Thread tailor

Lee, Janghaeng; Wu, Hong; Ravichandran, M.; Clark, Nathan

doi:10.1145/1815961.1815996

Cited by 63 publications

(6 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, if there are eight tasks and four cores, then each core will perform two tasks, and the total time will be roughly the amount of time needed to perform two tasks on a single core. In contrast, nine tasks would take much more time, because the last would execute alone [87].…”

Section: Load Balancingmentioning

confidence: 99%

Parallel Performance Problems on Shared-Memory Multicore Systems: Taxonomy and Observation

Atachiants

Doherty

Gregg

2016

IIEEE Trans. Software Eng.

View full text Add to dashboard Cite

The shift towards multicore processing has led to a much wider population of developers being faced with the challenge of exploiting parallel cores to improve software performance. Debugging and optimizing parallel programs is a complex and demanding task. Tools which support development of parallel programs should provide salient information to allow programmers of multicore systems to diagnose and distinguish performance problems. Appropriate design of such tools requires a systematic analysis of the problems which might be identified, and the information used to diagnose them. Building on the literature, we put forward a potential taxonomy of parallel performance problems, and an observational model which links measurable performance data to these problems. We present a validation of this model carried out with parallel programming experts, identifying areas of agreement and disagreement. This is accompanied with a survey of the prevalence of these problems in software development. From this we can identify contentious areas worthy of further exploration, as well as those with high prevalence and strong agreement, which are natural candidates for initial moves towards better tool support. 0098-5589 (c)

show abstract

Section: Load Balancingmentioning

confidence: 99%

Parallel Performance Problems on Shared-Memory Multicore Systems: Taxonomy and Observation

Atachiants

Doherty

Gregg

2016

IIEEE Trans. Software Eng.

View full text Add to dashboard Cite

show abstract

“…Speculatively executing multiple threads add significant overhead in the former. Few solutions aim to minimize resource utilization using work-stealing [5,25] and other proposals target to avoid oversubscription in [7,20]. These solutions require extensive offline profiling and are not effective on new unseen programs.…”

Section: Related Workmentioning

confidence: 99%

“…There has, in fact, been significant work from the compiler and runtime communities in improving workloadaware thread selection. Schemes that try to combine offline models with runtime tuning [11,14,16,20] can exploit prior knowledge of the program but are limited by the assumptions of the environment.…”

Section: Introductionmentioning

confidence: 99%

Celebrating diversity: a mixture of experts approach for runtime mapping in dynamic environments

Emani

O’Boyle

2015

Proceedings of the 36th ACM SIGPLAN Conference on Programming Language Design and Implementation

View full text Add to dashboard Cite

Matching program parallelism to platform parallelism using thread selection is difficult when the environment and available resources dynamically change. Existing compiler or runtime approaches are typically based on a one-size fits all policy. There is little ability to either evaluate or adapt the policy when encountering new external workloads or hardware resources. This paper focuses on selecting the best number of threads for a parallel application in dynamic environments. It develops a new scheme based on a mixture of experts approach. It learns online which, of a number of existing policies, or experts, is best suited for a particular environment without having to try out each policy. It does this by using a novel environment predictor as a proxy for the quality of an expert thread selection policy. Additional expert policies can easily be added and are selected only when appropriate. We evaluate our scheme in environments with varying external workloads and hardware resources. We then consider the case when workloads use affinity scheduling or are themselves adaptive and show that our approach, in all cases, outperforms existing schemes and surprisingly improves workload performance. On average, we improve 1.66x over OpenMP default, 1.34x over an online scheme, 1.25x over an offline policy and 1.2x over a state-of-art analytic model. Determining the right number and type of experts is an open problem and our initial analysis shows that adding more experts improves accuracy and performance.

show abstract

“…A work stealing application can also shutdown existing threads while preserving program correctness, so long as one thread is left executing. Additionally, automatic methods exist to combine program threads into a single instruction stream [4].…”

Section: Offline Model Construction Dynamic Configuration Of Multithrmentioning

confidence: 99%

Using utility prediction models to dynamically choose program thread counts

Moore

Childers

2012

2012 IEEE International Symposium on Performance Analysis of Systems &Amp; Software

View full text Add to dashboard Cite

Multithreaded applications can simultaneously execute on a chip multiprocessor computer, starting and stopping without warning or pattern. The behavior of each program can be different, interacting in unexpected ways, including causing competition for CPU cycles, which harms performance.To maximize program performance in this type of dynamic execution environment, the interactions among applications must be controlled. These interactions can be controlled by carefully choosing the number of threads for each multithreaded application (i.e., a system configuration). To choose a configuration, we advocate using program utility models to predict application behavior. Only a system that is capable of predicting and analyzing performance under multiple configurations can choose the best configuration and thus robustly meet its performance goals. In this paper, we present such a system.Our approach first gathers profile data. The profile data is used by multiple linear regression to build a utility model. The model takes into account program scalability, susceptibility to interference, and any inherent leveling off of performance as a program's thread count is increased. A utility model is constructed for each application. When the system workload changes, the utility models are consulted to find the new configuration that maximizes system performance while meeting each program's quality of service goals. We use multithreaded applications from PARSEC to evaluate our approach. Compared to the best traditional policy, which does not consider variances in the dynamic workload, our approach simultaneously improves system throughput by 19.3% while meeting user performance constraints 28% more often.

show abstract

Thread tailor

Cited by 63 publications

References 36 publications

Parallel Performance Problems on Shared-Memory Multicore Systems: Taxonomy and Observation

Parallel Performance Problems on Shared-Memory Multicore Systems: Taxonomy and Observation

Celebrating diversity: a mixture of experts approach for runtime mapping in dynamic environments

Using utility prediction models to dynamically choose program thread counts

Contact Info

Product

Resources

About