Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• A near-term outlook for big data
• 2012: The Hadoop infrastructure market booms
• Defining Hadoop: the Players, Technologies and Challenges of 2011

There is a ton of computing involved in this sport, and it all serves a different purpose.  On the one hand, there’s the horsepower required to compute different airspeed dynamics over the car’s form, while on the other there’s the massively parallel computing required to analyze the streams of data thrown off the cars in real time. And there’s a lot of gear that is packed up and travels from race site to race site.

The great F1 traveling IT show

Ahead of the show I talked to Mehul Kapadia, VP of strategic alliances and sponsorships from Tata Communications, about how some of the Formula 1 computing happens in the field. And one of the coolest things is that everything happens in the field — from the broadcasting of the race to TV and to apps to the creation of a DVD highlights video covering the whole season, most of the video and web output of the F1 occurs on a race site.

The place it occurs is a broadcast center that contains 160 tons of different gear, from servers to video-editing modules owned by F1. That gear is transported to each race on a jumbo jet and is designed to be set up in 24 hours and dismantled in 12 hours. The video feeds from the track to the rest of the viewing world is handled by satellite, but Kapadia is hoping that soon that will change. Tata is one of the largest IP backbone companies in the world and is hoping that after this year it can help transition the F1 world from satellite to fiber.

The trend of sending more and more HD-quality sports content via fiber networks instead of via satellite is one I’ve been hearing about for several years. Yet it’s a slow transition, in part because broadcasters and organizations that manage the sports are hesitant to mess with a formula that works. For all of the F1 races Tata provides a gigabit of connectivity to the track itself, which is more than 10 times the capacity than F1 had prior to the Tata sponsorship.

“In the middle of Austin this isn’t a big deal, but some of these sites are fairly remote,” said Kapadia.

And because the race teams are using on-site connectivity to stream data from hundreds of sensors inside the car to their teams that perform real-time calculations on-site, the extra capacity is going to get used. Those calculations help the racing teams tell their drivers when to come in for a pit stop or how to handle the course or other drivers. Somewhere in all that information, there’s almost certainly a great big data case study.

How much will F1 put in the cloud?

Tata is also hosting the Formula 1 web sites and apps in its data centers. And like most sports, the demand on the web site fluctuates incredibly — hitting peaks right before and during the race itself. It also experiences interest from all over the world, since F1 is a truly international sport. This poses geographical challenges and also means that any prime time isn’t constrained by a certain time zone. Tata hosts some of the site’s content on its on-demand InstaCompute cloud, but it also has built a content delivery network using technology it acquired through the purchase of BitGravity in 2011.

But still much of the F1 IT work is handled on site at the broadcast center, which means that F1 IT staff are on the road a lot — even if it’s to exotic locales. By putting more of its operations in the cloud and by having fat pipes connecting the racetracks, Kapadia hopes that he can take F1 online and let more take place from remote locations. Because as cool as it is to have a traveling IT infrastructure and support team, it’s also a logistical challenge — and something better connectivity and the cloud could help solve.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• A near-term outlook for big data
• Dissecting the data: 5 issues for our digital future
• What Amazon’s new Kindle line means for Apple, Netflix and online media

Given how anxious people are about the battery life on their mobile phones and how much more computing these devices are handling, one would think venture capital firms would rush to back Adapteva, which launched in May 2011. But Olofsson couldn’t find investors. He blames it on the reluctance of venture firms to back chip startups — and they certainly are leery of investing in capital-intensive hardware startups– but it could have been any number of reasons: bad business plan, a realization that handset makers weren’t going to swap out a Qualcomm application processor for an untried Adapteva option, or something else.

But instead of packing it in, Olofsson has turned to Kickstarter to bring his vision of supercomputer power in a tiny, low-power package to the market. He wants to sell a processor on a stripped-down board in two sizes as well as open source the software that will be needed to operate and program the chip. Called the Parallella project, the plan is to offer the 16-core board to those who pay $99 with the goal of raising $750,000. If the team can reach a stretch goal of $3 million it will also offer the 64-core version of its chip for $199.

