NEWPORT NEWS, VA. – Work on computing the behaviors of the smallest bits of matter in the universe at the U.S. Department of Energy's Thomas Jefferson National Accelerator Facility has just gotten a nod from Intel®, as the laboratory becomes the newest Intel® Parallel Computing Center. Jefferson Lab has also just installed its newest parallel computing cluster, featuring Intel® architecture, which is set to go into production in October.
“Being named an Intel Parallel Computing Center establishes us as a center of excellence regarding parallel, high-performance code on Intel’s architectures,” said Chip Watson, leader of Jefferson Lab’s High Performance Computing group and an IPCC principal investigator. “This announcement is a recognition of our activity both in terms of producing modernized code and in terms of our outreach and leadership in the community in this area.”
Jefferson Lab conducts discovery-caliber research in exploring the atomic nucleus and its fundamental constituents, such as protons and neutrons and these particles’ building blocks: quarks and gluons. The theory of how these particles interact -- Quantum Chromodynamics, or QCD -- is far too complex to solve with pencil on paper, so the researchers have turned to supercomputers to compute it in the form of lattice QCD, or LQCD.
The main thrust of the work involving Intel® is the continuing development of the Chroma code, which is a code that is used to solve LQCD. In collaboration with Intel®, the researchers have been working on developing and improving an open source library of equation solvers for Chroma and on optimizing the performance of Chroma specifically on the Xeon Phi™ processor platform.
“We have been working with Intel’s parallel computing labs for a long time on writing high-performance code for Intel Xeon Phi libraries. The purpose of this center is to continue modernizing the freely available Chroma LQCD code, so that it will run well on the new generations of hardware,” said Bálint Joó, a member of Jefferson Lab’s High Performance Computing group and a principal investigator for the IPCC.
The high-performance code will soon be put to the test on Jefferson Lab’s newest parallel computing cluster, SciPhi-XVI. The new cluster features 200 nodes of the second-generation Xeon Phi™ chips, which were released in 2016 and are codenamed Knights Landing. Each node has 64 computing cores, and individual nodes within the machine are connected via the Intel® Omni-Path Architecture, which supports a 100 Gigabit/second data transfer rate between the nodes.
“Becoming an IPCC recognizes the close collaboration between Jefferson Lab and Intel, and the designation recognizes our role as pathfinders in using Intel’s next-generation architectures,” said Amber Boehnlein, Jefferson Lab’s Chief Information Officer. “Our collaboration ensures that the Chroma code will run faster: providing better results, more quickly. And this is essential to compare the best possible theoretical predictions to new experimental results, a cornerstone of the Jefferson Lab 12 GeV Physics Program.”
The SciPhi-XVI machine was delivered and installed in mid-August and is currently undergoing beta testing in preparation for full deployment October 1.
“Now that our newest machine has arrived, we’re excited about rolling it out for lattice QCD calculations using the Chroma code,” explained Watson. “We’re also excited to be exploring the use of these chips for analyzing the data that is generated in Jefferson Lab’s experimental physics program.”
Watson and Joó’s work on Chroma is funded through the U.S. Department of Energy Office of Science. The research is carried out in collaboration with partner institutions, such as DOE’s National Energy Research Scientific Computing Center (NERSC) Exascale Science Application Program (NESAP) and the Scientific Discovery through Advanced Computing (SciDAC) program. NERSC is a DOE Office of Science User Facility at Lawrence Berkeley National Laboratory.
The scientists are sharing the fruits of their work by continuing to make the code optimizations they have generated for Chroma publicly available, benefiting the worldwide community of scientists who use the code in their research. They also anticipate seeking out new collaborations in expanding their outreach to the next generation of computational scientists.
Contact: Kandice Carter, 757-269-7263, firstname.lastname@example.org