BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:OmniX - an OS architecture for omni-programmable systems
DTSTART:20191003T161500
DTEND:20191003T171500
DTSTAMP:20260414T105940Z
UID:b83d3475db9ebd0a26384768b8d5170d81192670a109463470f2b3f2
CATEGORIES:Conferences - Seminars
DESCRIPTION:Mark Silberstein\, Associate Professor at the EE department at
  the Technion - Israel Institute of Technology\nAbstract:. Future systems 
 will be omni-programmable: alongside CPUs\, GPUs\, Security accelerators a
 nd FPGAs\, they will execute user code near-storage\, near-network\, and n
 ear-memory. Ironically\, while breaking power and memory walls via hardwar
 e specialization and near data processing\, emerging programmability wall 
 will become a key impediment for materializing the promised performance an
 d power efficiency benefits of omni-programmable systems. I argue that the
  root cause of the programming complexity lies in todays CPU-centric opera
 ting system (OS) design which is no longer appropriate for omni-programmab
 le systems.\n \nIn this talk I will describe the ongoing efforts in my la
 b to design an accelerator-centric OS called OmniX [HotOS'17]\, which exte
 nds standard OS abstractions into accelerators\, while maintaining a coher
 ent view of the system among all the processors. In OmniX\, near-data and 
 compute accelerators may directly invoke tasks and access I/O services amo
 ng themselves\, excluding the CPU from the performance-critical data and c
 ontrol plane operations\, and turning it into a "yet another" accelerator 
 for sequential computations. I will show how OmniX design principles have 
 been successfully applied to GPUs\, Programmable NICs and Intel SGX.\n \n
 Bio: Mark Silberstein is an Associate Professor at the EE department at th
 e Technion - Israel Institute of Technology where he is heading the Accele
 rated Computing Systems Lab. His research is centered around OSes for comp
 ute and I/O accelerator architectures\, which led to several publications 
 in ISCA\, ASPLOS\, OSDI\, PACT\, Eurosys and USENIX ATC\, all of which str
 ive to systematically minimize or eliminate the dependence of accelerated 
 systems on the host CPU. He is working on practical ways to protect agains
 t side channels\, in particular Intel SGX\, though he was more successful 
 at attacks at first and discovered the Foreshadow speculative execution bu
 g. He is a regular contributor to the SIGARCH blog.\n 
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR