NERSC Fields Its First Student Cluster Competition Team
Five former NERSC interns and one current intern have joined forces to participate in the Student Cluster Competition at ISC16, June 20-22 in Frankfurt, Germany, marking the first time that NERSC has fielded a student cluster competition team. The all-female team comprises two high schoolers and four college undergrads and is being guided by Rebecca Hartman-Baker, acting group lead for NERSC’s User Engagement Group. This is not Hartman-Baker's first time organizing a student cluster competition. She has also coached teams for the Student Cluster Competition at SC13 and SC14. Since being assembled last fall, the team—which is scattered all over the U.S. and internationally this summer—has been meeting weekly via the web to come up to speed on the competition and cluster requirements.
Preparing for a Hybrid Approach to Quantum Computing
Jarrod McClean and his Lawrence Berkeley National Laboratory colleagues want to predict the chemistry and properties of advanced compounds before scientists go into the lab to make them, but today's most powerful systems could take years to perform the calculations. Quantum computers could do the job, but it will be years before large-scale quantum computers are a reality. That's why McClean, a Luis Alvarez Fellow in Computing Science, and his colleagues in the lab’s Computational Chemistry, Materials and Climate (CCMC) Group are pursuing a third course: a high-performing hybrid.
Students from Oakland’s Life Academy Get a Glimpse of Scientific Computing
Ten students and two teachers from Oakland’s Life Academy spent the morning of Friday, June 3, learning about scientific supercomputing, what it takes to run a center like NERSC and the importance of high-speed networking. They also received a hands-on lesson in Arduino, an open-source electronics prototyping platform.
During the visit Jackie Scoggins of NERSC’s Operations Technology Group led a tour of the machine room and gave a presentation on how the operations group supports the systems and the scientists who use them. Kate Mace of ESnet’s Science Engagement Team gave a talk comparing “big data” on commercial apps to Big Data on research apps, pointing out how an entire day’s worth of Instagram posts is only a third as much data as that produced by one experiment at the ALS, and how networks underlie all data transfers.
Life Academy is a small public high school in Oakland emphasizing health and bioscience studies.
Hartman-Baker, Cook Kick-off Summer Student Talks
Rebecca Hartman-Baker and Brandon Cook, both of NERSC, will present the first talk of the 2016 summer student series followed by a machine room tour. Held at 10:30 a.m. this Friday, June 10 in Wang Hall (Building 59) room 4102, their talk will address "High-Performance Computing at NERSC." Closed-toe shoes are required to tour the machine room where NERSC's supercomputers are housed. For safety reasons, sandals or other open-toe shoes are not permitted on the machine room floor.
The series of weekly talks and events will continue through the 12-week program. Summer students are encouraged to attend and other CS staff are welcome.
Workshop on Algorithms for Modern Massive Data Sets
Registration closes June 12 for the 2016 Workshop on Algorithms for Modern Massive Data Sets (MMDS 2016). Co-sponsored by the Berkeley Institute for Data Science (BIDS), MMDS 2016 takes place June 21 - 24 on the campus of UC Berkeley and includes a four-day series of academic workshops addressing algorithmic and statistical challenges in modern large-scale data analysis.
The goals of this series of workshops are to explore novel techniques for modeling and analyzing massive, high-dimensional and non-linearly structured scientific and internet data sets and to bring together computer scientists, statisticians, mathematicians and data analysis practitioners to promote the cross-fertilization of ideas.
This Week's CS Seminars
Tuesday, June 7
Zettar ZX: A Multi-dimensional Scale-out Data Transfer Software Solution and Reference DTN Cluster Design
2:30 – 3:30 p.m., Wang Hall - Bldg. 59, Room 4016Les Cottrell and Chin Fang, SLAC National Accelerator Laboratory
We present high performance data transfer methods to facilitate current and future distributed data-intensive science projects. Drawing on examples from SLAC hosted projects as well as the oil & gas industry, we present future-facing data transfer concerns which require awareness to Terabit networking and motivate technology designs around these needs. Our effort covers the development and validation of a multi-dimensional scalable software solution (scaling in: available computation power, network interfaces, and IOPS), and a companion reference DTN cluster design. We present a cluster-oriented software architecture, supplemented with elements from peer-to-peer technology along with two patent pending methods of pre-transfer data processing. Using two proof of concept clusters we present benchmark results in moving sets of both small and large files across a 5000 mile 100Gbps WAN (provided by ESnet). Other key findings of this effort so far include the critical importance of using tiered HPC storage, with an all-NVMe SSD based top-tier, benefits of bypassing the CPU, kernel, and file systems as practically as possible. Several unique benefits of the solution are presented along with how it's being evaluated and extended. We describe a plan to use the BayExpress 400G network to demonstrate future possibilities in XFEL science. The success will pave the way for other distributed data-intensive science projects to leverage DOE Scientific User Facilities such as NERSC.
Link of the Week: Backyard Supercomputer Rock
When cartoon characters Phineas and Ferb decided to build a supercomputer (in their backyard) during season three of the Disney series, there had to be a song. And you have to hear it. (Thanks to Jackie Scoggins for the tip.)