A Review of Data Placement Optimisation for Data-Parallel Component Composition
نویسندگان
چکیده
Constructive methods for parallel programming are characterised by the composition of optimised, parallel software components. This paper concerns data placement, a key cross-component optimisation for regular data-parallel programs. This article is a survey of data placement optimisation techniques. The main contributions are (1) a uniform terminology, which identifies analyses of the problem which have proven fruitful, (2) a taxonomy of versions of the problem, distinguished by the efficiency with which they can be solved, (3) a discussion of open problems, challenges and opportunities for further progress in the area, and (4) a discussion of the significance of these results for constructive methods in parallel programming. We observe, in particular, the role of skeletons in restricting program graph structure to ensure that optimisation is tractable.
منابع مشابه
A Linear Algebra Formulation for Optimising Replication in Data Parallel Programs
In this paper, we present an efficient technique for optimising data replication under the data parallel programming model. We propose a precise mathematical representation for data replication which allows handling replication as an explicit, separate stage in the parallel data placement problem. This representation takes the form of an invertible mapping. We argue that this property is key to...
متن کاملEecient Interprocedural Data Placement Optimisation in a Parallel Library
This paper describes a combination of methods which make interprocedural data placement optimisation available to parallel libraries. We propose a delayed-evaluation, self-optimising (DESO) numerical library for a distributed-memory multicomputer. Delayed evaluation allows us to capture the control-ow of a user program from within the library at runtime, and to construct an optimised execution ...
متن کاملEfficient Interprocedural Data Placement Optimisation in a Parallel Library
This paper describes a combination of methods which make interprocedural data placement optimisation available to parallel libraries. We propose a delayed-evaluation, self-optimising (DESO) numerical library for a distributed-memory multicomputer. Delayed evaluation allows us to capture the control-ow of a user program from within the library at runtime, and to construct an optimised execution ...
متن کاملEmergency department resource optimisation for improved performance: a review
Emergency departments (EDs) have been becoming increasingly congested due to the combined impacts of growing demand, access block and increased clinical capability of the EDs. This congestion has known to have adverse impacts on the performance of the healthcare services. Attempts to overcome with this challenge have focussed largely on the demand management and the application of system wide p...
متن کاملRuntime Interprocedural Data Placement Optimisation for Lazy Parallel Libraries (Extended Abstract)
We are developing a lazy, self-optimising parallel library of vector-matrix routines. The aim is to allow users to parallelise certain computationally expensive parts of numerical programs by simply linking with a parallel rather than sequential library of subroutines. The library performs interprocedural data placement optimisation at runtime, which requires the optimiser itself to be very eec...
متن کامل