BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.devconf.info//devconf-cz-2026//talk//G78NSF
BEGIN:VTIMEZONE
TZID:CET
BEGIN:STANDARD
DTSTART:20001029T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-devconf-cz-2026-G78NSF@pretalx.devconf.info
DTSTART;TZID=CET:20260619T144500
DTEND;TZID=CET:20260619T152000
DESCRIPTION:GPU costs are spiraling\, yet clusters waste 30–40% of capaci
 ty due to static allocation. A GPU assigned to a pod sits idle between inf
 erence calls\, model loading\, startup and nobody else can use it. It gets
  worse when VM-based and containerized workloads run on separate clusters.
  The pool is siloed. No sharing\, no reclaim\, just waste.\nThis talk fixe
 s that at the scheduling layer using two upstream Kubernetes projects i.e 
 KubeVirt\, which brings VM workloads under native Kubernetes scheduling\, 
 and Dynamic Resource Allocation (DRA)\, which replaces the rigid device pl
 ugin model with a flexible\, claim-based API. Together they enable GPU sha
 ring across VMs and containers on a single cluster.\nWe'll walk through re
 al scheduling data\, the DRA resource claim model\, and how KubeVirt VM li
 fecycle integrates with DRA's structured parameter API. No theory-heavy sl
 ides. just the problem\, the architecture\, and what works.
DTSTAMP:20260430T125203Z
LOCATION:D0207 (capacity 90)
SUMMARY:The Life of a GPU: From Wasted Resource to Shared Asset with KubeVi
 rt and DRA - Basavaraju G\, Rishika Kedia
URL:https://pretalx.devconf.info/devconf-cz-2026/talk/G78NSF/
END:VEVENT
END:VCALENDAR