Some of the participants in the CTO Design Challenge that conceived the new computer. (From left) Mark Anderson, Larry Smarr, Brad Holtz, Steve Coy, Ty Carlson, Mike Riddle, Ken Kreutz-Delgado, Tim Stonehocker, David Schoenberger, and Hugh Bradlow. (Photo by Kris Krüg)
New startup building ‘desktop supercomputer,’ seeking big breakthroughs using chips that work like the human brain
The first computer system incorporating IBM’s TrueNorth computer chip was conceived in Deer Valley, Utah last month, in a mere four days at the Future in Review conference.
Under the direction of Mark R. Anderson, a handpicked group of computer executives from different companies designed a desktop supercomputer system as part of a CTO challenge at Anderson’s annual conference, also known as FiRE.
They designed the Pattern Computer as a general purpose computer, one that runs on any operating system, and is highly efficient, extensible, scalable and unbelievably fast. It will be built using commodity hardware, connect to the new 100Gbps Pacific Research Platform, and run TrueNorth and other new brain-inspired chips as co-processors.
Hound CTO Tim Stonehocker talks with Kamran Elahian and Brett Horvath at FiRe. Photo: Kris Krug
By Berit Anderson
After more than ten years of development and with more than 200 million users on parent company SoundHound, virtual assistant Hound seems poised to take on Siri, Cortana and Google Now all at once.
One of the things that sets Hound apart is context: CTO Tim Stonehocker explained that Hound can actually remember your preferences and the context of previous questions to allow you to speak naturally.
What areas are you searching within? What do you consider walking distance? What is your definition of ‘open late’? Hound will remember all of these things to answer more complicated queries that include inherent personal judgments and assessments: For example, find me restaurants within walking distance of home that are open late.
“I feel like when I’m speaking to Siri I have to speak like Siri, like a robot,” said moderator and Scout co-founder Brett Horvath. Hound, he said, really picks up well on accents and natural language — a symptom of SoundHound’s years of musical and sound analysis.
But even calling Hound a virtual assistant does it a kind of disservice. Hound has already scaled the interface across 50 domains, setting it on-par with Siri and Cortana. But it has also launched Houndify, a development platform available to developers by request, to help it scale to all domains and across devices.
“We have thousands and thousands of other companies .. who are developing things using our platform,” said SoundHound investor Kamran Elahian. Elahian is the chairman and co-founder of Global Catalyst Partners.
What will that mean? Stonehocker described seeing a SoundHound developer eying the company espresso machine a few weeks ago. He could tell, he said, what he was thinking, “and I encouraged him to take it apart.”
“Now we have a Houndified coffee machine.”
In a remarkable demo video, Stonehocker showed Hound brewing an espresso while talking about the weather in Park City, the altitude and what football games are on this week.
As Elahian sees it, Hound could be a breakthrough technology for the Internet of Things. “Our vision has been that Hound would be the one that would unify all these things … When you’re talking to your coffee machine about your favorite sports team and you move on to your toaster, it should know [what you were talking about].”
As the first investor in SoundHound, Elahian also recounted how he’d seized the opportunity to fill Hound’s first $5 million round when the two were still just Stanford masters students with a great idea.
“I told them, I really like what you’re doing. I really like the vision. Don’t meet with Kleiner Perkins. Don’t meet with Sequoia. I’ll give you the $5 million.”
On Wednesday morning, Ed Butler, BBC Reporter, led a panel discussion introducing six of twelve FiRe Starter companies — top startups chosen by the Strategic News Service (SNS) for world promotion at the FiRe conference. These companies displayed innovative technologies that are making great strides towards improving our world.
Goldman noted that in the late 1990s the tech market, characterized by the dotcom bubble, stood in stark relief to today’s more rational market behavior. He sees an increased focus on fundamentals, a sense of discipline which makes it harder to take companies public. While Goldman noted that the presence of roughly 120 unicorns in the market has led to some overvaluation, the structure of valuation itself has shifted. While capital is in high supply, fewer companies have venture arms today. He noted that most of the money is coming from the hedge and mutual funds, and the big ones; Blackrock, Fidelity, and T. Row Price.
Further, Goldman explained that much of the valuation increases today occur while companies are private, a reversal from previous valuation patterns. Public companies just aren’t seeing the increases they used to.
Doug Jamison noted that after the dotcom bust, sub-50 million dollar IPOs have faded away. This fundamentally changes structure of market. Jamison noted that the challenge of unicorns is their allure; they suck money from smaller firms, often those involved in complex but critical deep-science innovation. Jamison stated a clear concern for this loss of diversification, noting “My thinking is…is that good for the market?”
