December 6, 2013
Professor Kerry Key first got access to a parallel computer at UC San Diego (UCSD) 10 years ago, when he was a graduate student at Scripps Institution of Oceanography. His advisor bought a small cluster for their research program and Key was tasked with managing the system. However, without formal computer training, this task quickly became a huge time sink and disrupted his scientific research.
"TSCC has been a huge eye-opener, giving me easy access to a powerful parallel computer." — Prof. Kerry Key, SIO
Four years ago, when Key was a postdoctoral scholar at UCSD, he received an email from the university’s San Diego Supercomputer Center (SDSC) announcing the Triton Resource cluster. After enrolling for a free trial account, Key found it to be a huge benefit, since he could focus on using the cluster for research while SDSC staff was responsible for managing the system. Key migrated to the new Triton Shared Compute Cluster (TSCC) rolled out earlier this year as part of the UCSD Research Cyberinfrastructure (RCI) program.
"What we run overnight on TSCC would take almost a year if run on a desktop computer. It is just so dramatic how much work we can get done now."
Scientifically, Key’s research focuses on using electromagnetic waves to study seafloor geology. Using one of Scripps’ research vessels, Key’s team deploys an array of electromagnetic receivers onto the seabed. A transmitter is then towed close to the seabed while broadcasting low- frequency electromagnetic waves that diffuse through the geology and are recorded by the receivers. This data is used to create and refine geological models of the targeted sea floor area.
These datasets can be overwhelming, with hundreds of thousands of transmitter positions. Using TSCC, researchers can model the datasets on hundreds of processors simultaneously, with each processor modeling a different transmitter. Using a desktop computer for this task would take thousands of times longer.
Key and his students recently submitted a research paper to Nature, which would have been unimaginable without relying on TSCC to expedite the modeling process.
"I can even do my supercomputing tasks remotely!"
As a busy researcher who travels frequently, TSCC lets Key log into TSCC remotely while on the road or at sea. He can launch jobs, track progress, and download results from TSCC at airports, research vessels, and hotels using a simple WiFi connection or even mobile devices. TSCC has transformed the way Key performs his research, while greatly improving his efficiency.
"SDSC provides solid support and quick response. I would recommend TSCC to my cost-conscious colleagues."
Key compared the TSCC node price with commercial offerings and found that for the same budget, he could get a much better deal from TSCC when taking into account the software, hardware, facility, and SDSC’s support and expertise. TSCC researchers can enjoy these special rates because UC San Diego provides a subsidy to each faculty member using the RCI services. Key is happy to speak with anyone who is interested in learning more about the TSCC service for their own computing needs.
FIONA: innovative Big Data Network Appliance Read story...
View slides from
May 21, 2014
Download a new research publication on RCI Storage
Intel Presents Xeon Phi developer training event at SDSC on Thursday, May 1, 2014 (see details)
The TSCC received a switch hardware upgrade and a Lustre upgrade on Sunday, March 9 and Monday, March 10. Read the details...