BEGIN:VCALENDAR VERSION:2.0 PRODID:Linklings LLC BEGIN:VTIMEZONE TZID:America/Denver X-LIC-LOCATION:America/Denver BEGIN:DAYLIGHT TZOFFSETFROM:-0700 TZOFFSETTO:-0600 TZNAME:MDT DTSTART:19700308T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:-0600 TZOFFSETTO:-0700 TZNAME:MST DTSTART:19701101T020000 RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20260422T000615Z LOCATION:503-504 DTSTART;TZID=America/Denver:20231113T090000 DTEND;TZID=America/Denver:20231113T173000 UID:submissions.supercomputing.org_SC23_sess447@linklings.com SUMMARY:PMBS23: The 14th International Workshop on Performance Modeling, B enchmarking, and Simulation of High-Performance Computer Systems DESCRIPTION:Verifying Performance Guidelines for MPI Collectives at Scale\ n\nMPI collective communication operations are crucial for high-performanc e computing, making the efficient implementation of collective algorithms essential for optimal application performance. While most MPI libraries pr ovide several algorithms for a specific collective operation, each may wor k bette...\n\n\nSascha Hunold (Technical University of Vienna)\n---------- -----------\nPower Analysis of NERSC Production Workloads\n\nPower has bec ome a key limiting factor in supercomputing. Understanding the power sign atures of current production workloads is essential to address this limit and continue to advance scientific computing at scale. This paper analyzes the power characteristics of NERSC production workloads at the s...\n\n\n Zhengji Zhao, Ermal Rrapaj, Sridutt Bhalachandra, Brian Austin, Hai Ah Nam , and Nicholas Wright (Lawrence Berkeley National Laboratory (LBNL))\n---- -----------------\nEvaluating the Potential of Elastic Jobs in HPC Systems \n\nIt is generally assumed that elastic parallel applications, with the a bility to dynamically resize their process count, would provide numerous b enefits to High-Performance Computing (HPC) systems and applications. Sup porting this capability, however, requires significant effort at several l ayers of...\n\n\nDavid Eberius, Md. Wasi-ur- Rahman, and David Ozog (Intel Corporation)\n---------------------\nA Performance Model for Estimating t he Cost of Scaling to Practical Quantum Advantage\n\nWe present a simple p erformance model to estimate the qubit-count and runtime associated with l arge-scale error-corrected quantum computations. Our estimates extrapolate current usage costs of quantum computers and show that computing the grou nd state of the 2D Hubbard model, which is widely believe...\n\n\nDaan Cam ps, Katherine Klymko, Brian Austin, and Nicholas J. Wright (Lawrence Berke ley National Laboratory (LBNL))\n---------------------\nAdaptive Stopping Rule for Performance Measurements\n\nPerformance variability in complex co mputer systems is a major challenge for accurate benchmarking and performa nce characterization, especially for tightly-coupled large-scale high-perf ormance computing systems. Point summaries of performance may be both unin formative, if they do not capture the ful...\n\n\nViyom Mittal, Pedro Brue l, Dejan Milojicic, and Eitan Frachtenberg (Hewlett Packard Enterprise (HP E))\n---------------------\nRisk-Aware Scheduling Algorithms for Variable Capacity Resources\n\nThis work focuses on the design of scheduling algori thms for independent jobs that are submitted to a platform whose resource capacity varies over time. Jobs are submitted online and assigned on a tar get machine by the scheduler, which is agnostic to the rate and amount of resource variation. The ...\n\n\nLucas Perotin (ENS Lyon), Chaojie Zhang (Microsoft Research), Rajini Wijayawardana (University of Chicago), Anne Benoit and Yves Robert (ENS Lyon), and Andrew Chien (University of Chicago )\n---------------------\nSPEChpc 2021 Benchmarks on Ice Lake and Sapphire Rapids Infiniband Clusters: A Performance and Energy Case Study\n\nWe as sess fundamental performance, power, and energy characteristics of the SPE Chpc 2021 benchmark suite on two clusters based on Intel Ice Lake and Sapp hire Rapids CPUs using MPI only. We use memory bandwidth, data volume, and scalability metrics in order to categorize the benchmarks and pinpoint r. ..\n\n\nAyesha Afzal and Georg Hager (Friedrich-Alexander University, Erla ngen-Nuremberg; Erlangen National High Performance Computing Center) and G erhard Wellein (Friedrich-Alexander University, Erlangen-Nuremberg; Depart ment of Computer Science)\n---------------------\nReducing Memory Requirem ents for the IPU Using Butterfly Factorizations\n\nHigh Performance Comput ing (HPC) benefits from different improvements during last decades, specia lly in terms of hardware platforms to provide more processing power while maintaining the power consumption at a reasonable level. The Intelligence Processing Unit (IPU) is a new type of massively paral...