Olofsson says he was inspired by the market and hobbyist community that’s building around Arduino boards and the Raspberry Pi, a low-power and low-cost computer. Earlier this month researchers at the University of Southampton made a supercomputer using LEGO bricks and Raspberry Pi modules. Olofsson acknowledged that project but pointed out that the Adapteva chip could deliver a lot more power — the 64-core version of his board delivers 51 gigahertz (compared to the 1.4 gigahertz processor inside the Samsung Galaxy 3) while consuming only five watts (that’s still a lot for a phone).

The Parallella boards will cost more however. Arduino boards or Raspberry Pi computers cost roughly $35 each as opposed to $99. But Oloffson is undaunted. He says researchers are already playing around with Adapteva chips for supercomputing and other projects, and aims to build a community. Kickstarter’s recent changes on how hardware companies will list their hardware caused minor launch snags for the Adapteva team, but nothing major.

In fact, the Kickstarter changes, which were designed to emphasize the funding nature of the platform instead of having people who backed products thinking of it more like a store, fit with Olofsson’s ideals. For him, bringing massively parallel computing to phones and other devices is a mission, not just a business.

“I feel like we have a better mousetrap here and yet the adoption has been very slow,” said Olofsson. ” And so we want to speed that up. As for the building a community, that’s our goal. We are talking a pretty scary step in opening our architecture. If this works we’re open sourcing and open licensing all of our SDKs. That’s the point of no return.”

Indeed. After raising a Series A of financing and taking on a convertible note to get to this point, Oloffson is betting his hopes on the Kickstarter community. There will be no Series B for Adapteva. In exchange for the community support, he’s stripped down his board and is opening up parts of his business that would be impossible for another chip company, where IP is everything. Make no mistake, this is a Hail Mary pass for his company, but it’s not one other chip startups could necessarily follow.

If Olofsson succeeds it may seem to be a new way of backing capital-intensive hardware firms, but in reality Adapteva spent $500,000 even getting the first version of its chips made. The military and its Series A strategic investor bore that cost, but declined to support the move beyond the $10,000 boards that are currently all Adapteva has to get people to embrace its new design. Let’s see if the Kickstarter community decides to give Adapteva and massively multicore parallel computing a whirl.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Crowdfunding’s rapid growth and future opportunity
• Facebook’s IPO filing: ideas and implications
• GigaOM Research highs and lows from CES 2013

Apple's A5, manufactured by Samsung.

Apple might have quad-core iPhone and iPad devices coming in 2012, according to some code discovered deep in Apple’s iOS 5.1 pre-release software on Friday. We’ve been hearing a lot about the possibility of a quad-core A6, successor to the A5 processor that currently powers iPhone 4S and iPad 2 devices, but now there’s reason to believe those new processors are being actively tested.

The iOS 5.1 beta contains references to quad-core chips in a hidden system panel that describes core configurations supported by iOS software, according to 9t05Mac. In the latest beta, it’s been updated to include a reference to “/cores/core.3″, which follows “/cores/core.0″ and /cores/core.1″,  references to single and dual-core devices, respectively. A “core.2″ designation is skipped, which would represent a triple-core processor according to the naming system.

Apple is thought to be testing the iPad 3 and possibly a next-gen iPhone on the iOS 5.1 beta, which could mean those devices are packing a quad-core processor, although it’s also possible these references are related to a much more distant release. But since Android phones with quad-core processors are already making their debuts, and quad-core Android tablets have already hit the market, it does seem likely Apple would want to anticipate the wider uptake of that tech in its 2012 mobile lineup.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Research In Motion: future scenarios for its fate
• Why iPad 2 Will Lead Consumers Into the Post-PC Era
• Analyzing the wearable computing market

SuVolta

SuVolta doesn’t want to design chips, it wants to make the process that fabrication plants will use to build the devices. Its technology cuts the energy used in chips in half, and requires a fairly simple tweak of the chemicals layered onto the chip during the manufacturing process. The resulting chips made using SuVolta’s process are just as fast but consume about half the power.

