China’s New Supercomputer is Tops

June 17, 2013

Today, at the opening session of the 2013 International Supercomputing Conference in Leipzig, Germany, the latest update (the 41st version) to the Top 500 list of supercomputers was announced, and a new Chinese system, the Tianhe-2, has taken first place honors.  The system achieved performance of 33.86 petaflops per second (3.386 × 1016 floating point operations per second) on the LINPACK benchmark; the Tianhe-2 (in English, Milky Way-2) will be deployed at the National Supercomputer Center in Guangzho, China, by the end of the year. The system has 16,000 nodes, each with multiple Intel processors, for a total of 3,120,000 processor cores.

The Titan system at Oak Ridge National Laboratory, ranked number 1 in the November, 2012 list, and the Sequoia system, at the Lawrence Livermore National Laboratory, previously ranked number 2, have both moved down one place, to number 2 and number 3, respectively.  The two system are still noteworthy as being among the most energy-efficient in use.  Titan delivers 2,143 Megaflops/Watt, and Sequoia 2,031.6 Megaflops/Watt.

The total capacity of the list has continued to grow quickly.

The last system on the newest list was listed at position 322 in the previous TOP500 just six months ago.  The total combined performance of all 500 systems has grown to 223 petaflop/sec, compared to 162 petaflop/sec six months ago and 123 petaflop/sec one year ago.

You can view the top ten systems, and the complete list, at this page.


A Supercomputer ARM Race?

May 28, 2013

The PC World site has a report of an interesting presentation made at the EDAworkshop13 in Dresden, Germany, this month, on possible future trends in the high-performance computing [HPC] market.  The work, by a team of researchers from the Barcelona Supercomputing Center in Spain, suggests that we may soon see a shift in HPC architecture, away from the commodity x86 chips common today, and toward the simpler processors (e.g., those from ARM) used in smart phones and other mobile devices.

Looking at historical trends and performance benchmarks, a team of researchers in Spain have concluded that smartphone chips could one day replace the more expensive and power-hungry x86 processors used in most of the world’s top supercomputers.

The presentation material is available here [PDF].  (Although PC World calls it “a paper”, it is a set of presentation slides.)

As the team points out, significant architectural shifts have occurred before in the HPC market.  Originally, most supercomputers employed special purpose vector processors, which could operate on multiple data items simultaneously.  (The machines built by Cray Research are prime examples of this approach.)  The first Top 500 list, published in June 1993, was dominated by vector architectures  — notice how many systems are from Cray, or from Thinking Machines, another vendor of similar systems.  These systems tended to be voracious consumers of electricity; many of them required special facilities, like cooling with chilled water.

Within a few years, though, the approach had begun to change.  A lively market had developed in personal UNIX workstations, using RISC processors, provided by vendors such as Sun Microsystems, IBM, and HP.   (In the early 1990s, our firm, and many others in the financial industry, used these machines extensively.)  The resulting availability of commodity CPUs made building HPC system using those processors economically attractive.  They were not quite as fast as the vector processors, but they were a lot cheaper.  Slightly later on, a similar transition, also motivated by economics, took place away from RISC processors and toward the x86 processors used in the by-then ubiquitous PC.

Top 500 Architectures

Top 500 Processor Architectures

The researchers point out that current mobile processors have some limitations for this new role:

  • The CPUs are mostly 32-bit designs, limiting the amount of usable memory
  • Most lack support for error-correcting memory
  • Most use non-standard I/O interfaces
  • Their thermal engineering does not necessarily accommodate continuous full-power operation

But, as they also point out, these are implementation decisions made for business reasons, not insurmountable technical problems.  They predict that newer designs will be offered that will remove these limitations.

This seems to me a reasonable prediction. Using more simple components in parallel has often been a sensible alternative to more powerful, complex systems.  Even back in the RISC workstation days, in the early 1990s, we were running large simulation problems at night, using our network of 100+ Sun workstations as a massively parallel computer.  The trend in the Top 500 lists is clear; we have even seen a small supercomputer built using Raspberry Pi computers and Legos.  Nature seems to favor this approach, too; our individual neurons are not particularly powerful, but we have a lot of them.