\n\n\nSeyedKazem Shekofteh, Christian Alles, and Holger Fröning (Heidelberg University, In stitute of Computer Engineering (ZITI))\n---------------------\nPMBS23: Th e 14th International Workshop on Performance Modeling, Benchmarking, and S imulation of High-Performance Computer Systems\n\nThe PMBS23 workshop is c oncerned with the comparison of high-performance computing systems through performance modeling, benchmarking or through the use of tools such as si mulators. We are particularly interested in research which reports the abi lity to measure and make tradeoffs in software/hardwar...\n\n\nSteven Wrig ht (University of York), Simon Hammond (National Nuclear Security Administ ration (NNSA)), and Stephen Jarvis (University of Birmingham)\n----------- ----------\nPMBS23 – Afternoon Break\n---------------------\nLatency and B andwidth Microbenchmarks of US Department of Energy Systems in the June 20 23 Top 500 List\n\nAs a rule, Top 500 class supercomputers are extensively benchmarked as part of their acceptance testing process. However, barring publicly posted LINPACK / HPCG results, most benchmark results are often inaccessible outside the hosting institution. Moreover, these higher level benchmarks do not prov...\n\n\nChristopher M. Siefert, Carl Pearson, and Stephen L. Olivier (Sandia National Laboratories); Andrey Prokopenko (Oak Ridge National Laboratory (ORNL)); and Jonathan Hu and Timothy J. Fuller ( Sandia National Laboratories)\n---------------------\nPMBS23 – Morning Bre ak\n---------------------\nHardware Specialization: Estimating Monte Carl o Cross-Section Lookup Kernel Performance and Area\n\nHardware specializat ion is one of the promising directions in the post-Moore era. It is imper ative to understand how hardware specialization paradigms can benefit HPC. An essential question revolves around estimating the theoretical perform ance of an optimally specialized architecture without requ...\n\n\nKazutom o Yoshii, John Tramm, and Bryce Allen (Argonne National Laboratory (ANL)); Tomohiro Ueno and Kentaro Sano (RIKEN Center for Computational Science (R -CCS)); and Andrew Siegel and Pete Beckman (Argonne National Laboratory (A NL))\n---------------------\nPhysical Oscillator Model for Supercomputing\ n\nA parallel program together with the parallel hardware it runs on is no t only a vehicle to solve numerical problems, it is also a complex system with interesting dynamical behavior: resynchronization and desynchronizati on of parallel processes, propagating phases of idleness, and the peculiar effect...\n\n\nAyesha Afzal and Georg Hager (Friedrich-Alexander Universi ty, Erlangen-Nuremberg; Erlangen National High Performance Computing Cente r) and Gerhard Wellein (Friedrich-Alexander University, Erlangen-Nuremberg ; Department of Computer Science)\n---------------------\nPMBS23 – Welcome \n---------------------\nA Reinforcement Learning-Based Backfilling Strate gy for HPC Batch Jobs\n\nHPC systems employ a scheduling technique called “backfilling”, wherein low-priority jobs are scheduled earlier to use the available resources that are waiting for the pending high-priority jobs. B ackfilling relies on job runtime to calculate the start time of the ready- to-schedule jobs ...\n\n\nElliot Kolker-Hicks and Di Zhang (University of North Carolina at Charlotte) and Dong Dai (University of North Carolina, C harlotte)\n---------------------\nPMBS23 – Lunch Break\n------------------ ---\nComparative Evaluation of Bandwidth-Bound Applications on the Intel X eon CPU MAX Series\n\nWe explore the performance of Intel Xeon MAX CPU Ser ies, representing the most significant new variation upon the classical CP U architecture since the Xeon Phi. Given a large on-package high-bandwidth memory, the bandwidth-to-compute ratio has significantly shifted compared to other CPUs on the mark...\n\n\nIstván Z. Reguly (Pázmány Péter Catholi c University, Hungary)\n---------------------\nModeling Data Locality of S parse Matrix-Vector Multiplication on the A64FX\n\nOne of the novel featur es of the Fujitsu A64FX CPU is the sector cache. This feature enables hard ware-supported partitioning of the L1 and L2 caches and allows the program mer control of which partition is used to place data in. This paper perfor ms an in-depth study of how to apply the sector cache t...\n\n\nSergej-Ale xander Breiter (Ludwig-Maxmilians-Universität München), James D. Trotter ( Simula Research Laboratory), and Karl Fürlinger (Ludwig-Maxmilians-Univers ität München)\n\nTag: Modeling and Simulation, Performance Measurement, Mo deling, and Tools\n\nRegistration Category: Workshop Reg Pass\n\nSession C hairs: Simon Hammond (National Nuclear Security Administration (NNSA)); St ephen Jarvis (University of Birmingham, UK); and Steven A. Wright (Univers ity of York, England) END:VEVENT END:VCALENDAR