Automatic Identification and Parallelisation of General Reduction Operations

Add to your list(s) Download to your calendar using vCal

• Philip Ginsbach, University of Edinburgh
• Wednesday 23 November 2016, 15:15-16:15
• SS03.

If you have a question about this talk, please contact Alan Mycroft.

Discovering and exploiting reduction parallelism hasbeen studied for many years but no current approach scales well beyond simplistic scalar reductions with low arithmetic intensity. In this paper we develop a new approach that automatically detects a wide class of performance critical reductions in established benchmark suites. This approach is based on a constraint formulation of the reduction idiom and has been implemented as an LLVM pass. We use a customized constraint solver to identify program subsets that adhere to the constraint specification. Once discovered, we automatically generate parallel code to exploit the reduction. This approach is robust and was evaluated on C versions of three well known benchmark suites: NAS , Parboil and Rodinia. We detected 84 scalar reductions and 6 histograms, outperforming existing approaches and show that exploiting histograms gives significant performance improvement.

This talk is part of the Computer Laboratory Programming Research Group Seminar series.

This talk is included in these lists:

• All Talks (aka the CURE list)
• Cambridge Centre for Data-Driven Discovery (C2D3)
• Cambridge talks
• Chris Davis' list
• Computer Laboratory Computer Architecture Group Meeting
• Computer Laboratory Programming Research Group Seminar
• Department of Computer Science and Technology talks and seminars
• Interested Talks
• SS03
• School of Technology
• Trust & Technology Initiative - interesting events
• bld31
• ndk22's list
• ob366-ai4er
• rp587

Note that ex-directory lists are not shown.