First Petaflop Computer to be Retired

March 31, 2013

I’ve posted notes here about the Top500 project, which publishes a semi-annual list of the world’s fastest computer systems, most recently following the last update to the list, in November 2012.

An article at Ars Technica reports that the IBM Roadrunner system, located at the US Department of Energy’s Los Alamos National Laboratory, will be decommissioned and, ultimately, dismantled.  The Roadrunner was the first system whose performance exceeded a petaflop (1 petaflop = 1 × 1015 floating point operations per  second).  It held the number one position on the Top 500 list from June, 2008 through June 2009; it was still ranked number two in November, 2009.  The Roadrunner system contained 122,400 processor cores in 296 racks, covering about 6,000 square feet.  It was one of the first supercomputer systems to use a hybrid processing architecture, employing both IBM PowerXCell 8i CPUs  and AMD Opteron dual-core processors

The system is being retired, not because it is too slow, but because its appetite for electricity is too big.   In the November 2012 Top 500 list, Roadrunner is ranked at number 22, delivering 1.042 petaflops and consuming 2,345 kilowatts of electricity.  The system ranked as number 21, a bit faster at 1.043 petaflops, required less than half the power, at 1,177 kilowatts.

It will be interesting to see how the list shapes up in June, the next regular update.


Watson Goes to College

March 9, 2013

Back in early 2011, I wrote a number of posts here about IBM’s Watson system, which scored a convincing victory over human champions in the long-running TV game show, Jeopardy!.   Since then, IBM with its partners has launched efforts to employ Watson in a variety of other fields, including marketing, financial services and medical diagnosis, in which Watson’s ability to assimilate a large body of information from natural language sources can be put to good use.

Now, according to a post on the Gigaom blog, Watson will, in a sense, return to its roots in computer science research.  IBM has supplied a Watson system to the Rensselaer Polytechnic Institute [RPI] in Troy, NY.  According to Professor James Hendler, author of the post, and head of the Computer Science department at RPI, one focus of the work with Watson will be expanding the scope of information sources the system can use.

One of our first goals is to explore how Watson can be used in the big data context.  As an example, in the research group I run, we have collected information about more than one million datasets that have been released by governments around the world. We’re going to see what it takes to get Watson to answer questions such as “What datasets are available that talk about crop failures in the Horn of Africa?”.

Some of the research work with Watson will also be aimed at gaining more understanding of the process of cognition, and the interplay of a large memory and sophisticated processing.

By exploring how Watson’s memory functions as part of a more complex problem solver, we may learn more about how our own minds work. To this end, my colleague Selmer Bringsjord, head of the Cognitive Science Department, and his students, will explore how adding a reasoning component to Watson’s memory-based question-answering could let it do more powerful things.

The Watson system is being provided to RPI as part of a Shared University Research Award granted by IBM Research.  It will have approximately the same capacity as the system used for Jeopardy!, and will be able to support ~20 simultaneous users.  It will be fascinating to see what comes out of this research.

The original IBM press release is here; it includes a brief video from Prof. Hendler.


Happy Birthday, Raspberry Pi

March 6, 2013

I’ve posted here several times before about the Raspberry Pi single-board Linux computer, developed by the Raspberry Pi Foundation. This past Friday, March 1, marked the one-year anniversary of the market introduction of the Pi.  (OK, if you want to be picky, it went on sale on February 29.)  Sales of the diminutive computer have greatly exceeded anyone’s expectations, with more than one million units sold in the first year.  As Liz writes on the Raspberry Pi Foundation’s blog,

It’s been a crazy, wonderful year, and usually I’d have a lot to say about it. We never thought we’d find ourselves in the position we’re in today, with a million Pis sold, a sprawling community, real evidence that kids are picking the Pi up and learning with it, and new friends from all over the world.

