BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:GstVulkan in Production: Lessons from the Real World
DTSTART;VALUE=DATE-TIME:20261011T120000Z
DTEND;VALUE=DATE-TIME:20261011T123000Z
DTSTAMP;VALUE=DATE-TIME:20261010T001724Z
UID:indico-contribution-620@indico.freedesktop.org
DESCRIPTION:Speakers: Anders Hellerup Madsen ()\nTo gain cross-platform ca
 pabilities in our video processing pipelines\, we at Veo Technologies deci
 ded to convert our custom GPU-dependent GStreamer elements from CUDA to Vu
 lkan. Today\, several of our production video processing pipelines run ent
 irely on GstVulkan\, processing hundreds of thousands of sports recordings
  every week\, having performance parity with the CUDA pipelines.\n\n- We c
 urrently run the same pipelines on Nvidia Tesla T4\, Nvidia Jetson\, MacOS
 \, and other GPU vendors.\n- We gained hands-on experience with GstVulkan 
 in production.\n- We converted many CUDA kernels to Vulkan and Slang.\n- W
 e learned a lot about Vulkan synchronisation…\n- We solved a lot of chal
 lenges and upstreamed a few patches to GstVulkan along the way.\n\nIn this
  talk\, we will share our experiences with using GstVulkan in production: 
 which strategies we used\, which challenges we ran into\, and what we gain
 ed.\n\nhttps://indico.freedesktop.org/event/14/contributions/620/
LOCATION:Impact Hub Prague
URL:https://indico.freedesktop.org/event/14/contributions/620/
END:VEVENT
END:VCALENDAR