 

While we continued to work on BladeSystem and entry level BCS platforms, in everything we did we were trying to find ways to get Unbound ideas into our part of the product line.  When there were some changes to Intel CPU roadmaps in 2006, it became clear that a lot of what we wanted to do would have to wait for at least another generation, but we still kept the ideas current.

Hitting reset

In early 2007 some of our teams from Richardson, Texas, and Ft. Collins, Colorado, took me up on the offer to see what we could do and they were stunned to realize how capable the platform could be.  They immediately imagined a union of BladeSystem and Superdome technologies. This would mean a reset to the whole program that was already years into development.

Fortunately for us, Martin Fink was heading up BCS and when he saw the proposal he immediately grasped the possibilities. If he hadn’t re-directed us, we would have probably felt justified in “staying the course” and giving customers what they had traditionally asked for.

Once Martin effected this massive reset, his marketing lead, Michelle Weiss, told him to go and figure out what he thought the long-term story should be so that we'd have something to carry us through the delay while we made the shift. When Martin went off to figure it out was when his "Machine" thread started. 

The next year, I started as BCS Chief Technologist working directly for Martin. Just before I started working for him, I came to Ft. Collins and booked an hour for us to talk one-on-one. He started laying out this vision of his and I’m nodding my head and nodding my head. After an hour, he asked, “Have you guys already figured this out too?”

Ours was a slightly different take, but in essence, the answer was yes. It was clear what my group had been working towards and what he was talking about were very complimentary, two lines of thought coming together.

RICHARD LEWINGTON

Technical Communicator, Hewlett Packard Labs. Worked as Fink’s communications specialist at Labs.

Martin was talking externally about the needs behind the Machine before we started making it, discussing the “tectonic shifts” in the industry that he saw coming. He had been testing the waters at conferences, to get a feel for whether they thought we were insane or not.

They did not.

At Tech Con 2013 in Anaheim, Stan Williams hosted a “birds of a feather” track. Tech Con attracts inventors from all over the company – Labs, Enterprise Group, but also people from, say, the marketing analytics department. It’s pretty representative.

The “birds of a feather” track is a let’s get a beer and get together and talk about what we should be doing type thing. This one was on The Machine, which Stan laid out explicitly for the attendees. It was an internal temperature taking. The enthusiasm was amazing. There was not simply a lack of negative comments, but lot of how can I get involved comments, of holy sh-t that’s cool comments.

MARTINA TRUCCO

Manager, Strategy and Communications, Hewlett Packard Labs

I remember that we didn’t know how many people would show up to that first birds of a feather session. The room was designed to seat 30 to 40 people. But there were easily 100 people squeezed into the session when I got there, with more spilling out into the hallway!

KIRK BRESNIKER

HPE Fellow, Chief Architect for The Machine. With a background in HP Business Critical Systems, he was appointed to a leading role by Martin Fink.

In Las Vegas in 2014 at Discover, the company’s customer conference, Martin made The Machine public for the first time.

Making the announcement that we were going to build The Machine energized the entire company.

To read the other chapters in the series, click here

0 Kudos
About the Author

Curt_Hopkins

Managing Editor, Hewlett Packard Labs