Optimal Footprint Symbiosis in Shared Cache

Wang, Xiaolin; Li, Yechen; Luo, Yingwei; Hu, Xiameng; Brock, Jacob; Ding, Chen; Wang, Zhenlin

doi:10.1109/ccgrid.2015.153

Cited by 15 publications

(26 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Locality-performance Correlation: A recent study by Wang et al shows that the HOTL-based miss ratio prediction has a linear relationship between execution time, with a coefficient of 0.938 [12]. They measure execution times and miss ratios of all 1820 4-programs co-run groups from a set of 16 SPEC programs.…”

Section: Discussion On Optimalitymentioning

confidence: 96%

“…Xiang et al reported on average 23 times slowdown from the full-trace footprint analysis [16]. Wang et al developed a sampling method called adaptive bursty footprint (ABF) profiling, which takes on average 0.09 second per program [12]. To have reproducible results, our implementation uses the full-trace footprint.…”

Section: Discussion On Optimalitymentioning

confidence: 98%

“…The correctness has been evaluated and validated for solo-use cache, including the two initial studies of the footprint theory [15], [16]. Independent validation can be found in the use of the footprint theory in optimal program symbiosis in shared cache [12], optimal memory allocation in Memcached [2], and a study from the OS community on server cache performance for disk access traces [13].…”

Section: Discussion On Optimalitymentioning

confidence: 99%

“…The footprint measurement we use is the same as the implementation by Xiang et al [16] and Wang et al [12] Xiang et al reported on average 23 times slowdown from the full-trace footprint analysis. Wang et al developed a sampling method called adaptive bursty footprint (ABF) profiling, which takes on average 0.09 second per program.…”

Section: A Methodologymentioning

confidence: 99%

“…Following the methodology of Wang et al [12], we randomly choose 16 programs from SPEC 2006: perlbench, bzip2, mcf, zeusmp, namd, dealII, soplex, povray, hmmer, sjeng, h264ref, tonto, lbm,omnetpp, wrf, sphinx3 and use the first reference input. The selection is representative, and the number of co-run groups is not too many to visualize.…”

Section: A Methodologymentioning

confidence: 99%

See 4 more Smart Citations

Optimal Cache Partition-Sharing

Brock

Ding

et al. 2015

2015 44th International Conference on Parallel Processing

Self Cite

View full text Add to dashboard Cite

When a cache is shared by multiple cores, its space may be allocated either by sharing, partitioning, or both. We call the last case partition-sharing. This paper studies partition-sharing as a general solution, and presents a theory an technique for optimizing partition-sharing. We present a theory and a technique to optimize partition sharing. The theory shows that the problem of partition-sharing is reducible to the problem of partitioning. The technique uses dynamic programming to optimize partitioning for overall miss ratio, and for two different kinds of fairness.Finally, the paper evaluates the effect of optimal cache sharing and compares it with conventional solutions for thousands of 4-program co-run groups, with nearly 180 million different ways to share the cache by each co-run group. Optimal partition-sharing is on average 26% better than freefor-all sharing, and 98% better than equal partitioning. We also demonstrate the trade-off between optimal partitioning and fair partitioning.

show abstract

Section: Discussion On Optimalitymentioning

confidence: 96%

Section: Discussion On Optimalitymentioning

confidence: 98%