A profile-based tool for finding pipeline parallelism in sequential programs

Traditional static analysis fails to auto-parallelize programs with a complex control and data flow. Furthermore, thread-level parallelism in such programs is often restricted to pipeline parallelism, which can be hard to discover by a programmer. In this paper we propose a tool that, based on profiling information, helps the programmer to discover parallelism. The programmer hand-picks the code transformations from among the proposed candidates which are then applied by automatic code transformation techniques.

This paper contributes to the literature by presenting a profiling tool for discovering thread-level parallelism. We track dependencies at the whole-data structure level rather than at the element level or byte level in order to limit the profiling overhead. We perform a thorough analysis of the needs and costs of this technique. Furthermore, we present and validate the belief that programs with complex control and data flow contain significant amounts of exploitable coarse-grain pipeline parallelism in the program’s outer loops. This observation validates our approach to whole-data structure dependencies. As state-of-the-art compilers focus on loops iterating over data structure members, this observation also explains why our approach finds coarse-grain pipeline parallelism in cases that have remained out of reach for state-of-the-art compilers. In cases where traditional compilation techniques do find parallelism, our approach allows to discover higher degrees of parallelism, allowing a 40% speedup over traditional compilation techniques. Moreover, we demonstrate real speedups on multiple hardware platforms.

Parallel Programming of General-Purpose Programs Using Task-Based Programming Models

The prevalence of multicore processors is bound to drive most kinds of software development towards parallel programming. To limit the difficulty and overhead of parallel software design and maintenance, it is crucial that parallel programming models allow an easy-to-understand, concise and dense representation of parallelism. Parallel programming models such as Cilk++ and Intel TBBs attempt to offer a better, higher-level abstraction for parallel programming than threads and locking synchronization. It is not straightforward, however, to express all patterns of parallelism in these models. Pipelines are an important parallel construct, although difficult to express in Cilk and TBBs in a straightfor- ward way, not without a verbose restructuring of the code. In this paper we demonstrate that pipeline parallelism can be easily and concisely expressed in a Cilk-like language, which we extend with input, output and input/output dependency types on procedure arguments, enforced at runtime by the scheduler. We evaluate our implementation on real applications and show that our Cilk-like scheduler, extended to track and enforce these dependencies has performance comparable to Cilk++.

Service user involvement in drug treatment programs: Barriers to implementation and potential benefits for client recovery

Service user forums have the potential for improving awareness of services, empowering service users and strengthening community partnerships within an inclusive treatment and rehabilitation framework. The research aimed to investigate perspectives about service user involvement in order to inform the development of effective service user forum(s) in west Ireland. A total of 30 interviews with key service providers and 12 interviews with service users were conducted, with interview questions focusing on: (1) awareness of the Service User Support Team and (2) barriers to service user involvement and the development of service user forums in the region. An integrated data collection and thematic analysis was undertaken. Current levels of service user involvement were low, restricted by one-way communication and appeared grounded in user-provider power differentials and stigma relating to drug dependency. Service providers queried the actual terms of reference, capacity and training that would be needed for service user forums to advocate and lobby for service users. The use of existing support groups, creation of internet user forums and rotation of rural meetings were recommended to promote engagement among service users. The research underscores the need for transparency, resources and a framework for good practice that reflects a participatory approach


Read More: http://informahealthcare.com/doi/abs/10.3109/09687637.2012.671860

Formal Analysis of MPI-based Parallel Programs

Most parallel computing applications in highperformance computing use the Message Passing Interface (MPI) API. Given the fundamental importance of parallel computing to science and engineering research, application correctness is paramount. MPI was originally developed around 1993 by the MPI Forum, a group of vendors, parallel programming researchers, and computational scientists. However, the document defining the standard is not issued by an official standards organization but has become a de facto standard © 2011 ACM.

Large Scale Verification of MPI Programs Using Lamport Clocks with Lazy Update

We propose a dynamic verification approach for large-scale message passing programs to locate correctness bugs caused by unforeseen nondeterministic interactions. This approach hinges on an efficient protocol to track the causality between nondeterministic message receive operations and potentially matching send operations. We show that causality tracking protocols that rely solely on logical clocks fail to capture all nuances of MPI program behavior, including the variety of ways in which nonblocking calls can complete. Our approach is hinged on formally defining the matches-before relation underlying the MPI standard, and devising lazy update logical clock based algorithms that can correctly discover all potential outcomes of nondeterministic receives in practice. can achieve the same coverage as a vector clock based algorithm while maintaining good scalability. LLCP allows us to analyze realistic MPI programs involving a thousand MPI processes, incurring only modest overheads in terms of communication bandwidth, latency, and memory consumption. © 2011 IEEE.

