BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20230831T095755Z
LOCATION:Schwarzhorn
DTSTART;TZID=Europe/Stockholm:20230628T140000
DTEND;TZID=Europe/Stockholm:20230628T160000
UID:submissions.pasc-conference.org_PASC23_sess160@linklings.com
SUMMARY:MS6D - Tuning Low-Overhead Node-Level Load Balancing for Computati
 onal Science and Engineering Simulations on Current and Emerging Supercomp
 uters
DESCRIPTION:Minisymposium\n\nInter-node dynamic load balancing speeds up s
 cientific simulations on a supercomputer. Given that nodes of emerging sup
 ercomputers have a much larger number and variety of processors than previ
 ous supercomputers, as well as the continued need for scalability of appli
 cations, intra-node load balancing is increasingly needed. Repurposing int
 er-node load balancing techniques for intra-node load balancing is not pra
 ctical due to heavyweight processor virtualization and because of the cost
 s incurred from load balancing itself, e.g., data movement from task migra
 tion, are more intricate and harder to predict. Node-level load balancing 
 requires novel low-cost load balancing strategies and empirical auto-tunin
 g of such strategies. This mini-symposium brings forward state-of-the-art 
 research involving such node-level load balancing from the level of (1) nu
 merical algorithms via the SLATE library, (2) adaptive and intelligent run
 time systems via Charm++’s node-level load balancing, (3) compilers via af
 finity and loop transformations in LLVM’s OpenMP, and (4) low-level runtim
 e libraries via discussion of the Argobots runtime system. By attending th
 is mini-symposium, one can expect lively discussions, research exchanges, 
 and ideas for future work between the audience and presenters on support f
 or node-level load balancing and its synergy with other techniques within 
 HPC software.\n\nLLVM’s OpenMP Loop Transformations to Enhance Locality in
  OpenMP Loop Scheduling\n\nThis talk will present OpenMP loop transformati
 ons, e.g., tiling, in a way that complements existing and novel dynamic sc
 hedules for OpenMP loops and OpenMP tasking.  It can be used for auto-tuni
 ng for obtaining the optimal composition of multiple loop transformations 
 and parameters without the cost...\n\n\nMichael Kruse (Argonne National La
 boratory)\n---------------------\nLow-Level HPC System Software for Task P
 arallelism\n\nThis talk will discuss the process of enhancing node-level d
 ynamic scheduling for MPI+X applications, expressed in higher-level models
  for task parallelism, such as OpenMP, and capable of targeting both CPU a
 nd GPU processors.  The main focus of the talk will be on low-level HPC sy
 stem software impl...\n\n\nVivek Sarkar (Georgia Institute of Technology)\
 n---------------------\nTunable Tasking in PARSEC for Numerical Linear Alg
 ebra Routines Used in HPC\n\nThis talk will show that a specialization of 
 node-level load balancing for each numerical linear algebra routine of the
  SLATE library is needed due to intricacies of each routine. In particular
 , it will show an auto-tuned low-overhead loop scheduling strategy with cu
 stomized data layouts speeds up d...\n\n\nGeorge Bosilca (University of Te
 nnessee)\n---------------------\nIntelligent Node-Level Locality-Aware Loa
 d Balancing and Optimizations with Charm++'s CkLoop\n\nIn this talk, we di
 scuss novel load balancers in Charm++/CharmMPI and the associated node-lev
 el library called CkLoop. We also will discuss how the design of these loa
 d balancers carefully re-uses intelligent heuristics from inter-node load 
 balancers of Charm++. Furthermore, we will show how inter- ...\n\n\nKavith
 a Chandrasekar (University of Illinois Urbana-Champaign) and Harshitha Men
 on (Lawrence Livermore National Laboratory)\n\nDomain: Computer Science, M
 achine Learning, and Applied Mathematics 
\n\nSession Chair: Vivek Kale (S
 andia National Laboratories)
END:VEVENT
END:VCALENDAR