Name: Matrix multiplication with Files, reproducibility example, without data persistence Contact Person: support-compss@bsc.es
Access Level: public
License Agreement: Apache2
Platform: COMPSs

Description

Matrix multiplication is a binary operation that takes a pair of matrices and produces another matrix.

If A is an n×m matrix and B is an m×p matrix, the result AB of their multiplication is an n×p matrix defined only if the number of columns m in A is equal to the number of rows m in B. When multiplying A and B, the elements of the rows in A are multiplied with corresponding columns in B.

In this implementation, A and B are square matrices (same number of rows and columns), and so it is the result matrix C. Each matrix is divided in N blocks of M doubles (N hardcoded to 2, and M hardcoded to 8). The multiplication of two blocks is done by a multiply task method with a simple three-nested-loop implementation. When executed with COMPSs, the main program generates N^3^ tasks arranged as N^2^ chains of N tasks in the dependency graph.

Reproducibility

To reproduce the exact results of this example, follow the instructions at the Workflow Provenance section at COMPSs User Manual, WITHOUT data persistence, PyCOMPSs application.

Execution instructions

Usage:

runcompss --lang=python src/matmul_files.py inputs_folder/ outputs_folder/

where:

• inputs_folder/: Folder where A and B matrices are located
• outputs_folder/: Folder with the resulting C matrix

Execution Examples

runcompss --lang=python src/matmul_files.py inputs/ outputs/
runcompss src/matmul_files.py inputs/ outputs/
python -m pycompss src/matmul_files.py inputs/ outputs/

Build

No build is required