Slashdot Mirror
When Lofar Meets Stella
Roland Piquepaille writes "The LOFAR (Low Frequency Array) telescope is a new IT radio-telescope which will use about 20,000 simple radio antennae when it's completed in 2008. At this time, it will cover an area with a diameter of 360 kilometers centered over the Netherlands. Its small radio antennae will detect radio wavelengths up to 30 meters, and because the ionosphere can bend some of these radio waves, the Lofar images might be somewhat blurry. So all the information captured by these antennae will be digitized and sent to a computing facility at a rate of 22 terabits/second today, and almost 50 terabits/second in 2010. This is the reason why Lofar needs Stella, an IBM supercomputer installed recently in Groningen, also in the Netherlands, to process signals from up to 13 billion light years from Earth. Stella consists of 12,000 PowerPC microprocessors and has a computing power of 27.4 teraflops. This overview contains more details and a picture about the Lofar-Stella interaction."
Stella consists of 12,000 PowerPC microprocessors and has a computing power of 27.4 teraflops.
I love it when slashdot talks dirty.
When Lofar Meets Stella
The LOFAR (Low Frequency Array) telescope is a new IT radio-telescope which will use about 20,000 simple radio antennae when it's completed in 2008. At this time, it will cover an area with a diameter of 360 kilometers centered over the Netherlands. Its small radio antennae will detect radio wavelengths up to 30 meters, and because the ionosphere can bend some of these radio waves, the Lofar images might be somewhat blurry. So all the information captured by these antennae will be digitized and sent to a computing facility at a rate of 22 terabits/second today, and almost 50 terabits/second in 2010. This is the reason why Lofar needs Stella, an IBM supercomputer installed recently in Groningen, also in the Netherlands, to process signals from up to 13 billion light years from Earth. Stella consists of 12,000 PowerPC microprocessors and has a computing power of 27.4 teraflops. Read more...
Let's start with the opening paragraphs of an article from New Scientist, "Huge radio telescope boasts supercomputer brain."
One of the world's most powerful supercomputers is to be the brain of a revolutionary new radio telescope called LOFAR. The telescope will look back to the time of the very first stars, map our galaxy's magnetic field and perhaps discover the mysterious sources of high-energy cosmic rays.
Instead of one large rigid dish, LOFAR will use thousands of simple radio antennae. Their signals will be woven together at the University of Groningen in the Netherlands by STELLA, the new supercomputer, which was launched on Tuesday and is unofficially ranked as the third most powerful on the planet.
LOFAR needs its own supercomputer because it aims to detect radio wavelengths of up to 30 metres. Such long-wave radio images are blurry, and the only way to make them sharper is to build a vast array of detectors spread over hundreds of kilometres.
Now, let's move to the General Information section of the LOFAR website for more specific information.
LOFAR is the first telescope of this new sort, using an array of simple omni-directional antennas instead of mechanical signal processing with a dish antenna. The electronic signals from the antennas are digitised, transported to a central digital processor, and combined in software to emulate a conventional antenna. The cost is dominated by the cost of electronics and will follow Moore's law, becoming cheaper with time and allowing increasingly large telescopes to be built.
So LOFAR is an IT-telescope. The antennas are simple enough but there are a lot of them - 25000 in the full LOFAR design. To make radio pictures of the sky with adequate sharpness, these antennas are to be arranged in clusters that are spread out over an area of ultimately 350 km in diameter. (In phase 1 that is currently funded 15000 antenna's and maximum baselines of 100 km will be built).
Below is a general diagram of the LOFAR-STELLA interaction picked from the System section of the LOFAR website (Credit: LOFAR).
A diagram of the LOFAR environment
Details are scarce about the STELLA supercomputer, built by IBM using some of its Blue Gene/L technology. Reuters gave some information last week in "Europe's Biggest Supercomputer Eavesdrops on Stars."
Running on 12,000 PowerPC microprocessors, the computer can execute 27.4 Teraflops, or 27.4 trillion floating-point operations, per second.
The new computer will consume 150 Kilowatts of power -- the equivalent of 2,500 60-watt light bulbs -- which is considered economical for a supercomputer, IBM said.
If you understand Dutch, you also can read this news release about this supercomputer.
Now we have to wait to see if the happy couple of Lofar and Stella can produce images as beautiful as Hubble gave us during the last decade.
Sources: Various websites
I get pissed off.
P.S. stop posting prostoalex submissions, too.
Lofar: Stella? STELLA!!!
We recently had heard in the office over one of the Yellow Machine that's made by Anthology Solutions.
