|
|
|
Computing activities of the RMKI Virgo Group
What we have for computing
Storage
- Currently a ~40 TB disk server dedicated for the Virgo Collaboration is available. This is a DPM (Disk Pool Manager) server.
- Data is accessible via the
rfio , gridftp , http , htttps , srm protocols.
- Server name:
grid253.kfki.hu with alias virgo-se.kfki.hu
Grid User Interface
- A shared user interface is available in the KFKI AFS
 system.
- To use the UI source the configuration file:
. /afs/kfki.hu/project/virgo/gliteUI/current/etc/profile.d/grid-env.sh
- If you don't have AFS account for our cell contact: gridadm@rmki.kfki.hu
- If you don't have a digital certificate, please read this.
GPU machines
-
grid251.kfki.hu,
-
opal4.rmki.kfki.hu
The following GPUs are available:
- 3 nVidia, GTX 295
- 3 ATI Radeon 5970
- 2 nVidia tesla C2050
In case of interest test login accounts can be requested. Please send an e-mail to Gergely.Debreczeni@rmki.kfki.hu
What we are/were doing
Virgo data transfer issues
Feasibility study of data transfer from Cascina with Grid tools
Read the description on this page.
Running analysis pipelins on the EGEE Grid
A feasibility study on how to use EGI Grid resources for the execution of gravitational
wave search pipelines. Read the implementation plan here.
Running CBC analysis pipelins in Bologna computer center
The CNAF computer center is part of the EGI Grid, as such its resources can be
accessed only through the Grid middleware and Virtual Organisation. The CBC inspiral pipeline
software is heavility based on the condor batch system which is not supported and not used by the
majority of EGI Grid sites. The goal of this porject is to set up a framewrok which allows
the execution of CBC and CW analysis pipelines on Grid resources without
modifying the software already in use. Read the description here
Hough-transformation implemented on GPUs
The Hough transformation is used for the search of gravitational waves emitted by rotating neutron stars. The sensitivity of
the search is computationally bounded. The goal of this project is to implement and paralellize the algorithm on GPU
devices which could possible results a significant improvement of the search sensitivity. Read the description here. |