Goldman pointed out further concerns in modern investment. He noted that venture has been a great asset class for endowments, but that volatile investments are also a result. Noted Goldman, “I was talking to a banker friend of mine, and he was saying ‘With an IPO, guys would do three months of due diligence. With these late stage private investments, they’re doing 2 weeks.” In other words, then, the difference in discipline between public and late-stage private is unbelievable. Goldman also raised concerns that companies who start to miss their numbers in these hyped private deals end up risking their ability to go public.
Doug Jamison highlighted the amount of legitimate valuation in the market, stating “If you parse the system, there are a lot of companies where valuations are not inaccurate.” He noted that companies, particularly those in the life sciences and energy fields, have suffered a downturn in valuation that also places them in more stable and accurate territory. Citing the shift of capital from limited partnerships towards a smaller subset, Jamison pointed out that if this is a rearrangement, it could be positive and disruptive. If the contraction is more long term, however, the diversification decrease will close the doors on deserving investments.
Goldman also noted that the number of parallel tech trends today is high, with 8 or 9 disruptive fields emerging now. While noting that most companies have business models that make sense long term (an improvement on the many companies in 2000 who had no business or revenue model to speak of) Goldman pointed out that things don’t go up and to the right forever. It is still difficult to see whether some of the valuations we see today really make sense.
When asked about crowdfunding as a disruptive way to raise capital in modern investing, both men expressed concern. Goldman noted problems with investing without long term commitment and knowledge, highlighting a lack of just reinforcement for long term investments in the crowfunding world. Jamison agreed, stating that while the democracy of investment brought by crowdfunding was good in the short term, investing isn’t solely about the money. He stated similar concerns with long-term investments, noting that “Invesment is like playing poker, but you get to count the cards.” If you can take in info and measure probability, you benefit. This may be lost in short-term crowfunding. As an example, Jamison queried, “Take drug trials, what crowdfunding body will be there 7 years in?” In terms of quick-to-market investments however, Jamison sees crowdfuning as a beneficial model.
Generally, then, both experts see secondary markets to capital as both an emerging realm of threat and possibility. However, the overall sentiment was troubled in regards to these sources of capital, with Jamison stating “We need secondary markets, but I just don’t see the demand”, while Goldman worried that such tools provided executives with a mechanism to take money off the table, asking “Can senior executives get distracted by not being hungry enough to go public, and therefore not be well aligned with investors?”
As for what is undervalued? Jamison sees a phenomenon in which Biotech firms have taken up much of the investment capital and left many sectors behind. Top of the list: IoT hardware firms.
Shane Wall, CTO of Hewlett- Packard, sat down with Ed Butler at FiRe today to have a chat about the future of 3D printing. While Wall’s predictions and descriptions ranged from 3–30 year timelines, one thing was clear; the future is now.
In particular, Wall described HP’s pivot towards a new, blended reality. What does that mean? It means we live in an era which regularly blends our physical world with the digital. From the Internet of Things to wearable tech, we regularly send and receive digital data to track and interact with our physical activities. The physical and the digital are no longer two distinct categories.
A leading example of this shift is 3D printing, one of HP’s new focus areas. According to wall, 3D transformation is tantamount to the future of production. Wall explained that “3D transformation […] is the next industrial revolution, and we mean it in that full sense.”
So far we hominids have made products the same way throughout history. We stage in raw materials, manufacture them into value-added products, and then stage those products in distribution centers and optimize the entire chain. He who optimizes the chain best (whether with production efficiency, better marketing, or faster distribution) wins.
That system, says Wall, will be turned on its head over the next 30 years. We will move to a model of democratized design in production. It will be simpler than ever to design the products we actually want and print them instantly.
Wall cautioned that to-date 3D printing has seen lots of hype, and often in its least valuable areas. Media portrayal of 3D technology tends to focus on the consumer side, highlighting cheap home printers that tend to produce low quality goods very slowly. This is a system of production of what are termed “extruded parts”, essentially a “glue gun on a table” model with low quality output. That total market, Wall noted, is a tiny percent of the broader market served by 3D printing. The real progress, he stated, will be in larger commercial and industrial applications, at least in the near term.
Commercial methods will focus on “centered laser-ing”. Using this method, a laser traps a bed of powder and fuses it, resulting in better quality parts. This technology is still slow and limited in quality, but the next 3–5 years, according to Wall, will see a major transformation in efficiency.
What is more, HP is developing a new 3D printing methodology, multi-jet fusion. Due to be released sometime next year, Wall explained that multi-jet will revolutionize 3D printing. Operating 10 times faster than other processes at just 20% of the cost with a significantly extended materials list, multi-jet allows control at the voxel level (a voxel is, in essence, a 3D pixel). The implication? You could print something as complex as a fully functional guitar with color and material differentiation, fluidity between rigid and flexible parts, and electronics.
