Title: Auto-tuning for large-scale image processing by dynamic analysis method on multicore platforms

Authors: Yan Wang; Brian Vinter

Addresses: Niels Bohr Institute, University of Copenhagen, Copenhagen, Denmark; Niels Bohr Institute, Blegdamsvej 17, DK-2100, Copenhagen, Denmark ' Niels Bohr Institute, University of Copenhagen, Copenhagen, Denmark; Niels Bohr Institute, Blegdamsvej 17, DK-2100, Copenhagen, Denmark

Abstract: This paper describes a general-purpose method of improving execution performance of the in-memory data, particularly in the case of large-scale image processing on different multicore platforms. To process large-scale arrays, the method of tiling is widely used to achieve high performance. However, frequently accessing the memory system by multithreads is bound to cause system bottleneck. Our optimisation strategies are automatic thread scheduling and data/task partitioning. Those methods that attempt to take advantage of spatial and temporal locality can reduce memory traffic remarkably. According to the hardware configurations, a scheduler automatically partitions the images into tiled blocks of pre-determined size. Then it fuses all the operations for the same blocks to reduce the rate of cache miss. The parallel task execution is more effective than other traditional parallel libraries, such as openMP. Moreover, the optimisation on space-filling curves that optimises the locality of neighbouring tiled blocks can also contribute to the fast memory access.

Keywords: multicore computing; locality; tiling; operation fusion; image processing; embedded systems; auto-tuning; dynamic analysis; large-scale arrays; optimisation strategies; automatic thread scheduling; data-task partitioning; fast memory access; space-filling curves; parallel task execution.

DOI: 10.1504/IJES.2016.077784

International Journal of Embedded Systems, 2016 Vol.8 No.4, pp.313 - 322

Received: 23 Sep 2013
Accepted: 10 Apr 2014

Published online: 15 Jul 2016 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article