Ethics in neuroscience graduate training programs:Views and models from Canada

Consideration of the ethical, social, and policy implications of research has become increasingly important to scientists and scholars whose work focuses on brain and mind, but limited empirical data exist on the education in ethics available to them. We examined the current landscape of ethics training in neuroscience programs, beginning with the Canadian context specifically, to elucidate the perceived needs of mentors and trainees and offer recommendations for resource development to meet those needs. We surveyed neuroscientists at all training levels and interviewed directors of neuroscience programs and training grants. A total of 88% of survey respondents reported general interest in ethics, and 96% indicated a desire for more ethics content as it applies to brain research and clinical translation. Expert interviews revealed formal ethics education in over half of programs and in 90% of grants-based programs. Lack of time, resources, and expertise, however, are major barriers to expanding ethics content in neuroscience education. We conclude with an initial set of recommendations to address these barriers which includes the development of flexible, tailored ethics education tools, increased financial support for ethics training, and strategies for fostering collaboration between ethics experts, neuroscience program directors, and funding agencies. © 2010 the Authors. Journal Compilation © 2010 International Mind, Brain, and Education Society and Blackwell Publishing, Inc.

Detection of visual-field deterioration by Glaucoma Progression Analysis and Threshold Noiseless Trend programs

BACKGROUND: To compare the ability of Glaucoma Progression Analysis (GPA) and Threshold Noiseless Trend (TNT) programs to detect visual-field deterioration.

METHODS: Patients with open-angle glaucoma followed for a minimum of 2 years and a minimum of seven reliable visual fields were included. Progression was assessed subjectively by four masked glaucoma experts, and compared with GPA and TNT results. Each case was judged to be stable, deteriorated or suspicious of deterioration

RESULTS: A total of 56 eyes of 42 patients were followed with a mean of 7.8 (SD 1.0) tests over an average of 5.5 (1.04) years. Interobserver agreement to detect progression was good (mean kappa = 0.57). Progression was detected in 10-19 eyes by the experts, in six by GPA and in 24 by TNT. Using the consensus expert opinion as the gold standard (four clinicians detected progression), the GPA sensitivity and specificity were 75% and 83%, respectively, while the TNT sensitivity and specificity was 100% and 77%, respectively.

CONCLUSION: TNT showed greater concordance with the experts than GPA in the detection of visual-field deterioration. GPA showed a high specificity but lower sensitivity, mainly detecting cases of high focality and pronounced mean defect slopes.

Home-based child development interventions for preschool children from socially disadvantaged families

BACKGROUND
Social disadvantage can have a significant impact on early child development, health and wellbeing. What happens during this critical period is important for all aspects of development. Caregiving competence and the quality of the environment play an important role in supporting development in young children and parents have an important role to play in optimising child development and mitigating the negative effects of social disadvantage. Home-based child development programmes aim to optimise children's developmental outcomes through educating, training and supporting parents in their own home to provide a more nurturing and stimulating environment for their child.

OBJECTIVES
To determine the effects of home-based programmes aimed specifically at improving developmental outcomes for preschool children from socially disadvantaged families.

SEARCH STRATEGY
We searched the following databases between 7 October and 12 October 2010: Cochrane Central Register of Controlled Trials (CENTRAL) (2010, Issue 4), MEDLINE (1950 to week 4, September 2010), EMBASE (1980 to Week 39, 2010), CINAHL (1937 to current), PsycINFO (1887 to current), ERIC (1966 to current), ASSIA (1987 to current), Sociological Abstracts (1952 to current), Social Science Citation Index (1970 to current). We also searched reference lists of articles.

SELECTION CRITERIA
Randomised controlled trials comparing home-based preschool child development interventions with a 'standard care' control. Participants were parents with children up to the age of school entry who were socially disadvantaged in respect of poverty, lone parenthood or ethnic minority status.

DATA COLLECTION AND ANALYSIS
Two authors independently selected studies, assessed the trials' risk of bias and extracted data.

RESULTS
We included seven studies, which involved 723 participants. We assessed four of the seven studies as being at high risk of bias and three had an unclear risk of bias; the quality of the evidence was difficult to assess as there was often insufficient detail reported to enable any conclusions to be drawn about the methodological rigour of the studies. Four trials involving 285 participants measured cognitive development and we synthesised these data in a meta-analysis. Compared to the control group, there was no statistically significant impact of the intervention on cognitive development (standardised mean difference (SMD) 0.30; 95% confidence interval -0.18 to 0.78). Only three studies reported socioemotional outcomes and there was insufficient data to combine into a meta-analysis. No study reported on adverse effects.

AUTHORS’ CONCLUSIONS
This review does not provide evidence of the effectiveness of home-based interventions that are specifically targeted at improving developmental outcomes for preschool children from socially disadvantaged families. Future studies should endeavour to better document and report their methodological processes.

Inference and Declaration of Independence in Task-Parallel Programs

The inherent difficulty of thread-based shared-memory programming has recently motivated research in high-level, task-parallel programming models. Recent advances of Task-Parallel models add implicit synchronization, where the system automatically detects and satisfies data dependencies among spawned tasks. However, dynamic dependence analysis incurs significant runtime overheads, because the runtime must track task resources and use this information to schedule tasks while avoiding conflicts and races.
We present SCOOP, a compiler that effectively integrates static and dynamic analysis in code generation. SCOOP combines context-sensitive points-to, control-flow, escape, and effect analyses to remove redundant dependence checks at runtime. Our static analysis can work in combination with existing dynamic analyses and task-parallel runtimes that use annotations to specify tasks and their memory footprints. We use our static dependence analysis to detect non-conflicting tasks and an existing dynamic analysis to handle the remaining dependencies. We evaluate the resulting hybrid dependence analysis on a set of task-parallel programs.