Supercomputing News
CAASTRO is using some of the most powerful computers in the world to process the torrents of data coming from our innovative wide-field telescopes.
Excerpts from the article 'iVEC Selects SGI to Accelerate Radio Astronomy Research' on benzinga.com, 14 December 2011
The Pawsey Centre, established as a result of the initiative, will be a petascale supercomputing facility managed and operated by iVEC, and will support radio astronomy research objectives.
The SGI Rackable system selected consists of ten water chilled racks and two storage racks. There are a total of one hundred cluster nodes, with 96 containing Nvidia® C2075 GPUs plus four custom server builds each containing Nvidia M2090 GPUs. The solution also includes 192 Intel® Xeon® X5650 2.66GHz 6-core processors, as well as eight Intel® Xeon® X5680 3.33GHz 6-core processors and 7.95 terabytes of total memory. Working with DataDirect Networks, SGI also provided 900 terabytes of the DDN-powered SGI® Infinite Storage 16000 arrays utilizing the Lustre® filesystem.
Excerpts from the article 'Swinburne's gSTAR heralds 'mega science'' in the Swinburne Magazine, 12 December 2011
Now, a new 120-teraflop supercomputer driven by graphic processing units (GPUs) is set to accelerate the rate of science and knowledge discovery at Swinburne, crunching data 10 times faster than the Green Machine. It means the equivalent of the diamond planet search could potentially be done in a week.Swinburne Pro-Vice-Chancellor for Research Matthew Bailes says the new GPU Supercomputer for Theoretical Astrophysics Research (gSTAR) is among the top six supercomputers in Australia, and will probably rank in the world’s top 200 machines.
Professor Bailes says the surge in processing power since Swinburne invested $1 million in the Green Machine in 2007 shows how quickly the power balance shifts between supercomputers.“gSTAR can perform more operations in a few seconds than the earliest supercomputers could conduct over their lifetime,” he says.
The system is supported by a quad data rate (QDR) InfiniBand network that transfers data to GPUs and large memory nodes, each with 512 gigabytes of memory, at 1000 times the speed of wireless internet connection.
Excerpts from the article 'Big Blue prototypes software for big, big data' in The Register, 4 December 2011
With as much as Exabyte of data as its daily dump, the SKA will demand new techniques just so astronomers can use the facility’s output.
Working with New Zealand-based radio astronomer, Dr Melanie Johnston-Hollitt from Wellington’s Victoria University, IBM has created the Information Intensive Framework prototype.
Under the framework, data will be classified into astronomical concepts, and overlaid with a guided search facility for faster data access and fewer errors. IBM says the prototype has also suggested further improvements to achieve the SKA’s performance demands.
"Undertaking research on exa-scale datasets will force radio astronomers into a new, as yet, unexplored regime of automated processing, imaging and analysis,” Dr Johnston-Hollitt said in IBM’s announcement.
“Surveys on even SKA precursor telescopes such as ASKAP and MWA are expected to produce catalogues of tens of millions of radio sources. How we organise and classify these data, which we will have in the next three years, is a significant challenge. We will need new solutions to fully realize the vast scientific potential of these datasets and it's fantastic that organisations like IBM are prepared to take up that challenge.”
If the A/NZ team wins the SKA contract, data will have to be pre-processed close to the telescopes (most of which would be in the remote north-west of Western Australia), then sent back to the Pawsey Supercomputing Centre in Perth for storage and analysis.
Excerpts from the article 'CSIRO opens tender for Pawsey petascale powerhouse' in
The Register, 21 November 2011:
Vendors to beat a path to Perth
The CSIRO has opened the tender for petascale iron for the Pawsey Centre, and is conducting an industry briefing in Perth on November 30.
Listed in the RFP, which is open until 6 January 2012, are a petascale supercomputer and realtime computer; HSM storage and tape library; networking, including a firewall, border router, and on-site Ethernet fabric; a visualization engine; virtual machines for data analysis; and systems integration.
The build will be a significant ramp-up of the Perth-based supercomputing centre, which already boasts a “pathfinder” project using HP ProLiant Blades running 9,600 Intel Xeon 5600 cores to deliver 87.02 TeraFLOPS.
Construction is due to start on the Pawsey Centre buildings early next year, with “white space” available for hardware installation by next November. CSIRO says 2.3 MW will be available for the supercomputer cell, 425 kW for I/O, and 25 kW for the tape cell.
Key applications for the new petascale machine will be processing and storing SKA Pathfinder (ASKAP) radio telescope data.
The realtime computing (RTC) system, to be delivered first, will need to deliver 200 double-precision teraflops peak performance, 45 Tbytes/s aggregate memory bandwidth, must use conventional CPUs (not graphics accelerators), with at least 1 Pbyte of usable storage. I/O to the scratch file system will need to run at 5 Gbytes/s in and out, and the system will have to be able to deliver 10,000 metadata operations per second on a user’s “home” file system.
Eighteen “datamover” nodes will have to demonstrate 180 Gbps network bandwidth and support 5 Gbytes/s “data ingest”. In April 2014, CSRIO wants a phase two system that delivers 1,000 DP teraflops against the Linpack benchmark
Excerpts from the article 'Behind iVEC's 'big science' supercomputer' in The Register,
10 October 2011:
Putting Australia’s West on the map
Each of its 96 nodes has two six-core Intel Xeon X5650s, one NVIDIA Tesla C2050 GPU, and 48 GB of RAM, but the SGI “Fornax” supercomputer opened late September as part of Western Australia’s Pawsey Centre project is still a test bed in some ways.
Located at the University of Western Australia, the iVEC@UWA machine is being managed and operated by iVEC, and is the second pathfinder in the Pawsey project (the other being iVEC@Murdoch, which came oneline last year). Among other things, the Pawsey Centre would ultimately operate the supercomputing facilities that will be built if Australia’s bid for the Square Kilometer Array astronomy project is successful.
However, while astronomy is the big national attention-grabber, the Pawsey Centre systems architect Guy Robinson is at pains to emphasise that Fornax supercomputer will carry workloads for a host of different scientific ventures, including geosciences, biosciences, materials science, chemistry, meteorology and climate science.
Astronomers, for example, need to be able to stream data into the centre at very high rates – Robinson spoke of 40 Gbps streams coming from each individual instrument in a radioastronomy array – and that has to be handled without disrupting the other users in the centre.
When the data has been pre-processed and stored – because, as Professor Bryan Gaensler of the Centre for All-Sky Astrophysics told The Register, the data raw has to be filtered or there would be too much to store – astronomers typically work with 6 TB files. “The signal rate coming down each individual telescope – those numbers are so frightening that I haven’t memorized them!” Robinson said.
Excerpts from the article 'Just-in-time processing power growth predicted to handle SKA demand' in Computerworld, 3 November 2011:
ASKAP will consist of an array of 36 radio dishes, each 12 metres in diameter, and is being built at the Murchison radio astronomy site in Western Australia. It is scheduled for completion in 2013.
Crucial to growing the processing power is not only the capacity of hardware, but the ability to co-ordinate the processing of many clustered nodes each with a large number of processing cores.
The consensus view of the exaflop machine is a processor running at 1 GHz in each of a thousand cores in each node of million in a cluster. Clusters of 262,000 nodes are within the capacity of current technology, “so the meganode is not that scary,” says Tim Cornwell “but the kilocore is really scary.”
If you extrapolate the Moore’s Law increase in the processing capacity of the most powerful supercomputers now available, he says, “the SKA demand will meet the processing trend line in 2022 when the SKA itself is due to be built.”
The algorithms needed to interpret the raw flow of data and deduce what objects are out there are also being redeveloped continually.