A major concern in this field is intellectual property. Clearly the ability to print something on demand opens a world of illicit copying, from basic consumer goods to machine parts, guns to fake drugs. However, Wall noted that we can track digital representations and diagrams online. What is more, 3D watermarks can be placed in products from specific printers. While it will be far from simple, Wall noted that technology exists that allows tracking to show how a part was printed, who owns it, and which printer created it.
Perhaps most fascinating in Wall’s interview was his outlook on the future of 3D parts. How could 3D printing get more mind-boggling? Wall sees 4D self-assembly printing coming forward in the next 30 years. 4D parts would be printed with specific characteristics at their boundaries. These characteristics will be activated to stimulate joining for self-assembly.
In short, picture that you have multiple parts with certain properties on their edges; place them together, add water, and they assemble themselves; or, as noted by BBC Senior Broadcast Journalist Ed Butler, “No more Ikea struggles”.
Wall explained that as parts increase in strength and application (we are already manufacturing airplane parts using 3D printing technology) there will be societal implications. The removal of human participation in production requires some forethought and planning, to be sure. Another caveat is that 3D printing will not remove the need for the basic materials, such as rare earth metals, that we find problematic today.
All caveats aside, Wall noted that the system HP is creating today, a system wherein open platform printers allow us to advance faster than any individual firm could do so, will lead to a new era of production. It will be an era of rapid social change, massive industrial advances, and a streamlined lifestyle previously unimaginable.
The takeaway? Buckle your 3D printed, flexible-rigid integrated, self-assembling seat belts.
Dharmendra Modha has something to share with the world, and it’s making a splash. Modha spoke today at the SNS FiRe conference in Park City about the creation of IBM’s DARPA-funded TrueNorth chip, a pattern recognition processor capable of vastly exceeding all previous processing capabilities while using a fraction of the power.
Mohda described his humble beginnings, working his way from 100 neuron models all the way through rats and monkeys to human scale. He noted that by modeling a processor after the human brain, we can mimic a social network of 10 billion neurons. A real time simulator of that magnitude would require 12 gigawatts of power, or the entire generation capacity of the nation of Singapore.
This is instructive. As our processors have gotten faster and hotter, we have reached the upper limits of speed and heat in processing. The brain, on the other hand, is extremely slow and cool while processing massive quantities of information. A bee brain, for example, is a supercomputer the size of a postage stamp with the power load of a hearing aid battery.
The architecture of brains, Mohda explained, is parallel, distributed, slow, and modular. If you connect cortices from any direction they can easily communicate. The TrueNorth chip operates more efficiently, therefore, in almost every way. As well as the above characteristics, it is far cooler than its predecessors. The end result? We can achieve pattern recognition using large quantities of censors while using orders of magnitude less power.
According to Mohda, “We are literally creating synaptic supercomputers, very very quickly.” The underlying concept is a sea-change in our understanding of processing. Until now, Mohda notes, we have been trying to put a square peg in the proverbial round hole in the processing world. Rather than picture small shifts in power and capacity, Mohda says, “We need to envision an entirely new ecosystem. Chips come and go, ecosystems endure.”
What are the applications? Imagine glasses for the visually impaired that help them navigate complex environments without any need for internet connections. Imagine a tumbleweed robot that can roll into a burning house and search for missing children. Imagine censors that can measure ocean temperature, turbidity and more to track pollution.
This, Mohda states, will transform “not only business, but society itself […]We see a whole new generation of chips, systems, programming languages, software ecosystems, algorithms, and applications; a completely different way to make the world work.”
Where does this put us now? Mohda, with a sly grin, noted that we are at a point with synaptic supercomputing best described by Winston Churchill. “This is not the end, it is not even the beggining of the end. But it is, perhaps, the end of the beginning.”
In short, Mohda says, “The brain remains, and will remain, the frontier of discovery for at least a century.”
“Mark and I have been waiting for this moment for many many months,” announced Larry Smarr, the Founding Director of Calit2 at Future in Review Wednesday morning.
He was referring to FiRe Chair and SNS CEO Mark Anderson, who took to the stage with Smarr to announce the creation of the UCSD Pattern Recognition FiRe Lab. The lab will develop machine learning that can identify patterns using IBM’s TrueNorth chip. (TrueNorth inventor Darmendra Modha is also a speaker at FiRe 2015.) Then it will put that technology to work analyzing big data to identify patterns and learn about the world — from genomics to the study of the microbiome to physics and engineering.
IBM & DARPA are contributing TrueNorth. The Cal system, the University of Tokyo, Lawrence Berkeley Lab, Lawrence Livermore Lab & Stanford will act as research partners in the lab.
