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:20260724T151408Z
LOCATION:Bldg. 6 - 104
DTSTART;TZID=Europe/Stockholm:20260701T140000
DTEND;TZID=Europe/Stockholm:20260701T160000
UID:submissions.pasc-conference.org_PASC26_sess163@linklings.com
SUMMARY:MS5F - Single and Reduced Precision for Weather and Climate Models
DESCRIPTION:Access the recording\n\nOrganizer(s): Balthasar Reuter (ECMWF)
 , and Claudia Frauen (DKRZ)\n\nHigh-resolution weather and climate models 
 are rapidly approaching the long-standing goal of simulating one year per 
 day at 1 km global resolution, thanks to GPU-accelerated pre-Exascale and 
 Exascale HPC systems. Achieving this performance sustainably, however, req
 uires more than traditional code optimization - it demands HPC co-design, 
 where scientific developers, numerical analysts, and computational experts
  collaborate to rethink algorithms for modern hardware. One promising stra
 tegy is the use of single- or reduced-precision arithmetic. Lower-precisio
 n computations improve memory bandwidth, cache efficiency, and energy usag
 e, and enable exploitation of specialized GPU units optimized for fast, lo
 w-precision operations. At the same time, reducing precision raises import
 ant questions about scientific trust, numerical stability, and reproducibi
 lity, particularly in chaotic, multi-scale Earth system models. This minis
 ymposium will bring together experts from climate science, numerical model
 ing, and HPC to share experiences, challenges, and best practices in adopt
 ing lower-precision computations. Topics include algorithmic reformulation
 , precision-aware software design, verification and validation strategies,
  and precision analysis tools. While focused on weather and climate models
 , the lessons learned are broadly applicable across computational science,
  offering guidance for building performant, trustworthy, and future-ready 
 models on next-generation HPC platforms.\n\nReducing and Validating Precis
 ion in the ICON model\n\nReducing model precision can significantly reduce
  the memory footprint of a model, and increase utilization on future AI-ba
 sed hardware. However, identifying and validating single precision model c
 omponents without compromising on the model accuracy of scientific experim
 ents remains a key challenge....\n\n\nDylan Kierans (DKRZ)\n--------------
 -------\nSingle Precision Trajectories in 4D‑Var Data Assimilation\n\nFour
 ‑dimensional variational (4D‑Var) data assimilation is underpinning the ge
 neration of high‑quality initial conditions required for ECMWF’s weather f
 orecasts and Destination Earth's (DestinE) Digital Twins. A major cost dri
 ver in this system is the nonlinear (NL) model tra...\n\n\nJorge Bandeiras
 , Patrick Gillies, Balthasar Reuter, and Ioan Hadade (ECMWF)\n------------
 ---------\nPerformance, Precision and Accuracy: Mixed-Precision Modelling 
 for Weather and Climate Applications at the Met Office\n\nNumerical calcul
 ations using floating point numbers suffer from accumulated round-off erro
 r and moreover, they are not associative. This is problematic for  weather
  and climate applications as they rely on global sums in PDE solvers and t
 he numerical answer can then depend on the parallel decomposit...\n\n\nChr
 istopher Michael Maynard (Met Office)\n---------------------\nDoes Halving
  the Precision Yield Twice the Speed?\n\nModern GPU architectures offer su
 bstantially higher throughput for single-precision arithmetic compared to 
 double precision, making reduced-precision computation an attractive strat
 egy for accelerating weather and climate models. In theory, halving the pr
 ecision could yield up to twice the performan...\n\n\nDmitry Alexeev (NVID
 IA Inc.)\n\nDomain: Climate, Weather, and Earth Sciences\n\nSession Chairs
 : Balthasar Reuter (ECMWF) and Claudia Frauen (DKRZ)
END:VEVENT
END:VCALENDAR