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:20250822T115806Z
LOCATION:Room 5.0B15 & 16
DTSTART;TZID=Europe/Stockholm:20250617T160000
DTEND;TZID=Europe/Stockholm:20250617T163000
UID:submissions.pasc-conference.org_PASC25_sess135_msa274@linklings.com
SUMMARY:A GPU-Accelerated Unified API for Singular Values Enabling Reprodu
 cability Across Architectures and Data Types
DESCRIPTION:Evelyne Ringoot, Rabab Alomairy, Valentin Churavy, and Alan Ed
 elman (Massachusetts Institute of Technology)\n\nWe present a portable, GP
 U-accelerated implementation of a QR-based Singular Value algorithm in Jul
 ia, that allows code reproducibility across several different GPU vendors.
  Singular Value Decomposition (SVD) is a fundamental numerical tool in sci
 entific computing and machine learning, providing optimal low-rank matrix 
 approximations with applications ranging from dimensionality reduction to 
 data compression and signal processing.  Our implementation leverages Juli
 a’s multiple dispatch and metaprogramming capabilities, integrating with t
 he GPUArrays and KernelAbstractions frameworks to provide a unified type-,
  and hardware-agnostic API. It supports diverse GPU architectures and data
  types, including half precision and Apple Metal. We benchmark the algorit
 hm against several state-of-the-art linear algebra libraries and confirm p
 erformance reproducibility through a unified API. We explore GPU kernel op
 timization through parameter tuning to enable efficient parallelism and im
 proved memory locality. Performance results on multiple GPU backends and d
 ata types demonstrate  scalability combined with reproducibility, highligh
 ting Julia’s suitability for high-performance linear algebra in heterogene
 ous environments.\n\nDomain: Chemistry and Materials, Engineering, Life Sc
 iences, Physics, Computational Methods and Applied Mathematics\n\nSession 
 Chair: Samuel Omlin (ETH Zurich / CSCS)\n\n
END:VEVENT
END:VCALENDAR