hh.sePublications
Change search
Link to record
Permanent link

Direct link
BETA
Yang, Mingkun
Publications (2 of 2) Show all publications
Essayas, G., Yang, M., Cedersjö, G., Ul-Abdin, Z., Gaspes, V., Janneck, J. W. & Svensson, B. (2014). Realizing Efficient Execution of Dataflow Actors on Manycores. In: Randall Bilof (Ed.), Proceedings: 2014 International Conference on Embedded and Ubiquitous Computing: EUC 2014: August 2014, Milano, Italy. Paper presented at The 12th IEEE International Conference on Embedded and Ubiquitous Computing (EUC 2014), Milan, Italy, Aug. 26-28, 2014 (pp. 321-328). Los Alamitos, CA: IEEE Computer Society, Article ID 6962305.
Open this publication in new window or tab >>Realizing Efficient Execution of Dataflow Actors on Manycores
Show others...
2014 (English)In: Proceedings: 2014 International Conference on Embedded and Ubiquitous Computing: EUC 2014: August 2014, Milano, Italy / [ed] Randall Bilof, Los Alamitos, CA: IEEE Computer Society, 2014, p. 321-328, article id 6962305Conference paper, Published paper (Refereed)
Abstract [en]

Embedded DSP computing is currently shifting towards manycore architectures in order to cope with the ever growing computational demands. Actor based dataflow languages are being considered as a programming model. In this paper we present a code generator for CAL, one such dataflow language. We propose to use a compilation tool with two intermediate representations. We start from a machine model of the actors that provides an ordering for testing of conditions and firing of actions. We then generate an Action Execution Intermediate Representation that is closer to a sequential imperative language like C and Java. We describe our two intermediate representations and show the feasibility and portability of our approach by compiling a CAL implementation of the Two-Dimensional Inverse Discrete Cosine Transform on a general purpose processor, on the Epiphany manycore architecture and on the Ambric massively parallel processor array. © 2014 IEEE.

Place, publisher, year, edition, pages
Los Alamitos, CA: IEEE Computer Society, 2014
Keywords
dataflow languages, compilation framework, code generation, manycore, CAL
National Category
Embedded Systems
Identifiers
urn:nbn:se:hh:diva-26991 (URN)10.1109/EUC.2014.55 (DOI)000358149800046 ()2-s2.0-84908625634 (Scopus ID)978-0-7695-5249-1 (ISBN)978-1-4799-7609-6 (ISBN)
Conference
The 12th IEEE International Conference on Embedded and Ubiquitous Computing (EUC 2014), Milan, Italy, Aug. 26-28, 2014
Projects
HiPEC
Funder
Knowledge FoundationSwedish Foundation for Strategic Research
Available from: 2014-11-05 Created: 2014-11-05 Last updated: 2018-03-22Bibliographically approved
Yang, M., Savas, S., Ul-Abdin, Z. & Nordström, T. (2013). A Communication Library for Mapping Dataflow Applications on Manycore Architectures. In: Tomas Nordstrom & Zain-ul-Abdin (Ed.), Proceedings of the 6th Swedish Multicore Computing Workshop: . Paper presented at 6th Swedish Multicore Computing Workshop, MCC-2013, November 25-26 2013, Halmstad University, Halmstad, Sweden (pp. 65-68).
Open this publication in new window or tab >>A Communication Library for Mapping Dataflow Applications on Manycore Architectures
2013 (English)In: Proceedings of the 6th Swedish Multicore Computing Workshop / [ed] Tomas Nordstrom & Zain-ul-Abdin, 2013, p. 65-68Conference paper, Published paper (Refereed)
Abstract [en]

Dataflow programming is a promising paradigm for high performance embedded parallel computing. When mapping a dataflow program onto a manycore architecture a key component is the library to express the communication between the actors. In this paper we present a dataflow communication library supporting the CAL actor language. A first implementation of the communication library is created for Adapteva’s manycore architecture Epiphany that contains an onchip 2-D mesh network. Three different buffering methods, with and without direct memory access (DMA) transfer, have been implemented and evaluated. We have also made a preliminary study on the effect of mapping strategies of the actors onto the cores. The assessment of the library is based on a CAL implementation of a two dimensional inverse discrete cosine transform (2D-IDCT) and our own CAL-to-C compilation framework. As expected the results show that the most efficient actor to-core mapping strategy is to keep the communication to the nearest neighbor communication pattern as much as possible. Thus, the best way to place a pipelined sequence of computations like our 2D-IDCT is to place the actors into cores in a serpentine fashion. For this application we found that the simple receiver side buffer outperforms the more complicated buffering strategies that used DMA transfer.

National Category
Embedded Systems
Identifiers
urn:nbn:se:hh:diva-24019 (URN)
Conference
6th Swedish Multicore Computing Workshop, MCC-2013, November 25-26 2013, Halmstad University, Halmstad, Sweden
Available from: 2013-11-27 Created: 2013-11-27 Last updated: 2018-03-22Bibliographically approved
Organisations

Search in DiVA

Show all publications