This power reduction is cool, but it’s not the main reason why SuVolta’s on this list. SuVolta tweaks both the manufacturing process and the circuit design. But the process works best for systems on a chip, as opposed to stand alone processors. A System on a chip (SoC) is when multiple types of processors are placed on a single chip as an integrated package.

SoCs are common in the mobile world because they are a way to cram more functionality into a smaller package and they consume less power. SuVolta’s President and CEO Bruce McWilliams, believes SoCs will be the way of the future for how most chips are built.

Ambiq Micro

Ambiq is commercializing technology out of the University of Michigan to build a real-time clock designed for sensors. The clock consumes less power, but also takes over functions that currently involve other chips in order to reduce the power usage of the sensor even further (yup, it’s like an SoC microcontroller). Scott Hanson, the CEO and co-founder of Ambiq explains that today’s sensors usually contain a microcontroller, a clock that puts the chip to sleep and wakes it as necessary, a power supply, a sensor of some sort (typically a MEMs device) and a radio.

But Ambiq combines the clock and the microcontroller so the chip requires less power and takes up less space. Some proposed uses of the chip include implanting it inside the human body, or a chip that can run on tiny solar cells the size of a penny (see image).

As we put more sensors on devices and inside our infrastructure, Hansen believes we’re about to open up a new frontier for chip design firms who can build chips for the sensor web. Ambiq is his bet on this, but he expects many more. With an investment from ARM, he’s not the only one betting on a new generation of chips that will need specialized microcontroller and a smaller size, the British licensing company clearly sees an opportunity as well.

Adapteva

The demand for power in mobile devices and in the servers that power large web sites such as Facebook or Google has led to a boost for ARM, which licenses a chip architecture that trades performance speed for power efficiency. For phones this is fine, but for tablets and even servers, it may be time to think up an entirely new architecture. That’s where Adapteva comes in. The company has rethought a RISC-based architecture for chips and built massively multicore chips that are built to run in parallel or independently.

Much like an older startup called Tilera, which is also building massively multicore chips for data centers, Adapteva thinks that x86 doesn’t offer the energy efficiency needed, while ARM doesn’t offer the performance that next generation mobile devices such as tablets and servers will need. So it’s borrowing the concept of massively multicore chips from the high performance computing world and dialing it down for tomorrow’s mobile applications and up for the next generation of HPC. In the coming years, we’ll see more massively parallel chips, but we’ll also see a willingness to jettison the tried and true architectures as we embrace more specialty computing.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Cloud computing’s impact on chip and hardware design
• Cloud computing infrastructure: 2012 and beyond
• Ups and downs for cleantech in Q1

Jeff Martens CEO of CPUsage

Do you sleep? Have a laptop or desktop that sits idle during those eight hours? Need an extra $10 a month? If so, startup CPUsage has a proposition that you should hear. The eight-month-old startup wants to pay folks so it can harness their idle compute time to sell to corporations. CEO and Co-Founder Jeff Martens estimates an average user donating four hours of compute time every day could score about $10 a month.

Martens and his two other co-founders want to turn their Portland, Ore.-based startup into the Folding@home or SETI@home of the for-profit world. The goal is to enroll users and use their computers to help corporate customers (the startup already has two) speed up their analysis jobs. The company’s software breaks up a job into bits and sends those bits to the user’s computer for parallel processing. One customer uses the service for decoding agricultural DNA. Martens knows it’s not right for all jobs, as latency is high and there might be security concerns.

However, he stressed that each node only gets 1/500 of the data to process, which makes it harder to reassemble the job. The map above shows where the company has nodes today. Also, unlike Plura Processing, another company doing this sort of CPU harvesting for profit, CPUsage works directly with its members as opposed to through a game or other intermediary, so its software resides on the user’s hardware as opposed to harnessing CPU cycles through a browser. This allows for extra security over how information is treated, asserts Martens.