“Somehow biologically evolved computers have found a way to be a million times more energy efficient,” Smarr said. “We must be able to find a way to reverse engineer the brain to do that.”
“The pattern recognition processor is almost like a camera — it’s interfacing with the world in a very blunt way,” Mark explained.
Anderson first wrote about pattern computing in February of 2013. A year later Science published an article about Modha’s work on TrueNorth.
Ken Kreutz-Delgado, Professor of Intelligent Systems, Signal Processing, and Robotics at UCSD’s Jacobs School of Engineering, will direct the lab. He described the TrueNorth chip and pattern recognition lab Wednesday as “an opportunity for people who really want to understand patterns of real world data”.
Smarr said he also sees the lab as a way to attract world-class talent working on big data from around the world. “Could we find people around the world, the big minds working on big data applications, that we could quickly bring in,” he wondered.
One of those talents is Akihiro Nakao, a Professor of Applied Computer Science at the University of Tokyo. Nakao’s team developed FLARE, an open programmable deeply node architecture that he described Wednesday as “trying to define the new generation Internet.”
Bibop Gresta, Hyperloop Transportation Technologies’ Deputy Chairman and COO, speaks at Future in Review in Park City, Utah. Photo: Kris Krug.
Bibop Gresta’s name, fun as it is to say, was perhaps the least interesting thing about him to the crowd gathered at the Future in Review technology conference in Park City, Utah, Friday night. The international Italian VC also happens to be a former pop star, who claims on his LinkedIn profile to have become a software development director by the time he turned 15.
Just two years later, with plenty of crowdsourced input from JumpStarter users, Hyperloop Transportation Technologies has plans to break ground on its first stretch of tube in 2016.
Hyperloop passengers will travel in pressurized capsules, levitated on an air cushion inside of a reduced-pressure tube. Eventually, the craft will travel at speeds of up to 760 mph, although Gresta said Friday that the only time passengers would feel the gravitational force is during starts and stops. Aerodynamics, he explained, make windows impossible, but high-res screens inside each capsule will show what’s going by outside in real-time if you like. And there will be no lying down uncomfortably during the journey. Gresta described the passenger experience as plane-like.
“We are building it on completely renewable energy,” he announced, citing solar panels planned for the top of the tube, kinetic energy to reclaim wind movement, regenerative breaking and geothermal energy. A Hyperloop line between San Francisco and Los Angeles, he predicted, will cost at most $10 billion to build and will be able to transport 3,400 people an hour — 67,000 each day.
Laying high-speed rail along the same route would cost $68 billion, forty percent of which Gresta attributed to the cost of clearing the right-of-way. The Hyperloop, built on 5-foot pylons spaced 400-feet apart, won’t have that challenge.
One of the project’s biggest assets is the willingness of its team to forego pay — 450 top engineers from 21 nations have traded $11,000,000 in man hours in exchange for stock.
To avoid dealing with government regulations and citizen pushback, that first stretch of tube will be built as part of Quay Valley, CA — a proposed 750,000 person solar-powered development in the middle of California’s San Joaquin Valley. Early testers of the technology won’t be traveling at full-speed. The Quay Valley tube will be an exploratory eight kilometers long, limiting capsule speeds to ‘just’ 360 mph.
“We will [still] smoke the Japanese high-speed rail,” Gresta smirked.
But even if Hayes goes belly-up again, one gets the feeling that Hyperloop Transportation Technologies will be just fine.
“Im a pessimist on seeing the first Hyperloop in America, honestly,” Gresta said Friday, citing legal and regulatory challenges. There are plenty of other countries, he explained, that would be happy to draw a straight line between two points and simply clear the right-of-way for Hyperloop development. Hyperloop Transportation Technologies is already talking to four other interested nations.
Beyond that? Gresta sees a successful Hyperloop revolutionizing industries like freight and medical transport and said the company is in talks with the Port of Los Angeles, the Port of Singapore and twenty-one others.
“We are not having the technology yet to go underwater, but we are working on it,” he joked.
“China’s latest factory data is the worst in 3 years. What’s wrong with China’s business model? Mark Anderson is CEO of INVNT/IP, a consortium of US companies and experts who’ve put together a report, claiming that some 50% of Chinese growth in recent decades has been founded on the stealing of western business ideas, via old-fashioned industrial espionage and more sophisticated state-sponsored hacking. He exclusively tells the BBC the basis for his claims. And we also hear from Chinese author Edward Tse, who says the old stereotypes of Chinese companies leeching off western technology and possessing few ideas of their own is outdated. He’s spent years advising Chinese companies, and in a new book, China’s Disrupters, he claims a new genuinely entrepreneurial and innovative spirit has transformed the country’s business climate.”