The blog is a bit misleading: "Details are scarce about the STELLA supercomputer, built by IBM using some of its Blue Gene/L technology."
Details are plenty since what IBM gave to the project is a couple of racks of BlueGene so everything applies, scaled proportionately.
Here are some details: http://www.research.ibm.com/bluegene/ and http://www.llnl.gov/asci/platforms/bluegenel/
This is an IBM BlueGene system. I went down to Rochester, MN to see one of these systems.. very interesting architecture.. each "node" is a dual core system on a chip.. the compute node OS is a simple non-multi-tasking kernel with a simple linux-ish libc.
You cross-compile your application on a power4/5 linux box, and then submit it to the system.. they reduced the computational aspect to it's most basic components.. CPU/FPU, memory, and MPI interconnect..
compute nodes don't even have ethernet or drive controlers... all I/O is handled by another specialized I/O node, which provides data over one of the two MPI interconnects.
They still have "Stella" shouting contests to honor the scene in A Streetcar Named Desire (1951),featuring Marlon Brando's plea to Kim Hunter as Stella Kowalski. Wonder if that has something to do with the choice of names for the computer.
Rapidweather's Linux Screenshots.
1) Roland Piquepaille. See one of the many comments on what this guy does.
2) The blurb was written by some third grader, wasn't it?
IT radio-telescope? What is IT? Radio is an adjective, there's no need to hyphenate radio telescope.
"At this time"? Now? I thought it won't be completed until 2008?
We detect radio waves, not wavelengths.
"the Lofar images might be somewhat blurry"? Images? Since when do we get images from radio telescopes?
They're obviously not sending data at 22 terabits/sec today, since the telescope hasn't even been constructed yet.
"This" is why it needs stella? Oh, the antecedent is yet to come in the sentence.
I am Lofar of the Dish People! Much have I have detected, and much have I computed, for I am Lofar of the Dish People!
It's a big mess-O-sensors spread over a wide area.
Radio, seismic, atmospheric pressure, and "other".
With the big iron computer it will be possible to play around with all sorts of spatio-temporal signal processing. This has been done with optical telescopes to remove 'twinkles', SA-RADAR and SA-SONAR, and most intensively in oil exploration where 2-D arrays of seismic sensors coupled with 2-D arrays of seismic sources are used in oil exploration. The neat thing, just liek in oil exploration, is that the data from the different sensors can be looked at for correlations. air-pressure, seismic, and radio data all recorded around a significant geophysical event. Yes, I knwo this is 'fishing science', bu tit is fun...
Now I'm the grandest Tiger in the Jungle!
LOFAR: I am LOFAR of the Hill People! Much have I have seen, and much have I done, for I am LOFAR of the Hill People!
We speak of many things! Detecting radio wavelengths! Fire! The weakness of women!
Sorry, That was the first thing I thought of when I read LOFAR meets stella
Does anyone else think it is strange to install a supercomputer years before this telescope gets deployed? Processors do get faster and faster all the time.
Even though this telescope will not be placed in the most densily populated area, the Netherlands is hardly the place to go looking for 350 KM^2 of land to put a radio telescope on. It's nice for business and science here, but putting it on a field somewhere in East Germany, Tsech Republic etc. might be more economic.
Anyway, I'm buying stock in the storage business.
It might sound impressive, but it's a stupid idea. The main reason they need a supercomputer in the first place is so they can try and remove the effect of the interference - but "taking it all out in software" is exceedingly difficult. Especially if the RFI gets so bad that it saturates the receiver front-ends.
LOFAR (my office mate worked on it) used to be an international collaboration, but it broke a apart because the Dutch insisted it be build in their country, rather than in some place more sensible, like Western Australia.
Human genome = 3 billion base pairs = 6 GBit. Windows + Office = 20 Gbit. Which is more impressive?
You'd really like to have a directional antenna to block out signals that are not coming from the sky.
That's why it's a 'phased array'. The concept is pretty simple: if a signal is coming from a certain direction, it will arrive at antenna A a fraction of a second earlier than at antenna B. LOFAR measures the incoming signals and their phase at all the antennae, and then lets software 'untangle' this information to reconstruct the direction of the different signals.
There's an awful lot of man-made noise and very strong signals in this frequency range.
This is true, but most of it is limited to certain frequency intervals. There are still enough other frequencies left to observe in.
One lousy piece of rusty fence wire can intermix all that crud and rebroadcast all kinds of sum and difference gobs of spurious signals.
Actually I think the wavelengths used here are large enough to just ignore rusty fence wire ;)
Fact is that this 'idea' is up and running, and it works!