decoration
Site web de Philippe Dumont

Erbium: A deterministic, concurrent intermediate representation to map data-flow tasks to scalable, persistent streaming processes

Résumé

Tuning applications for multicore systems involve subtle concurrency concepts and target-dependent optimizations. This paper advocates for a streaming execution model, called Er- bium, where persistent processes communicate and synchronize through a multi-consumer multi-producer sliding window. Considering media and signal processing applications, we demonstrate the scalability and eciency advantages of streaming compared to data-driven scheduling. To exploit these bene ts in compilers for parallel languages, we propose an intermediate representation enabling the compilation of data- ow tasks into streaming processes. This intermediate representation also facilitates the application of classical compiler optimizations to concurrent programs.

Entrée bibtex

@inproceedings{Miranda:2010:EDC:1878921.1878924,
 author = {Miranda, Cupertino and Pop, Antoniu and Dumont, Philippe 
           and Cohen, Albert and Duranton, Marc},
 title = {Erbium: a deterministic, concurrent intermediate representation 
          to map data-flow tasks to scalable, persistent streaming processes},
 booktitle = {Proceedings of the 2010 international conference on Compilers, 
              architectures and synthesis for embedded systems},
 series = {CASES '10},
 year = {2010},
 isbn = {978-1-60558-903-9},
 location = {Scottsdale, Arizona, USA},
 pages = {11--20},
 numpages = {10},
 url = {http://doi.acm.org/10.1145/1878921.1878924},
 keywords = {compiler intermediate representation, data flow, kahn networks, 
             runtime system, shared memory, stream},
} 

Téléchargement