The idea of harnessing idle compute time stretches all the way back to 1999 when the SETI@home project was created to help listen for extraterrestrial life. Other non-profit projects followed, including Folding@home, which studies proteins to find cures for diseases. But Martens believes there is a market for folks who would be more interested in giving up their idle compute time if they were directly compensated.

Under his planned model, for every dollar CPUsage earns, about 45 cents goes back to the person who owns the computer CPUsage is harnessing. The company charges customers about 15 cents per CPU per hour in line with the pricing for Amazon’s medium-sized instances. Obviously, this isn’t a solution for everyone, but it might be useful for enough people to make a viable business.

And Martens thinks it could do some social good as well. He says the company is talking to the Portland School District to take on some of the idle computers in the district’s schools — which Martens thinks could generate $1 million for the district in a year. Another customer of CPUsage recently hit it big in the media for its ability to harness compute power to find lost gadgets.

For now, Martens goal is to raise in the range of $750,000 for a seed round to help expand the number of computers CPUsage will have in its system. He would also like to move from the current architecture that requires all jobs and traffic to flow through the company’s servers to more of a peer-to-peer Skype-like design to help cut down on bandwidth costs.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• From car to cloud: the future of the in-vehicle app landscape
• What’s Next for the Cloud? Distributed Architectures
• The role of converged infrastructure in the data center

We’ve been tracking news of dual-core processors for mobile devices throughout 2010, but have been looking to the future for actual products that will use them. Qualcomm, for example, recently announced a two-core version of its Snapdragon chip called the MSM8960. However, that silicon won’t be sampling to device-makers until sometime in 2011, so it won’t power products prior to that. Texas Instruments, also has a new dual-core chip in the pipeline, the OMAP4440, which boasts two 1.5 GHz computer cores. The company expects production of the OMAP4440 in the second quarter of 2011. While these chips are “coming soon,” Nvidia’s Tegra 2 will be in the Optimus 2X smartphone next month and rumors indicate it will power several tablets too.

So what exactly does a dual-core CPU in a phone bring to the user? Much like upgrading to a personal computer with the latest and greatest processor, these chips can improve the overall speed of a smartphone but still maintain judicious battery life. That’s important, because even the fastest mobile devices are essentially useless if the battery only lasts a few short hours. These chips can improve overall speed, handle 1080p video playback without breaking a sweat, and boost webpage loading by 33 percent, according to TI. In a lengthy whitepaper on multicore CPUs, Nvidia mentions a multi-tasking boost: A phone’s navigation app could run on one 1 GHz core while a streaming audio application could run on the other 1GHz core.

Indeed, dual-core chips are the future for more powerful, yet power efficient smartphones as Stacey noted in a GigaOM Pro report on the topic (subscription required). Back in April, she wrote:

As the lines between computers and mobile devices blur, traditional PC vendors are building phones and the traditional phone manufacturers are trying to build mobile PCs. But with mobility come constraints — particularly around power consumption and battery life. So the big task for every device manufacturer is figuring out how to cram all the functionality of a big computer into a tiny handset. Many chip firms believe tomorrow’s phones will be powered by multicore processors that deliver the performance the consumer wants without destroying the lengthy battery life such devices need.

While dual-core smartphones will bring immediate performance gains without sacrificing a device’s run-time, more potential awaits. Clearly a phone’s operating system can help manage processing power by leveraging two or more computing cores, but mobile app developers could achieve gains by writing applications optimized for multi-core use. Through the use of parallel computing, software can better leverage computers with multiple cores. By taking the lessons learned for building software to run on multiple cores in servers developers can build apps that deliver faster results because the processing can take place on the phone. Richer media and games are also a byproduct of multiple cores.

