Abstract
The Knights Landing (KNL) release of the Intel Many Integrated Core (MIC) Xeon Phi line of processors is a potential game changer for HEP computing. With 72 cores and deep vector registers, the KNL cards promise significant performance benefits for highly-parallel, compute-heavy applications. Cori, the newest supercomputer at the National Energy Research Scientific Computing Center (NERSC), was delivered to its users in two phases with the first phase online at the end of 2015 and the second phase now online at the end of 2016. Cori Phase 2 is based on the KNL architecture and contains over 9000 compute nodes with 96GB DDR4 memory. ATLAS simulation with the multithreaded Athena Framework (AthenaMT) is a good potential use-case for the KNL architecture and supercomputers like Cori. ATLAS simulation jobs have a high ratio of CPU computation to disk I/O and have been shown to scale well in multi-threading and across many nodes. In this paper we will give an overview of the ATLAS simulation application with details on its multi-threaded design. Then, we will present a performance analysis of the application on KNL devices and compare it to a traditional x86 platform to demonstrate the capabilities of the architecture and evaluate the benefits of utilizing KNL platforms like Cori for ATLAS production.
Highlights
In the multi-core computing era, processor chip trends such as increasing core multiplicity, decreasing memory per core, and increasing importance of vector processing are changing the way scientific software developers write efficient, scalable code
High energy physics (HEP) experiments such as ATLAS[1] are no exception to this paradigm shift
In order to ensure that ATLAS can efficiently utilize modern computing devices and devices of the future, a large campaign is underway to adopt a multi-threading concurrency model for parallelism and efficient use of memory resources[3][4]
Summary
In the multi-core computing era, processor chip trends such as increasing core multiplicity, decreasing memory per core, and increasing importance of vector processing are changing the way scientific software developers write efficient, scalable code. Modern computing devices such as Intel’s Xeon Phi line of many-core processors are good examples of what will be used more frequently in high performance computing facilities. These devices are best utilized with highlyparallel applications, so scientific computing models must adapt for greater concurrency and intelligent usage of memory resources.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
