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:20250822T115810Z
LOCATION:Campussaal - Plenary Room
DTSTART;TZID=Europe/Stockholm:20250617T103000
DTEND;TZID=Europe/Stockholm:20250617T110000
UID:submissions.pasc-conference.org_PASC25_sess150_posC102@linklings.com
SUMMARY:ACMP06 - Understanding HMM Performance for Enhanced HPC Portabilit
 y
DESCRIPTION:Nicholas Cassarino (University of North Carolina at Charlotte)
 \n\nHeterogeneous Memory Management (HMM) simplifies programming for heter
 ogeneous systems, making High-Performance Computing (HPC) devices more acc
 essible to domain scientists; however, it suffers from slow performance co
 mpared to other memory management approaches. HMM is an infrastructure pro
 vided by Linux to enable a more simple and universal usage of non-conventi
 onal memory, enabling usage of multiple devices without the need for devel
 opers to use runtime APIs for memory allocation and data transfer. This si
 mplification benefits domain scientists by reducing code complexity, makin
 g it easier to transition between systems, and enabling quicker adoption o
 f legacy code for complex heterogeneous systems, such as those found in HP
 C Centers.\nCurrently, HMM has slow performance compared to explicit memor
 y management and Unified Virtual Memory (UVM), a similar infrastructure pr
 ovided by NVIDIA for their GPUs. UVM requires driver specific APIs for all
 ocation, but once allocated the memory can be used by any device. Due to t
 he similarity between HMM and UVM, we expect any performance differences t
 o result from improper UVM driver implementation, challenges in utilizing 
 HMM correctly, or inefficient algorithms introduced by the abstraction. In
  our work, we conduct experiments to identify the root cause of the expect
 ed underlying issues and provide insights into their impact.\n\nSession Ch
 air: David Moxey (King's College London)\n\n
END:VEVENT
END:VCALENDAR