Abstract | ||
---|---|---|
There is a trend towards using accelerators to increase per- formance and energy eciency of general-purpose proces- sors. Adoption of accelerators, however, depends on the availability of tools to facilitate programming these devices. In this paper, we present techniques for automatically par- titioning programs for execution on accelerators. We call the o-loaded code regions sub-algorithms, which are parts of the program that are loosely connected to the remainder of the program. We present three heuristics for automati- cally identifying sub-algorithms based on control flow and data flow properties. Analysis of SPECint and MiBench benchmarks shows that on average 12 sub-algorithms are identified (up to 54), cover- ing the full execution time for 27 out of 30 benchmarks. We show that these sub-algorithms are suitable for o-loading them to accelerators by manually implementing sub-algorithms for 2 SPECint benchmarks on the Cell processor. |
Year | DOI | Venue |
---|---|---|
2009 | 10.1145/1531743.1531759 | Conference On Computing Frontiers |
Keywords | Field | DocType |
off-loaded code regions sub-algorithms,accelerators,sub-algorithms,present technique,mibench benchmarks,control flow,data flow property,partitioning program,cell processor,partitioning,specint benchmarks,full execution time,towards automatic program partitioning,o-loading,energy efficiency,data flow | Computer science,Efficient energy use,Parallel computing,Control flow,Remainder,Real-time computing,SPECint,Heuristics,Execution time,Operating system,Data flow diagram | Conference |
Citations | PageRank | References |
3 | 0.41 | 15 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Sean Rul | 1 | 102 | 4.89 |
Hans Vandierendonck | 2 | 629 | 54.43 |
Koen De Bosschere | 3 | 1659 | 117.74 |