BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Saber: Line-Rate Stream Processing on Heterogeneous Architectures 
 - Alexandros Koliousis (Imperial College London)
DTSTART:20150319T150000Z
DTEND:20150319T160000Z
UID:TALK57535@talks.cam.ac.uk
CONTACT:Eiko Yoneki
DESCRIPTION:Many real-time applications (e.g. financial data analytics\, o
 nline machine learning\, and system monitoring) need to process large amou
 nts of streaming data with high throughput and low latency. Window queries
  (for example\, sliding-window aggregates) are fundamental for data stream
  processing. However\, despite the computational power of modern data cent
 er servers\, nowadays equipped with heterogeneous processors such as CPUs 
 and GPGPUs\, window-based query applications remain compute-bound as they 
 lack support to harness these available resources.\n\nThis talk introduces
  Saber\, a hybrid data-parallel stream processing system implemented in Ja
 va that is capable of using both a multi-core CPU and a GPGPU to accelerat
 e window queries. Rather than pinning (part of) a query on a particular pr
 ocessor\, Saber runs it on both and employs a novel preference-based sched
 uling mechanism to maximise the share of computation that runs on the best
 -suited processor. When running on the GPGPU\, Saber interlaces the execut
 ion of consecutive data-parallel tasks to consume the excess system bandwi
 dth that would otherwise remain unused due to the latency of moving data f
 rom Java memory to the accelerator (and vice versa). Putting all together\
 , Saber can process typical window queries at line-rate\, shifting the ers
 twhile computational bottleneck to the network.\n\nBio: Alexandros Kolious
 is is a Research Associate at the Department of Computing\, Imperial Colle
 ge London. He is a member of the Experimental Software Systems and Large-S
 cale Distributed Systems groups. His research is currently focused on engi
 neering data parallel processing systems  that make best use of heterogene
 ous processors. Prior to joining Imperial College\, he was working on home
  network management at the School of Computing Science\, University of Gla
 sgow\, from where he also obtained his MSc and PhD degrees.\n
LOCATION:FW26\, Computer Laboratory\, William Gates Builiding
END:VEVENT
END:VCALENDAR
