Since the NESAP programme was unveiled in 2014, NERSC has been partnering with code teams and library and tool developers to prepare and optimize their codes for the Cori manycore architecture. Like NESAP, the NESAP for Data programme joins application teams with resources at NERSC, Cray and Intel. However, while the initial NESAP projects involve mostly simulation codes, NESAP for Data targets science applications that process and analyze massive datasets acquired from U.S. Department of Energy-supported experimental and observational sources, such as telescopes, microscopes, genome sequencers, light sources and particle physics detectors. The goal is to enable these applications to take full advantage of the Intel Xeon Phi Knights Landing (KNL) chipset on Cori.
The selected NESAP for Data projects are:
"We're very excited to welcome these new data-intensive science application teams to NESAP", stated Rollin Thomas, a Big Data architect in NERSC's Data Analytics and Services group who is coordinating NESAP for Data. "NESAP's tools and expertise should help accelerate the transition of these data science codes to KNL. But I'm also looking forward to uncovering and understanding the new performance and scalability challenges that are sure to arise along the way."
Through NESAP, the teams will have full access to Cori-KNL, plus NERSC Data Department expertise, testbeds and collaborations with vendors. Participants are already lining up to start submitting jobs to the KNL debug queues, according to Rollin Thomas. NESAP for Data participants will also be included in the various NESAP meetings and events, such as the upcoming dungeon session to be held in March at Intel's Portland, Oregon facility.
Like the initial NESAP programme, NESAP for Data also includes post-doctoral opportunities; the first NESAP post-doc to start working on the data projects is Zahra Ronaghi, who joined the NESAP programme in January. NERSC is now in the process of reviewing more applications to fill two additional post-doc positions within NESAP for Data.
"We've learned a tremendous amount in the last couple years working with our existing NESAP teams to prepare their applications for Knights Landing", stated Jack Deslippe, acting group lead for NERSC's Application Performance group. "We've developed an optimization strategy that the greater NERSC community can use to prepare for Cori. It's really exciting to get a chance to bring in some new data-centric applications - to apply some of what we've learned already but also to learn more from the unique challenges these apps face."