BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20210402T160551Z
LOCATION:Track 8
DTSTART;TZID=America/New_York:20201111T160000
DTEND;TZID=America/New_York:20201111T160500
UID:submissions.supercomputing.org_SC20_sess198_ws_whpc107@linklings.com
SUMMARY:Toward Modular Supercomputing: Resource Disaggregation and Virtual
 ization by Network-Attached Accelerators
DESCRIPTION:Workshop\n\nToward Modular Supercomputing: Resource Disaggrega
 tion and Virtualization by Network-Attached Accelerators\n\nNeuwirth\n\nTh
 is lightning talk presents the Network-Attached Accelerator (NAA) approach
 , which increases communication efficiency by moving accelerators into a m
 assively-parallel “stand-alone” computing cluster. The novel architecture 
 enables the direct communication between remote accelerators transparently
  through the Extoll high-speed network without any host interactions and a
 n optimal application-to-compute-resources mapping by disaggregating the a
 ccelerators from their PCI Express hosts and virtualizing the hardware ove
 r the network. From a user’s perspective, the NAA software environment pro
 vides a transparent mapping between the remote accelerators and the cluste
 r nodes by emulating the PCI Express subsystem for the underlying operatin
 g system. The architectural idea has been derived from the European Dynami
 cal Exascale Entry Platform (DEEP) project series, which introduces the Cl
 uster-Booster concept to enable a larger amount of codes exploiting the ad
 vantages of highly scalable systems, while improving the energy-efficiency
  and scalability of cluster computers. \n\nInitial prototype implementatio
 ns with different accelerators such as Intel Xeon Phi Coprocessors and NVI
 DIA GPGPUs of the Tesla generation show promising performance results for 
 both micro- and application benchmarks. For example, the communication tim
 e between remote GPUs can be reduced by up to 47%. Furthermore, the Networ
 k-Attached Accelerator approach serves as the prototype compute module for
  the Modular Supercomputing Architecture (MSA), which extends and generali
 zes the Cluster-Booster concept. The idea is to connect compute modules (s
 uch as cluster, data analytics, booster, and quantum computing modules) wi
 th different hardware and performance characteristics with each other to c
 reate a single heterogeneous system.\n\nTag: Education, Training and Outre
 ach, Professional Development, Workforce Development\n\nRegistration Categ
 ory: Workshop Reg Pass
END:VEVENT
END:VCALENDAR