BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.com//juliacon-2022//speaker//ZTATFJ
BEGIN:VEVENT
UID:pretalx-juliacon-2022-YPGNCS@pretalx.com
DTSTART:20220727T130000Z
DTEND:20220727T133000Z
DESCRIPTION:In the Fall Semester 2021 at ETH Zurich\, we designed and taugh
 t a new course: **Solving PDEs in parallel on GPUs with Julia**. We presen
 t technical and teaching experiences we gained: we look at our tech-stack 
 `CUDA.jl`\, `ParallelStencils.jl` and `ImplictGlobalGrid.jl` for GPU-compu
 ting\; and `Franklin.jl`\, `Literate.jl`\, `IJulia.jl`/Jupyter for web\, s
 lides\, and exercises. We look into the crash-course in Julia\, teaching s
 oftware-engineering (git\, CI) and project-based student evaluations.
DTSTAMP:20260316T202931Z
LOCATION:Blue
SUMMARY:Teaching GPU computing\, experiences from our Master-level course -
  Ludovic Räss\, Samuel Omlin\, Mauro Werder
URL:https://pretalx.com/juliacon-2022/talk/YPGNCS/
END:VEVENT
BEGIN:VEVENT
UID:pretalx-juliacon-2022-7FVVF3@pretalx.com
DTSTART:20220727T134000Z
DTEND:20220727T135000Z
DESCRIPTION:The accelerating outflow of ice in Antarctica or Greenland due 
 to a warming climate or the geodynamic processes shaping the Earth share c
 ommon computational challenges: extreme-scale high-performance computing (
 HPC) which requires the next-generation of numerical models\, parallel sol
 vers and supercomputers. We here present a fresh approach to modern HPC an
 d share our experience running Julia on thousands of graphical processing 
 units (GPUs).
DTSTAMP:20260316T202931Z
LOCATION:Blue
SUMMARY:GPU4GEO - Frontier GPU multi-physics solvers in Julia - Ludovic Rä
 ss\, Samuel Omlin\, Albert de Montserrat\, Boris Kaus
URL:https://pretalx.com/juliacon-2022/talk/7FVVF3/
END:VEVENT
BEGIN:VEVENT
UID:pretalx-juliacon-2022-AKVUKM@pretalx.com
DTSTART:20220727T152000Z
DTEND:20220727T153000Z
DESCRIPTION:We present an efficient approach for writing architecture-agnos
 tic parallel high-performance stencil computations in Julia. Powerful meta
 programming\, costless abstractions and multiple dispatch enable writing a
  single code that is usable for both productive prototyping on a single CP
 U and for production runs on GPU or CPU workstations or supercomputers. Pe
 rformance similar to CUDA C is achievable\, which is typically a large imp
 rovement over reachable performance with `CUDA.jl` Array programming.
DTSTAMP:20260316T202931Z
LOCATION:Purple
SUMMARY:High-performance xPU Stencil Computations in Julia - Ludovic Räss\
 , Samuel Omlin
URL:https://pretalx.com/juliacon-2022/talk/AKVUKM/
END:VEVENT
BEGIN:VEVENT
UID:pretalx-juliacon-2022-RJYBLA@pretalx.com
DTSTART:20220727T153000Z
DTEND:20220727T154000Z
DESCRIPTION:We present a straightforward approach for distributed paralleli
 zation of stencil-based Julia applications on a regular staggered grid usi
 ng GPUs and CPUs. The approach allows to leverage remote direct memory acc
 ess and was shown to enable close to ideal weak scaling of real-world appl
 ications on thousands of GPUs. The communication performed can be easily h
 idden behind computation.
DTSTAMP:20260316T202931Z
LOCATION:Purple
SUMMARY:Distributed Parallelization of xPU Stencil Computations in Julia - 
 Ludovic Räss\, Samuel Omlin
URL:https://pretalx.com/juliacon-2022/talk/RJYBLA/
END:VEVENT
END:VCALENDAR