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:20191206T142452Z
LOCATION:Analog 1\, 2
DTSTART;TZID=Europe/Stockholm:20180625T134500
DTEND;TZID=Europe/Stockholm:20180625T134900
UID:isc_hpc_ISC High Performance 2018_sess103_phd109@linklings.com
SUMMARY:(PhD12) From Molecular Dynamics towards a Node-Level Auto-Tuning L
 ibrary for N-Body Simulations
DESCRIPTION:PhD Forum\n\n(PhD12) From Molecular Dynamics towards a Node-Le
 vel Auto-Tuning Library for N-Body Simulations\n\nGratl, Tchipev, Seckler,
  Neumann, Bungartz\n\nMolecular Dynamics simulations are tools of great po
 tential for fields like Chemical or Biological Engineering. From a Compute
 r Science point of view, building highly scalable and versatile simulation
  codes proves to be an interesting but challenging task. Due to the nature
  of the problem, system characteristics have a drastic impact on the time 
 to solution, node-level performance and can influence runtime by up to ord
 ers of magnitude. To make matters worse these characteristics can also cha
 nge at runtime, thus requiring adapting solution strategies during runtime
 .\n\nOver the last years, ls1-mardyn was developed at the Chair of Scienti
 fic Computing in Computer Science at the Technical University of Munich. T
 his is a highly scalable code for molecular dynamics simulations of small 
 rigid multi-centered molecules. The code is developed in C++ with an effic
 ient vectorization and a hybrid OpenMP + MPI parallelization which was sho
 wn to be capable of excellent scaling up to thousands of nodes. ls1-mardyn
  also features a reduced memory mode through which it was possible to perf
 orm the largest particle simulation to date, containing about 2 * 10^13 mo
 lecules.\n\nIn every particle simulation short range, pairwise interaction
 s make up a significant portion of the overall runtime. While developing l
 s1-mardyn it became obvious that there is no silver bullet to provide high
 ly scalable and efficient approaches for every scenario. As an example, a 
 decision needs to be made concerning the underlying algorithm of a simulat
 ion. Classic choices are either Verlet lists, which provide good computati
 onal efficiency, or Linked Cells which excel at memory efficiency. Also, t
 ree structures provide interesting options. Further aspects that are diffi
 cult to get efficient are SIMD vectorization (data layout, intrinsics vs c
 ompiler generated), OpenMP patterns or optimization for accelerators like 
 Xeon Phi. For this reason, we want to start the open source project AutoPa
 s: a node-level library for short-range pairwise interactions. The vision 
 for the library is to be easy to use and to provide a powerful general-pur
 pose base to build full N-Body simulations on top of it. The library aims 
 to be highly flexible by providing different particle containers, OpenMP p
 arallelization patterns, and vectorization options.\n\nTwo of the core req
 uirements to achieve the goals are a modular code design and auto-tuning. 
 The first is necessary since different scenarios benefit from dedicated so
 lution techniques, so for optimal flexibility, these need to be exchangeab
 le in an easy way. The optimal combination of these techniques depends on 
 many characteristics and can change at runtime. Furthermore there exist to
 o many combinations of techniques to just test all in order to find the op
 timum empirically. Auto-tuning enables the code to find the optimal combin
 ation of aforementioned techniques on its own and adapt itself during runt
 ime by periodically reevaluating the current state of the simulation. One 
 approach to this is to create performance models for characteristics that 
 define the scenario and empirically compare the combinations the performan
 ce model predicted to be the most efficient.\n\nOur current research focus
 es on expanding the library and exploring ways how to employ performance m
 odeling for auto-tuning.\n\nPasses: Conference Pass, Exhibition Pass, Mole
 cular Research, Performance Analysis and Optimization, Scientific Software
  Development\n\nTag: Conference Pass, Exhibition Pass, Molecular Research,
  Performance Analysis and Optimization, Scientific Software Development
URL:https://2018.isc-program.com/presentation/?id=phd109&sess=sess103
END:VEVENT
END:VCALENDAR