Title :
Eliminating Redundant Computation and Exposing Parallelism through Data-Triggered Threads
Author :
Tseng, Hung-Wei ; Tullsen, Dean M.
Author_Institution :
Dept. of Comput. Sci. & Eng., Univ. of California, San Diego, CA, USA
Abstract :
Unlike threads in parallel programs created by conventional programming, data-triggered threads are initiated when a memory value is changed. By expressing computation through these threads, computation is executed only when the data changes and is skipped whenever the data does not change. The authors´ model achieves performance speedups of up to 5.9x, averaging 45.6 percent, with SPEC2000 benchmarks.
Keywords :
data handling; parallel programming; SPEC2000 benchmarks; conventional programming; data triggered threads; exposing parallelism; memory value; parallel programs; redundant computation elimination; Computational modeling; Data structures; Instruction sets; Load modeling; Parallel programming; Programming; dataflow languages; multithreaded processors; parallel;
Journal_Title :
Micro, IEEE
