Predicting the performance of parallel programs

Experimental Evaluation of BSP Programming Libraries

“…Also, a framework for communications-library independent BSP like programming is presented there. Other studies on the predictability of BSP algorithms are contained in [1,25] (using MPI) and [5] (using MPI and PUB).…”

“…The input and output data are distributed in a 2-dimensional grid of blocks with size n/ √ p × n/ √ p. The BSP running time of this algorithm is shown in Eq. (5).…”

Software Tools for Performance Modeling of Parallel Programs

“…This methodology is based on two tools: CALL and R. CALL [3,5] is a profiling tool to interact with the code in an easy, simple and direct way. It controls external monitoring tools through CALL drivers that implement an application performance interface to those external tools.…”

“…The methodology to obtain analytical performance models is described in section 3. This methodology is based on an enhanced version of the CALL instrumentation tool [3,5] and on the R statistical package [18]. Section 4 shows a case study of the proposed methodology.…”

Analytical Estimation of the Scalability of Iterative Numerical Algorithms on Distributed Memory Multiprocessors

“…Many parallel computation models have been proposed for distributed-memory multiprocessors. The most famous of these models are the BSP model family (see [2][3][4][5][6][7]) and the LogP model family (see [8][9][10][11][12][13][14]). Most of these models are low-level models and require detailed description of the structure of the algorithm to the level of code in a programming language or pseudocode.…”