BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20220812T074335Z
LOCATION:Osaka Room
DTSTART;TZID=Europe/Stockholm:20220629T120000
DTEND;TZID=Europe/Stockholm:20220629T123000
UID:submissions.pasc-conference.org_PASC22_sess167_msa210@linklings.com
SUMMARY:FPGAs for ML Inference in CMS
DESCRIPTION:Minisymposium\n\nFPGAs for ML Inference in CMS\n\nJames\n\nThe
  Level-1 (L1) Trigger of the CMS experiment at the CERN LHC selects potent
 ially interesting particle collision events at a rate of approximately 40 
 MHz, within a timespan of a few microseconds. As a result of the huge band
 width requirements, and tiny latency budget, this step of the online proce
 ssing chain utilises FPGA-based hardware. As CMS upgrades for the High Lum
 inosity LHC, the available computational power, and the complexity and vol
 ume of input data at L1 are increasing by an order of magnitude. This allo
 ws for unprecedented reconstruction and selection potential. The capabilit
 y of machine learning (ML) inference to implement these algorithms within 
 a fixed latency, and often in a highly parallelised manner is being exploi
 ted across the upgraded L1 trigger. In order to facilitate this massive gr
 owth in the number and diversity of algorithms being implemented in FPGAs,
  it is necessary to utilise new solutions for creating portable, flexible,
  and maintainable implementations that are accessible to those without har
 dware design knowledge. Tools such as High Level Synthesis languages, soft
  core libraries, and custom ML inference compilers will be discussed and e
 xplored. The results of these strategies as used for the CMS L1 trigger wi
 ll be presented.\n\nDomain: Computer Science and Applied Mathematics, Phys
 ics
END:VEVENT
END:VCALENDAR