BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Optimizing PyTorch AI Models with GPU Profiling and Triton Kernels
  - Part 1
DTSTART;VALUE=DATE-TIME:20260911T114000Z
DTEND;VALUE=DATE-TIME:20260911T134000Z
DTSTAMP;VALUE=DATE-TIME:20260922T182710Z
UID:indico-contribution-2064@events.ncbj.gov.pl
DESCRIPTION:Speakers: Konrad Klimaszewski (NCBJ)\, Michał Obara (NCBJ)\nT
 his training provides a practical introduction to optimizing PyTorch-based
  AI models on GPUs\, with a strong focus on performance profiling and cust
 om Triton kernels. The aim is to equip attendees with the skills needed to
  identify performance bottlenecks and accelerate GPU-based computations. B
 y the end of the training\, attendees will be able to:\n\n* manage CPU–G
 PU memory transfers and reason about performance\,\n* profile GPU code and
  interpret traces to spot bottlenecks\,\n* understand the motivation and p
 rinciples behind writing custom GPU kernels\,\n* write simple custom kerne
 ls in Triton and integrate them into PyTorch workflows\,\n* compare custom
  kernel performance against built-in PyTorch operations.\n\nTarget audienc
 e: Users who already work with PyTorch and want to accelerate and optimize
  their GPU-based numerical or AI computations.\n\nRequirements: Working kn
 owledge of PyTorch tensors and basic GPU concepts\; Python proficiency\; f
 amiliarity with undergraduate-level linear algebra.\n\nRequired tools: Bri
 ng your own laptop — a remote Jupyter Notebook environment with everythi
 ng needed will be provided.\n\nhttps://events.ncbj.gov.pl/event/468/contri
 butions/2064/
LOCATION:
URL:https://events.ncbj.gov.pl/event/468/contributions/2064/
END:VEVENT
END:VCALENDAR