But you hear from me all the time. So for today’s post I’ve asked members of the Pi family to share a few words with us about the way this year has looked to them instead.

The Raspberry Pi computer ships without an operating system installed; the customer needs to download an OS and store it on an SD memory card in order to make effective use of the device.  When the Pi was first introduced, the recommended OS was a version of the  Fedora Linux system.  That has now been supplanted by Raspbian Linux, a derivative of the highly-regarded Debian distribution. (A version of Arch Linux and a port of RISC OS are also available for the Pi.)

Raspbian is an impressive accomplishment, not least because the work was done, almost entirely, by two people: Mike Thompson and Peter Green.  Getting a suitable distribution built was a bit tricky.  The Raspberry Pi uses an ARMv6 procesor.  Debian has a port to the ARM architecture, but it is for a later version of the processor, the ARMv7.  In order to use the floating-point arithmetic hardware in the Pi’s processor, Thompson and Green had to rebuild ~19,000 packages, using an ARM cluster built in Thompson’s basement.  (There were also ~17,000 non-executable packages that did not need porting: help files, fonts, and man pages, for example.)  The full story of the Raspbian port is given in an article at Ars Technica.


More New Old Tech

January 19, 2013

I’ve mentioned here before some instances in which current technical problems have sometimes been amenable to old technologies, dusted off and updated a bit.  For example, there is the use of the venerable AC induction motor, patented by Nikola Tesla in 1888, in new electric vehicles, as well as the renewed interest in the use of DC power distribution for data centers.

Now, according to an article at Technology Review, another old technology, for a type of Diesel engine, is getting another look.  The basic design, called the Jumo engine, was originated back in the 1930s by Junkers, a German aircraft manufacturer.  It was dirty, but very efficient.  In contrast to a conventional Diesel engine, which uses a single piston per cylinder to compress air and fuel, the Jumo engine uses two pistons per cylinder, compressing the air-fuel mixture between them.  The efficiency advantage arises from expending less energy heating up the cylinder head, leaving more to drive the pistons.

A California company called Achates Power has updated the engine design to allow it to meet current emission standards, at least in a one-cylinder prototype.  The US Army has given Achates, together with a partner company, AVL Powertrain Engineering, a $4.9 million grant to develop a multi-cylinder prototype.  The company believes that the engine can be made smaller and cheaper than existing Diesel engines, while boosting fuel economy by 20%.  Compared to a gasoline engine, the fuel economy would of course look even better.

This is still a prototype, and the new design is not likely to make the 2014 model year for new cars.  Still, it is encouraging that progress can be made without requiring a “great leap forward” in every instance.


Requiring Black Boxes for Cars

January 5, 2013

Back in May of 2012, I wrote about the current and potential use of event data recorders [EDRs], so-called “black boxes”, in automobiles.  Similar devices have been used for years on commercial aircraft. and the data obtained from them has been of great value in understanding crashes and improving safety.   Many newer cars already have some kind of event recording device.  The Department of Transportation’s National Highway Traffic Safety Administration [NHTSA] has established some required standards for the amount and type of data that must be collected by installed EDRs; however, the installation of the devices was not required.

The NHTSA has now published a proposed regulation (Docket No. NHTSA-2012-0177) in the December 13, 2012 edition of the Federal Register (copies available as plain text or PDF) that would require the installation of EDRs in most autos and light trucks manufactured on or after September 1, 2014.  These devices would be required to meet the already existing standards for data collection.

The proposed regulation is open to public comment until February 11, 2013.  You can submit comments online using the docket page at the Regulations.gov web site (it also has a viewable copy of the rulemaking notice).  Alternatively, you can submit comments by mail or fax by following the instructions in the notice.  All submitted comments will become a matter of public record; online submissions can be viewed via the docket page.

There are legitimate privacy issues surrounding the collection of this data, and the ownership of the collected data needs to be clarified.  Still, there is a good case to be made, on safety grounds, for collecting the data; it should be possible to arrive at a reasonable trade-off.


%d bloggers like this: