archive-rs.com » RS » S » SCL.RS

Total: 94

Choose link from "Titles, links and description words view":

Or switch to "Titles and links view".
  • Facilities
    Grid site in Serbia it is the Tier 0 site in the Serbian Grid infrastructure providing all core services and management of national AEGIS Virtual Organization All computing and core service nodes at AEGIS01 IPB SCL Grid site are interconnected by the star topology Gigabit Ethernet network through three stacked high throughput Layer 3 switches each node being connected to the switch by two Gigabit Ethernet cables in channel bonding In terms of storage resources AEGIS01 IPB SCL provides 27 TB of disk space to the Grid community The site uses Scientific Linux 5 4 with gLite 3 2 middleware and supports national AEGIS regional SEE SEEGRID SGDEMO METEO ENV SEISMO pan European ATLAS and CMS and DESKTOPGRID Virtual Organizations AEGIS07 IPB ATLAS is the second largest Serbian Grid site is based on 128 Intel Xeon processors with 32 bit architecture and the total of 96GB of RAM The site supports several virtual organizations but is mainly dedicated to the ATLAS VO community of the CERN LHC experiment AEGIS08 IPB DEMO Grid site is used purely for educational training purposes It is based on Xen virtual machines deployed on a single node with two Intel Xeon Quad core CPUs with 16 GB of RAM It is used in various training events for demonstration of installation and configuration of different Grid services Local cluster consists of several nodes based on the new computing architectures IBM POWER6 PowerXCell the latest Intel and AMD Opteron CPUs The hardware is located in one IBM BladeCenter and in several separate 1U rack mount servers IBM BladeCenter has three kinds of servers two Intel Xeon E5405 two POWER6 and two IBM PowerXCell 8i processors within the H type chassis and one server based on latest Intel Nehalem Xeon processor with InfiniBand interconnection In terms of storage

    Original URL path: http://www.scl.rs/scl-facilities/facilities (2016-02-09)
    Open archived version from archive

  • Services
    Organizations BDII WMS LB VOMS LFC MyProxy SCL Repository Service SCL Repository Service provides RPMs produced by SCL and within projects SCL participates in as well as a collection of mirrors of FTP web and rsync sites of interest to academic users SCL SVN ViewVC service Scientific Computing Laboratory provides and maintains the source code repository dedicated to Serbian academic open source development and projects SCL participates in News 12

    Original URL path: http://www.scl.rs/scl-facilities/services (2016-02-09)
    Open archived version from archive

  • Tools
    simple web interfaces developed for Grid site monitoring and integrated presentation of the results provided by various monitoring tools Some of these scripts can be deployed on any general purpose computing cluster without the involvement of gLite middleware scl scripts The set of scl scripts is developed in the BASH scripting language It is intended to simplify the management of medium or large computer clusters Using these scripts it is possible to execute the same command on all nodes or to distribute the files on many nodes in an automated way The set scl scripts consists of five scripts and one configuration file per group of nodes or servers WatG Browser The WatG Browser What is at the Grid Browser is a web based Grid Information System GIS visualization application providing detailed overview of the status and availability of various Grid resources in a given gLite based e Infrastructure It is able to query and present data obtained from Grid information systems at different layers from local resource information system for a particular Grid service GRIS to the Grid site information system site BDII and to the top level information system for the whole Grid infrastructure top level BDII gFinger

    Original URL path: http://www.scl.rs/scl-facilities/tools (2016-02-09)
    Open archived version from archive

  • Monitoring
    design targeted at federations of clusters It leverages widely used technologies such as XML for data representation XDR for compact portable data transport and RRDtool for data storage and visualization It uses carefully engineered data structures and algorithms to achieve very low per node overheads and high concurrency Pakiti Pakiti tool provides a monitoring and notification mechanism for checking the patching status of installed packages on an RPM based Linux system It is using a client server model in which clients and servers are exchanging information using HTTP S SAM SAM Service Availability Monitoring is a framework used in EGEE for the monitoring of production Grid sites It provides a set of probes which are submitted at regular intervals and a database that stores test results In effect SAM provides monitoring of grid services from a user perspective BBmSAM BBmSam is a web application implemented in PHP and relying on a MySQL database for data storage This tool has been tested under different web servers Apache Microsoft IIS and can be used with any web server supporting PHP at list through CGI GStat GStat is an application designed to monitor EGEE LCG compatible Information Systems Its purpose is to detect faults verify the validity and display useful data from the Information System WatG Browser The WatG Browser What is at the Grid Browser is a web based Grid Information System GIS visualization application providing detailed overview of the status and availability of various Grid resources in a given gLite based e Infrastructure It is able to query and present data obtained from Grid information systems at different layers from local resource information system for a particular Grid service GRIS to the Grid site information system site BDII and to the top level information system for the whole Grid infrastructure top

    Original URL path: http://www.scl.rs/scl-facilities/monitoring (2016-02-09)
    Open archived version from archive

