Using Experimental Data to Improve the Performance Modelling of Parallel Linear Algebra Routines
نویسندگان
چکیده
The performance of parallel linear algebra routines can be improved automatically using different methods. Our technique is based on the modellisation of the execution time of each routine, using information generated by routines from lower levels. However, sometimes the information generated at one level is not accurate enough to be used satisfactorily at higher levels. Therefore, a remodelling of the routines is performed by using (applied appropriately) polynomial regression. A remodelling phase is proposed, and analysed with a parallel matrix multiplication.
منابع مشابه
Parallel Sparse Matrix Computations Using the PINEAPL Library: A Performance Study
The Numerical Algorithms Group Ltd is currently partic ipating in the European HPCN Fourth Framework project on Parallel Industrial NumErical Applications and Portable Libraries PINEAPL One of the main goals of the project is to increase the suitability of the existing NAG Parallel Library for dealing with computationally in tensive industrial applications by appropriately extending the range o...
متن کاملCtadel: A Generator of Efficient Numerical Codes Ctadel: A Generator of Efficient Numerical Codes
Algorithm Templates. According to Kant and Steinberg [82], linear solvers and related algorithms can be cast in the form of so-called algorithm templates. Templates are algebraic structures that capture the essentials of an algorithm, elaborated with problem-specific aspects, and can be transformed into a chosen target language. The SciNapse system [2] uses algorithm templates in the automatic ...
متن کاملPerformance Evaluation of Linear Algebra Routines
In this paper we presents a tool for the dynamic forecast of performance of linear algebra routine as well as communication between clusters. FAST (Fast Agent ’s System Timer) is a software package allowing client applications to obtain an accurate forecast of routine needs in terms of completion time, memory space, and number of communication, as well as current system availability. FAST relie...
متن کاملSolving PDE Problems on Parallel and Distributed Computer Systems Using the NAG Parallel Library
The NAG Parallel Library enables users to take advantage of the increased computing power and memory capacity ooered by multiple processors. It provides parallel subroutines in some of the areas covered by traditional numerical libraries, such as dense and sparse linear algebra, optimization, quadrature and random number generation, as well as utility routines for data distribution, input/outpu...
متن کاملReducing the time to tune parallel dense linear algebra routines with partial execution and performance modelling
We present a modelling framework to accurately predict time to run dense linear algebra calculation. We report the framework’s accuracy in a number of varied computational environments such as shared memory multicore systems, clusters, and large supercomputing installations with tens of thousands of cores. We also test the accuracy for various algorithms, each of which having a different scalin...
متن کامل