So the full gain from dual-core chips in smartphones may take time to realize, much as it took time for Nvidia to get a top-tier handset-maker to use its Tegra 2 chip. A dual-core phone announced in 2010 and shipping a few short weeks later is a bit of a surprise to me, but I actually expected that if one did appear this year, it would be running on Nvidia’s platform: last month I saw the window of opportunity open for the chip company and this month I see that LG stepped through it. And as Om noted back in May, the arrival of such chips is actually closing the mobile window for Intel (s intc), as it struggles to get its x86 chips to use less power.

Related content from GigaOM Pro (sub req’d):

• Marketing Handsets in the Superphone Era
• Today’s Smartphones Give Rise to Tomorrow’s Robots
• For Phones, the Future is Multiple Cores

Google built MapReduce to handle its particular needs, which are a far cry from the needs of most businesses. Database analyst Curt Monash told Computerworld that the study just reinforced his belief that MapReduce is better for limited tasks like text searching or data mining — you know, the things Google does on an epic scale. For tasks that require relational database capabilities at web scale, database sharding has become a favorite practice. I’ve heard Google itself uses SQL, MapReduce and/or sharding depending on the task. Companies like Aster Data Systems and Greenplum give companies the functionality of both MapReduce and SQL in one user-friendly package.

I think MapReduce (and its variants, like Hadoop) have received a lot of unnecessary adoration thanks to the fervor over cloud computing. Some people, it seems, associated Google, Yahoo and their web brethren with cloud computing, and thus surmised that in order to do cloud computing, you must do exactly what Google and Yahoo do. This, of course, is not the case. From a business perspective, cloud computing is just as much about saving money and making life easier as it is about doing massive amounts of computing. If you don’t have unique computing needs like the web giants, but just want eliminate the joys of owning and managing machines, there are plenty of clustered SQL solutions available in the cloud. Like most things in life, it’s just a matter of finding the right tool for the job.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Ups and downs for cleantech in Q1
• A near-term outlook for big data
• Real-­time query for Hadoop democratizes access to big data analytics

In this context, there is some good news for parallel programming. Data analysis software parallelizes fairly naturally. In fact, software written in SQL has been running in parallel for more than 20 years. But with “Big Data” now becoming a reality, more programmers are interested in building programs on the parallel model — and they often find SQL an unfamiliar and restrictive way to wrangle data and write code. The biggest game-changer to come along is MapReduce, the parallel programming framework that has gained prominence thanks to its use at web search companies.

To understand where we’re headed with parallel software, let’s look at what the computer industry has already accomplished. The branch of parallel research that has had the most success in the field is parallel databases. Rather than requiring the programmer to unravel an algorithm into separate threads to be run on separate cores, parallel databases let them chop up the input data tables into pieces, and pump each piece through the same single-machine program on each processor. This “parallel dataflow” model makes programming a parallel machine as easy as programming a single machine. And it works on “shared-nothing” clusters of computers in a data center: The machines involved can communicate via simple streams of data messages, without a need for an expensive shared RAM or disk infrastructure.

The MapReduce programming model has turned a new page in the parallelism story. In the late 1990s, pioneering web search companies built new parallel software infrastructure to manage web crawls and indexes. As part of this effort, they were forced to reinvent parallel databases –- in large part because the commercial database products at the time did not handle their workload well. Like SQL, the MapReduce framework is a parallel dataflow system that works by partitioning data across machines, each of which runs the same single-node logic.

SQL provides a higher-level language that is more flexible and optimizable, but less familiar to many programmers. MapReduce largely asks programmers to write traditional code, in languages like C, Java, Python and Perl. In addition to its familiar syntax, MapReduce allows programs to be written to and read from traditional files in a filesystem, rather than requiring database schema definitions. MapReduce is such a compelling entryway into parallel programming that it is being used to nurture a new generation of parallel programmers. Every Berkeley computer science undergraduate now learns MapReduce, and other schools have undertaken similar programs. Industry is eagerly supporting these efforts.

