This Week in HPC News
Last week, we noted the doubling in the number of research items coming from across the supercomputing spectrum and this week was no different. The uptick that we’ve seen this month in terms of new research and development efforts continues. The academic news was especially strong, with a number of Moore’s law themed items – see here, here and here.
Among the more noteworthy stories are a €700,000 HPC investment from a group of animal breeding researchers, who will use the supercomputer to perform calculations related to animal genetics as well as an innovative computational algorithm from researchers at Carnegie-Mellon University and the University of Maryland that significantly speeds up gene expression analysis. We also reported on the efforts of the HPC-SME to boost HPC adoption among the SME community in France.
Perhaps the most significant research-oriented story of the week came out of Texas Advanced Computing Center, where scientists have uncovered a link between Alzheimer’s disease and brain cancer that may pave the way for better treatment options and new medicines.
Moving on, let’s give a warm round of applause to all the students who participated in two separate student supercomputing challenges and all the supporting players who make these events possible. The 2014 Asia Student Supercomputer Challenge (ASC14) kicked off on Monday in Guangzhou, China. Held at the Sun Yat-sen University campus – home to the world’s fastest supercomputer Tianhe-2 – the event aims to motivate students to study supercomputing and improve students’ ability to tackle a range of real problems. Among the 16 teams who made it to the finals round is the Purdue University team, the Boilermakers, the only US team to do so. We’ll be reporting on the winners early next week, but just to make it into this event is a huge accomplishment.
But China was not the only venue with a student supercomputing challenge this week. The 24th annual New Mexico Supercomputing Challenge at the Los Alamos National Laboratory in Los Alamos, NM, welcomed more than 240 students (about 70 teams) from schools around the state. Open to any New Mexico high-school, middle-school or elementary school student, the learning event is geared to teaching a wide range of skills, including research, writing, teamwork, time management, oral presentations and computer programming. The top honors went to Eli Echt-Wilson and Albert Zuo from Albuquerque’s La Cueva High School for their computational model of deciduous tree growth with an emphasis on maximizing carbon sequestration.
Top News Items This Week
The big vendor story this week was IBM’s unveiling of POWER8, the next generation of its Power Systems architecture and servers. Although the servers are being targeted at what the firm is calling “an era of big data” their CAPI interconnect has some crying HPC. The new servers are the first physical products emerging from the OpenPOWER foundation, an industry alliance formed by IBM, Mellanox Technologies, NVIDIA, and Tyan, that makes POWER hardware and software available for open development, as well as POWER intellectual property licensable to other manufacturers. The group, which launched nine months ago, now has 25 members.
Altera becomes the first programmable logic company to integrate hardened IEEE 754-compliant, floating-point operators in an FPGA, delivering higher levels of DSP performance, designer productivity and logic efficiency. The hardened floating point DSP blocks are integrated in Altera’s 20 nm Arria 10 FPGAs and SoCs – currently shipping – as well as 14 nm Stratix 10 FPGAs and SoCs. The company says: “integrated hardened floating-point DSP blocks, combined with an advanced high-level tool flow, enable customers to use Altera’s FPGAs and SoCs to address an expanding range of computationally intensive applications, such as high-performance computing (HPC), radar, scientific and medical imaging.” Will this be the tipping point for FPGAs in HPC?
The Holland Computing Center HPC support team at the University of Nebraska have adopted Allinea Software’s unified debugging and profiling tools across four clusters – a total of around 20,000 compute cores – for the University of Nebraska campuses in Lincoln, Omaha and Kearney. HPC Applications Specialist, Adam Caprez, needed a solution that would free up user time. “The tools have made it so much easier to debug MPI or multi-threaded programs when someone comes to me with an issue,” he says. “I recently used Allinea DDT to help a faculty member to debug some MPI code that he’d inherited from another institution. Allinea DDT really helped us to narrow it down and identify what the esoteric problem was. It was really useful.”
The National Center for Atmospheric Research (NCAR) and the University of Colorado Boulder (CU-Boulder) announced their intent to join the Intel Parallel Computing Centers program. The program is part of the research and academic work that Intel supports in the quest for increasing efficiency and optimization of parallel microprocessor computer architectures. Participants work on developing methods to increase the performance of applications that use advanced microprocessor technologies and help train the next generation of scientists and engineers who will apply these new technologies to challenges of societal importance.