Resources in VO AstroGrid
Each Astrogrid-D resource is running Globus Middleware, Version 4.0.5 or later. On almost all resources a variant of Linux (SL, RHEL, SuSe or Debian) is installed.
Please lookup the actual status of the machines here: Grid-Status::AstroGrid
Workstations:
For serial jobs, experimenting, and interfacing with DGrid,
four 32-bit workstations and
five 64-bit workstations
are available. Most of the 32-bit and all of the 64-bit machines feature at least 2 processors or cores.
All workstations are configured with the fork-jobmanager of Globus,
| machine | arch |
ncpu (cores) |
ram (GB) |
GHz |
|---|---|---|---|---|
| buran.aei.mpg.de | 32 | 2 | 8 | 2.4 |
| supergrid.aei.mpg.de | 32 | 1 | 0.5 | 1.0 |
| gavo2.aip.de | 32 | 2 | 4 | 1.6 |
| gavo3.aip.de | 32 |
4 |
16 |
1.5 |
| astar.aip.de |
64 |
2 |
2 |
3.0 |
| cashmere.aip.de |
64 |
1 |
2 |
3.2 |
| dublin.aip.de | 64 |
2 |
2 |
2.4 |
| photon.aip.de | 64 |
2 |
2 |
3.0 |
| mintaka.ari.uni-heidelberg.de | 64 |
2 |
1 |
2.6 |
Cluster:
Two 32-bit cluster are available, with ~20 processors. Both clusters feature not only the fork-jobmanager but also PBS or Condor for job-submission.
Three (four) 64-bit clusters are availabe, with up to 512 processors, each featuring a PBS jobmanager and a fork-jobmanager.
Two of these clusters have additional specialized hardware for calculating gravitational interaction (GRAPE).
Not all PBS queues are available to grid-users. Usually you are informed about usage at login.
| machine | arch | nodes |
ncpu (cores) per node |
ram (GB) |
GHz |
LRM | Special |
|---|---|---|---|---|---|---|---|
| siemens.babylon.cs.uni-potsdam.de | x86_32 | 17 | 1 | 1.0 | 2.5 | PBS | |
| bladekemper21.informatik.tu-muenchen.d | x86_32 | 2 | 2 | 1.0 | 2.7 | Condor | |
| titan.ari.uni-heidelberg.de |
x86_64 |
32 |
2 |
2.0 |
3.2 |
PBS | GRAPE6a |
| damiana.aei.mpg.de |
x86_64 |
198 |
4 |
2.0 |
3.0 |
SGE |
|
| golowood.mao.kiev.ua |
x86_64 |
8 |
|
|
|
PBS |
GRAPE6a |
| p5io3.rzg.mpg.de | P5 |
19 |
8 / 4 |
- |
1.9 |
LL | OS: AIX |
Storage Nodes:
There are two D-Grid Storage nodes with ~200TB capacity and 30TB Storage located in LRZ.
The Storage nodes are available to the D-Grid VOs either by Globus middleware, or in one case, by dCache, and in the other, by OGSA-DAI.
| machine |
capacity (TB) |
Special |
|---|---|---|
| astrodata.astrogrid-d.org | 100 | dCache |
| alnitak.ari.uni-heidelberg.de | 15 | OGSA-DAI |
There are currently three types of resources available in AstroGrid-D
a) Shared resources:While you can expect resources that belong to a) or b) to be available instantly, there might be obstacles to overcome to get access to resources below c).
The partners allow access to grid-enabled machines of the institutes
b) Sponsored resources in AstroGrid-D
BMBF funded hardware, maintained by AstroGrid-D
c) Sponsored resources in D-Grid
BMBF funded hardware, maintained by other D-Grid institutions
The summary refers only to resources from a) or b).