Technically speaking, SQL has some advantages over MapReduce, including natural combinations of multiple data sets, and the opportunity for deep code analysis and just-in-time query optimizations. In that context, one of the most exciting developments on the scene is the emergence of platforms that provide both SQL and MapReduce interfaces within a single runtime environment. These are especially useful when they support parallel access to both database tables and filesystem files from either language. Examples of these frameworks include the commercial Greenplum system (which provides all of the above), the commercial Aster Data system (which provides SQL and MapReduce over database tables), and the open-source Hive framework from Facebook (which provides a SQL-like language over files, layered on the open-source Hadoop MapReduce engine.)

MapReduce has brought a new wave of excited, bright developers to the challenge of writing parallel programs against Big Data. This is critical: a revolution in parallel software development can only be achieved by a broad base of enthusiastic, productive programmers. The new combined platforms for data parallelism expand the options for these programmers and should foster synergies between the SQL and MapReduce communities. Longer term, these Big Data approaches to parallelism might provide the key to keeping other sectors of the software industry on track with Moore’s Law.

Joe Hellerstein is a professor of Computer Science at the University of California Berkeley and has written a white paper with more detail on this topic.

Slide courtesy of Green Plum

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• A near-term outlook for big data
• Real-­time query for Hadoop democratizes access to big data analytics
• The importance of putting the U and I in visualization

Like many changes in computer science, the rapid shift toward parallel computing is a function of technology trends in hardware. Most technology watchers are familiar with Moore’s Law, and the more general notion that computing performance doubles about every 18-24 months. This continues to hold for disk and RAM storage sizes, but a very different story has unfolded for CPUs in recent years, and it is changing the balance of power in computing — probably for good.

What Moore’s Law predicts, specifically, is the number of transistors that can be placed on an integrated circuit. Until recently, these extra transistors had been used to increase CPU speed. But, in recent years, limits on heat and power dissipation have prevented computer architects from continuing this trend. Basically, CPUs are not getting much faster. Instead, the extra transistors from Moore’s Law are being used to pack more CPUs into each chip.

Most computers being sold today have a single chip containing between two and eight processor “cores.” In the short term, this still seems to make our existing software go faster: one core can run operating systems utilities, another can run the currently active application, another can drive the display, and so on. But remember, Moore’s Law continues doubling every 18 months. That means your laptop in nine years will have 128 processors, and a typical corporate rack of 40-odd computers will have something in the neighborhood of 20,000 cores.

Parallel software should, in principle, take advantage not only of the hundreds of processors per machine, but of the entire rack — even an entire data center of machines. Since individual cores will not get appreciably faster, we need massively parallel software that can scale up with the increasing number of cores, or we will effectively drop off of the exponential growth curve of Moore’s Law. Unfortunately, the large majority of today’s software is written for a single processor, and there is no technique known to “auto-parallelize” these programs.

Worse yet, this is not just a legacy software problem. Programmers still find it notoriously difficult to reason about multiple, simultaneous tasks in the parallel model, which is much harder for the human brain to grasp than writing “plain old” algorithms. So this is a problem that threatens to plague even new, greenfield software projects.

There are some precedents for how to overcome this problem. Over the past 20 years, the main bright spot in parallel software development has been in high-volume data analysis. SQL has been a successful massively parallel programming language since the late 1980′s. Many legacy SQL programs parallelize naturally, and every SQL programmer continues to write inherently parallel code.

Unfortunately, SQL represents a tightly scoped (albeit critical) corner of the software industry. But in recent years, a new ecosystem of data-intensive parallel development has been growing around the MapReduce parallel programming framework, which allows programmers to write data-parallel code in familiar languages like C, Java, Python and Perl. In my next post, I’ll talk about how the lessons of SQL and the growing excitement about MapReduce may bring parallelism to a larger swath of the software market.

Joe Hellerstein is a professor of Computer Science at the University of California Berkeley and has written a white paper with more detail on this topic.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• A near-term outlook for big data
• Real-­time query for Hadoop democratizes access to big data analytics
• The importance of putting the U and I in visualization