  • PARADOX Cluster User Guide
    and canceling is the same as for other previously described types of jobs If the job binary was compiled with intel compiler than the appropriate intel modules should be loaded e g module load intel 14 0 1 4 2 4 Hybrid job submission The following example shows a typical hybrid job submission script bin bash PBS q standard PBS l nodes 4 ppn 16 PBS l walltime 10 00 00 PBS e PBS JOBID err PBS o PBS JOBID out module load openmpi 1 8 2 export OMP NUM THREADS 16 cd PBS O WORKDIR chmod x prog mpirun np 4 npernode 1 bind to none prog It is important to take note of the third line in this script which specifies how the processes should be laid out across compute nodes It is important to adjust the ppn parameter in accordance with OMP NUM THREADS environment setting to avoid oversubscribing a node The mpirun parameter npernode also controls the actual number of processes that will be assigned to a node because the PBS line only serves for the resource allocation Another detail is the bind to none parameter which is needed for OpenMPI versions from 1 8 on which allows threads to spread across the cores For more information about process and thread mapping please see mpirun documentation and the map by parameter 4 2 5 CUDA single node job submission CUDA jobs are very similar to previous types except that cuda module should be loaded and since login node has no GPU these programs can only run on compute nodes Here is an example of an CUDA job submission script bin bash PBS q standard PBS l nodes 1 ppn 1 PBS l walltime 10 00 00 PBS e PBS JOBID err PBS o PBS JOBID out cd PBS O WORKDIR chmod x prog module load cuda 5 5 prog 4 2 6 CUDA MPI Job submission CUDA jobs are very similar to previous types except that cuda module should be loaded and since login node has no GPU these programs can only run on compute nodes Here is an example of an CUDA MPI job submission script bin bash PBS q standard PBS l nodes 4 ppn 1 PBS l walltime 10 00 00 PBS e PBS JOBID err PBS o PBS JOBID out cd PBS O WORKDIR chmod x prog module load openmpi 1 6 5 module load cuda 5 5 mpirun prog 4 3 Compiling 4 3 1 Compiler flags C C Intel compilers icc and icpc Compilation options are the same except for the the C language behavior icpc manages all the source files as C files whereas icc makes a difference between both of them Basic flags o exe file names the executable exe file c generates the correspondent object file Does not create an executable g compiles in a debugging mode I dir name specifies the path where include files are located L dir name specifies the path where libraries are located l lib name asks to link against the lib libname library Optimizations O0 O1 O2 O3 optimization levels default O2 opt report generates a report which describes the optimization in stderr O3 required ip ipo inter procedural optimizations mono and multi files fast default high optimization level O3 ipo static ftz considers all the denormalized numbers like INF or NAN as zeros at runtime fp relaxed mathematical optimization functions Leads to a small loss of accuracy Preprocessor E preprocess the files and sends the result to the standard output P preprocess the files and sends the result in file i Dname value defines the name variable M creates a list of dependence Practical p profiling with gprof needed at the compilation mp mp1 IEEE arithmetic mp1 is a compromise between time and accuracy Fortran Intel compiler ifort Fortran compiler Basic flags o exe file names the executable exe file c generates the correspondent object file does not create an executable g compiles in debugging mode R E Debugging I dir name specifies the path where include files are located L dir name specifies the path where libraries are located l libname asks to link against the lib libname library Optimizations O0 O1 O2 O3 optimization levels default O2 opt report generates a report which describes the optimization in stderr O3 required ip ipo inter procedural optimizations mono and multi files fast default high optimization level O3 ipo static ftz considers all the INF and NAN numbers as zeros fp relaxed mathematical optimization functions Leads to a small loss of accuracy align all fills the memory up to get a natural alignment of the data pad makes the modification of the memory positions operational Run time check C or check generates a code which ends up in run time error ex segmentation fault Preprocessor E preprocess the files and sends the result to the standard output P preprocess the files and sends the result in file i Dname value defines the name variable M creates a list of dependences fpp preprocess the files and compiles Practical p profiling with gprof needed at the compilation mp mp1 IEEE arithmetic mpl is a compromise between time and accuracy i8 promotes integers on 64 bytes by default r8 promotes reals on 64 bytes by default module dir send read the files mod in the dir directory fp model strict Tells the compiler to strictly adhere to value safe optimizations when implementing floating point calculations and enables floating point exception semantics It might slow down your program Please refer to the man pages of the compilers for more information 4 3 2 Compiling MPI program Here is an example of a MPI program include stdio h include mpi h main int argc char argv int num procs my id int len char name MPI MAX PROCESSOR NAME MPI Init argc argv find out process ID and how many processes were started MPI Comm rank MPI COMM WORLD my id MPI Comm size MPI COMM

    Original URL path: http://www.scl.rs/PARADOXClusterUserGuide/ (2016-02-09)
    Open archived version from archive

  • Cooling System
    Links Search SCL Cooling System SCL LAB Cooling System 64bit Cluster Rack Cooling System 32bit Cluster Rack Cooling System News 12 February 2016 SCL Seminar by Jelena Smiljanic 12 January 2016 In memoriam Nikola Buric 28 December 2015 SCL Seminar by Milos Radonjic 25 December 2015 SCL Seminar by Milan Radonjic 18 December 2015 SCL Seminar by Nenad Vukmirovic 11 December 2015 SCL Seminar by Marija Mitrović Dankulov 09 December

    Original URL path: http://www.scl.rs/scl-facilities/10-infrastructure/facilities/356-cooling-system (2016-02-09)
    Open archived version from archive



  •