BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.devconf.info//devconf-cz-2025//speaker//QPS9QS
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-2025-LLXY7K@pretalx.devconf.info
DTSTART;TZID=CET:20250613T153000
DTEND;TZID=CET:20250613T160500
DESCRIPTION:Even though most observability solutions stay out of your way a
 fter deploying them\, you will most likely still need to get your data to 
 them somehow. Many projects recommend a so-called “Agent” or “Collec
 tor”. But what exactly is an agent\, and why do I need one?\n\nThis talk
  covers the design decisions behind observability agents\, the open protoc
 ols used by prominent examples\, and the different ways they work together
 . On top\, these agents also need to be deployed and configured. Container
  orchestration solutions simplify this a lot but provides their own challe
 nges.\n\nAfter this talk\, you will be able to reason about different depl
 oyment architectures and know which methods of gathering observability dat
 a to choose for specific situations.
DTSTAMP:20261016T055343Z
LOCATION:D0206 (capacity 154)
SUMMARY:License to Observe - Why observability solutions need agents - Domi
 nik Süß
URL:https://pretalx.devconf.info/devconf-cz-2025/talk/LLXY7K/
END:VEVENT
BEGIN:VEVENT
UID:pretalx-devconf-cz-2025-GEETLG@pretalx.devconf.info
DTSTART;TZID=CET:20250614T144500
DTEND;TZID=CET:20250614T152000
DESCRIPTION:Modern AI workloads rely on large GPU fleets whose efficient ut
 ilisation is crucial due to high costs. However\, gathering telemetry from
  these workloads to optimise performance is challenging because it require
 s manual instrumentation and adds performance overheads. Further\, it does
  not produce telemetry in a standardised format for commonly used visualis
 ation tools like Prometheus.\n\nThis talk explores the potential of levera
 ging eBPF to capture CUDA calls made to GPUs\, including kernel launches a
 nd memory allocations. Data from these probes can be used to export Promet
 heus metrics\, facilitating detailed analysis of kernel launch patterns an
 d associated memory usage. This approach offers significant benefits as eB
 PF imposes minimal overhead and requires no intrusive instrumentation. Our
  implementation is also open-source and available on GitHub.
DTSTAMP:20261016T055343Z
LOCATION:D0206 (capacity 154)
SUMMARY:Auto-instrumentation for GPU performance using eBPF - Marc Tuduri\,
  Dominik Süß
URL:https://pretalx.devconf.info/devconf-cz-2025/talk/GEETLG/
END:VEVENT
END:VCALENDAR