BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Introduction to GPU computing with PyTorch
DTSTART:20260327T090000Z
DTEND:20260327T144500Z
DTSTAMP:20260424T181500Z
UID:indico-event-462@spectra-indico-teszt.hinfra.hu
CONTACT:training@cyfronet.pl\;events-noreply@plgrid.pl
DESCRIPTION:Speakers: Michał Obara (Narodowe Centrum Badań Jądrowych)\,
  Konrad Klimaszewski (Narodowe Centrum Badań Jądrowych)\, Klemens Noga (
 ACC Cyfronet AGH)\n\nThe training provides a practical introduction to gen
 eral-purpose computing on graphics processing units (GPGPU) with a strong 
 focus on the PyTorch framework. The aim is to provide the skills needed to
  design\, implement\, and profile GPU-accelerated computations. By the end
  of the training\, attendees will be able to:\n\nuse PyTorch tensors on GP
 U to implement basic numerical algorithms\,\nuse PyTorch for linear algebr
 a\,\nmanage CPU–GPU memory transfers and reason about performance\,\npro
 file GPU code and spot the main bottlenecks\,\nwrite simple custom kernels
  in Triton and plug them into PyTorch workflows.\n\n \nLevel\n\nTarget au
 dience\nTraining is intended for users who would like to accelerate their 
 Python numerical computations using graphics processing units (GPGPUs).\nA
 genda\n\nIntroduction to GPU Multiprocessing\n\nGPGPU computing paradigm a
 nd typical application domains\,\nOverview of CUDA and hardware-agnostic a
 pproaches.\n\n\nIntroduction to PyTorch\n\nTensors: creation\, initialisat
 ion and parameters\,\nAggregation and shape operations\,\nIndexing\, slici
 ng and broadcasting\, boolean and masked tensors\,\nMatrix multiplication 
 and elementwise math\,\nLinear Algebra Using PyTorch.\n\n\nGPU acceleratio
 n using PyTorch\n\nMemory management in PyTorch\,\nComparing GPU vs CPU pe
 rformance on linear algebra workloads\,\nMotivation and basic principles o
 f performance profiling\,\nProfilers: setup\, tracing and visualisation.\n
 \n\nCustom GPU Kernels with Triton\n\nMotivation for writing custom kernel
 s (performance\, flexibility)\,\nOverview of Triton and its programming mo
 del\,\nImplementing basic kernels (e.g. vector operations\, simple reducti
 ons)\,\nIntegration with PyTorch and comparison to built-in operations.\n\
 n\n\nRequirements\n\nBasic programming proficiency in Python (control flow
 \, functions\, modules).\nFamiliarity with undergraduate-level mathematics
 : calculus and linear algebra (vectors\, matrices\, eigenvalues).\n\nVenue
 \nThe workshop will be held online via Zoom. The meeting link will be sent
  to registered participants.\nLanguage\nEnglish\nDuration\n6 hours\nRegist
 ration\nThe Registration and the Waiting list close automatically after 23
 rd March 2026. The Registration may close prematurely if the limit of part
 icipants is reached beforehand\, but the Waiting List will remain availabl
 e until the deadline above.\nAcknowledgements\nThis course is partially fu
 nded by the EuroCC 2 project.\nThe project has received funding from the E
 uropean High-Performance Computing Joint Undertaking (JU) under grant agre
 ement No 101101903. The JU receives support from the Digital Europe Progra
 mme and Austria\, Belgium\, Bulgaria\, Croatia\, Cyprus\, Czech Republic\,
  Denmark\, Estonia\, Finland\, France\, Germany\, Greece\, Hungary\, Icela
 nd\, Ireland\, Italy\, Latvia\, Lithuania\, Luxembourg\, Montenegro\, Neth
 erlands\, North Macedonia\, Norway\, Poland\, Portugal\, Romania\, Serbia\
 , Slovakia\, Slovenia\, Spain\, Sweden\, and Türkiye.\n\nhttps://spectra-
 indico-teszt.hinfra.hu/event/462/
LOCATION:online 
URL:https://spectra-indico-teszt.hinfra.hu/event/462/
END:VEVENT
END:VCALENDAR