BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Two Software-Based Models of Thread Level Speculation - Cosmin Oan
 cea (University of Cambridge)
DTSTART:20070601T141500Z
DTEND:20070601T151500Z
UID:TALK7522@talks.cam.ac.uk
CONTACT:Viktor Vafeiadis
DESCRIPTION:Although thread-level speculation (TLS) has proved to be a pro
 mising direction for exploiting the programs' inherent parallelism\, their
  application to mainstream use is still delayed. \nOn the one hand\, hardw
 are-centric approaches involve non-trivial and  expensive changes to the b
 asic cache-coherence infrastructure\, which have not as yet been integrate
 d in the design of the commercial processors.  On the other hand\, current
  software solutions use too much extra-memory\,  or incur non-constant and
  potentially high overheads\, or are not applicable to general program pat
 terns.\n\nWe present here two software-based TLS models that address the o
 bserved shortcomings of existing solutions:    First\, the speculative mem
 ory size is adaptable\, but in general much smaller than the one of the se
 quential program. Second\, the speculative operation overhead is nearly co
 nstant and comparable with the ``ideal'' case of the related approaches\, 
 to the result that the system's performance can be easily anticipated.  Th
 ird\, we achieve these without using any synchronization primitive\, as we
  find the latter prohibitively expensive. Fourth\, dependencies are tracke
 d at the instruction/iteration level\, and therefore our models tolerate w
 ell dependency-violations\, as long as they are rear events. Finally\, our
  solutions do not rely on specialized compiler support\, and hence\, they 
 can be easily integrated in a dynamic compiler or used directly by the  pr
 ogrammer.   \n\nWe show that our models yield significant speed-ups\, as h
 igh as 260% on a four processor SMP machine\, even when nearly all of the 
 memory access operations are guarded with speculative support. We argue th
 at due to these properties\, our framework is better suited than others to
  expose TLS to mainstream use. \n
LOCATION:GS15\, Computer Laboratory
END:VEVENT
END:VCALENDAR