Header menu link for other important links
X
Optimizing recursive task parallel programs
Published in Association for Computing Machinery
2017
Volume: Part F128411
   
Abstract
We present a new optimization DECAF that optimizes recursive task parallel (RTP) programs by reducing the task creation and termination overheads. DECAF reduces the task termination (join) operations by aggressively increasing the scope of join operations (in a semantics preserving way), and eliminating the redundant join operations discovered on the way. Further, DECAF extends the traditional loop chunking technique to perform load-balanced chunking, at runtime, based on the number of available worker threads. This helps reduce the redundant parallel tasks at different levels of recursion. We also discuss the impact of exceptions on our techniques and extend them to handle RTP programs that may throw exceptions. We implemented DECAF in the X10v2.3 compiler and tested it over a set of benchmark kernels on two different hardwares (a 16-core Intel system and a 64-core AMD system). With respect to the base X10 compiler extended with loop-chunking of Nandivada et al. [26] (LC), DECAF achieved a geometric mean speed up of 2.14× and 2.53× on the Intel and AMD system, respectively. We also present an evaluation with respect to the energy consumption on the Intel system and show that on average, compared to the LC versions, the DECAF versions consume 71.2% less energy. © 2017 Association for Computing Machinery.
About the journal
JournalData powered by TypesetProceedings of the International Conference on Supercomputing
PublisherData powered by TypesetAssociation for Computing Machinery
Open AccessNo
Concepts (12)
  •  related image
    Energy utilization
  •  related image
    Program compilers
  •  related image
    Semantics
  •  related image
    DATA PARALLEL
  •  related image
    GEOMETRIC MEAN
  •  related image
    JOIN OPERATION
  •  related image
    LOAD-BALANCED
  •  related image
    RECURSIONS
  •  related image
    REDUNDANT PARALLELS
  •  related image
    TASK PARALLEL
  •  related image
    USEFUL PARALLELISM
  •  related image
    Intelligent control