Parallel extenstion to Matlab using Java
ptroja/jpar
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
BUILD: 1. To build jPar you need to have JDK (Java Development Kit) with "javac" and "jar" tools installed. Start command line (shell) interface and add JDK tools directory to your PATH enviroment variableif if necessary: c:\jpar>set PATH="c:Program files\java\jdk1.6.0_13\bin\";%PATH% (Windows) Build the jpar.jar Java archive with command "compile.bat" (Windows) or "sh compile.sh" (Unix). NOTE: if you have Ant build tool (http://ant.apache.org/) installed just issue the command: "ant". This will build jpar.jar file in main directory. INSTALL: 1. Copy file .java.policy to home directory (in Windows use "Documents and Settings\Username" directory) or use "install.bat" (Windows) or "sh ./install.sh" (Unix) on every node, where you want to run jPar client or solver 2. Check whether the folder with Java binaries is in the system path writing "java" in the command window. If not, change the path variable, adding this directory. Run one instance of jpar server using "jpar_server.bat" (Windows) or "sh ./jpar_server.sh" (Unix) on node, where you want to run jpar clients 3. Every instance of Matlab has to be started from or have as a current the directory which contains jpar files 4. Each solver has to be started from Matlab session in jpar directory using: >> jpar_solver(['hostname']); where 'hostname' is the name of host where jpar server is running (defaults to localhost) 5. You can test parallelization with the example command: >> a = rand(100,100,10) + i*rand(100,100,10); >> [V,D] = jpar_client('eig', a) 6. To kill solvers use the command: >> jpar_client('kill'); 7. To see free solvers use the command: >> jpar_client('hosts'); REMARK for multiple jPar running on the same machine: To use multiple jPar servers on the same machine please pick up alternative Java RMI port number, other than default 1099. Then run each part of the jPar with this port number: 1. invoke jpar_server with the selected port number as a command line argument 2. start jpar_solver(hostname, port) with selected port number 3. modify rmiRegistryPort value in jpar_client.m script AUTHORS: Marek Majchrowski <M.Majchrowski@elka.pw.edu.pl>, Piotr Trojanek <P.Trojanek@elka.pw.edu.pl>, Andrzej Karbowski <A.Karbowski@ia.pw.edu.pl>. NOT IMPLEMENTED: * support for sparse matrices
About
Parallel extenstion to Matlab using Java
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published