Matches in SemOpenAlex for { <https://semopenalex.org/work/W36563690> ?p ?o ?g. }
- W36563690 abstract "We present a randomized work-stealing thread scheduler for fork-join multithreaded jobs that provides continual parallelism feedback to the job scheduler in the form of processor requests. Our A-STEAL algorithm is appropriate for large parallel servers where many jobs share a common multiprocessor resource and in which the number of processors available to a particular job may vary during the job’s execution. Assuming that the job scheduler never allots the job more processors than requested by the job’s thread scheduler, A-STEAL guarantees that the job completes in near-optimal time while utilizing at least a constant fraction of the allotted processors. Our analysis models the job scheduler as the thread scheduler’s adversary, challenging the thread scheduler to be robust to the system environment and the job scheduler’s administrative policies. For example, the job scheduler can make available a huge number of processors exactly when the job has little use for them. To analyze the performance of our adaptive thread scheduler under this stringent adversarial assumption, we introduce a new technique called “trim analysis,” which allows us to prove that our thread scheduler performs poorly on at most a small number of time steps, exhibiting near-optimal behavior on the vast majority. To be more precise, suppose that a job has work T1 and critical-path length T∞. On a machine with P processors, A-STEAL completes the job in expected O(T1/P +T∞+ L lgP ) time steps, where L is the length of a scheduling quantum and P denotes the O(T∞ +L lgP )-trimmed availability. This quantity is the average of the processor availability over all but the O(T∞ + L lgP ) time steps having the highest processor availability. When the job’s parallelism dominates the trimmed available, that is, P ≪ T1/T∞, the job achieves nearly perfect linear speedup. Conversely, when the trimmed mean dominates This research was supported in part by the Singapore-MIT Alliance and NSF Grants ACI-0324974 and CNS-0305606. Yuxiong He is a Visiting Scholar at MIT CSAIL and a Ph.D. candidate at the National University of Singapore. the parallelism, the asymptotic running time of the job is nearly the length of its critical path. We measured the performance of A-STEAL on a simulated multiprocessor system using synthetic workloads. For jobs with sufficient parallelism, our experiments indicate that A-STEAL provides almost perfect linear speedup across a variety of processor availability profiles. In these experiments, A-STEAL typically wasted less than 20% of the processor cycles allotted to the job. We compared A-STEAL with the ABP algorithm, an adaptive workstealing thread scheduler developed by Arora, Blumofe, and Plaxton that does not employ parallelism feedback. On moderateto heavy-loaded large machines with predetermined availability profiles, A-STEAL typically completed jobs more than twice as quickly, despite being allotted the same or fewer processors on every step, while wasting only 10% of the processor cycles wasted by ABP." @default.
- W36563690 created "2016-06-24" @default.
- W36563690 creator A5027618658 @default.
- W36563690 creator A5035990793 @default.
- W36563690 creator A5040302174 @default.
- W36563690 date "2008-01-01" @default.
- W36563690 modified "2023-09-27" @default.
- W36563690 title "Work Stealing with Parallelism Feedback" @default.
- W36563690 cites W1511733714 @default.
- W36563690 cites W1522538019 @default.
- W36563690 cites W1554326704 @default.
- W36563690 cites W1563776668 @default.
- W36563690 cites W1580785700 @default.
- W36563690 cites W1585314083 @default.
- W36563690 cites W1591963066 @default.
- W36563690 cites W1684247715 @default.
- W36563690 cites W1773176621 @default.
- W36563690 cites W1822025191 @default.
- W36563690 cites W1968020161 @default.
- W36563690 cites W1968447145 @default.
- W36563690 cites W1984263429 @default.
- W36563690 cites W1987632081 @default.
- W36563690 cites W1993621349 @default.
- W36563690 cites W1995070612 @default.
- W36563690 cites W2001397837 @default.
- W36563690 cites W2007390700 @default.
- W36563690 cites W2009122662 @default.
- W36563690 cites W2014071176 @default.
- W36563690 cites W2016559894 @default.
- W36563690 cites W2021084015 @default.
- W36563690 cites W2027491055 @default.
- W36563690 cites W2032401773 @default.
- W36563690 cites W2056714230 @default.
- W36563690 cites W2060847221 @default.
- W36563690 cites W2061048015 @default.
- W36563690 cites W2061810917 @default.
- W36563690 cites W2068666958 @default.
- W36563690 cites W2072725684 @default.
- W36563690 cites W2076726046 @default.
- W36563690 cites W2079656678 @default.
- W36563690 cites W2079687686 @default.
- W36563690 cites W2082897183 @default.
- W36563690 cites W2087388231 @default.
- W36563690 cites W2087610851 @default.
- W36563690 cites W2090247159 @default.
- W36563690 cites W2094587335 @default.
- W36563690 cites W2098147619 @default.
- W36563690 cites W2103738812 @default.
- W36563690 cites W2104680817 @default.
- W36563690 cites W2107696263 @default.
- W36563690 cites W2114888252 @default.
- W36563690 cites W2115774699 @default.
- W36563690 cites W2119887431 @default.
- W36563690 cites W2123219527 @default.
- W36563690 cites W2130566259 @default.
- W36563690 cites W2134723508 @default.
- W36563690 cites W2144287101 @default.
- W36563690 cites W2149663533 @default.
- W36563690 cites W2159972990 @default.
- W36563690 cites W2167003842 @default.
- W36563690 cites W2168426407 @default.
- W36563690 cites W2175675949 @default.
- W36563690 cites W2283605101 @default.
- W36563690 cites W2294090198 @default.
- W36563690 cites W2295428206 @default.
- W36563690 cites W2295944892 @default.
- W36563690 cites W2752885492 @default.
- W36563690 cites W3015460342 @default.
- W36563690 cites W3142328950 @default.
- W36563690 cites W1493797193 @default.
- W36563690 cites W3206388299 @default.
- W36563690 hasPublicationYear "2008" @default.
- W36563690 type Work @default.
- W36563690 sameAs 36563690 @default.
- W36563690 citedByCount "3" @default.
- W36563690 crossrefType "journal-article" @default.
- W36563690 hasAuthorship W36563690A5027618658 @default.
- W36563690 hasAuthorship W36563690A5035990793 @default.
- W36563690 hasAuthorship W36563690A5040302174 @default.
- W36563690 hasConcept C111873713 @default.
- W36563690 hasConcept C111919701 @default.
- W36563690 hasConcept C120314980 @default.
- W36563690 hasConcept C126255220 @default.
- W36563690 hasConcept C138101251 @default.
- W36563690 hasConcept C171627638 @default.
- W36563690 hasConcept C173608175 @default.
- W36563690 hasConcept C206729178 @default.
- W36563690 hasConcept C33923547 @default.
- W36563690 hasConcept C41008148 @default.
- W36563690 hasConcept C4822641 @default.
- W36563690 hasConcept C79974875 @default.
- W36563690 hasConceptScore W36563690C111873713 @default.
- W36563690 hasConceptScore W36563690C111919701 @default.
- W36563690 hasConceptScore W36563690C120314980 @default.
- W36563690 hasConceptScore W36563690C126255220 @default.
- W36563690 hasConceptScore W36563690C138101251 @default.
- W36563690 hasConceptScore W36563690C171627638 @default.
- W36563690 hasConceptScore W36563690C173608175 @default.
- W36563690 hasConceptScore W36563690C206729178 @default.
- W36563690 hasConceptScore W36563690C33923547 @default.