An Nvidia Tesla C870 has been installed in BG1. See the information at this site for an overview of this card. The CUDA programming environment has also been installed in /usr/local/cuda. Documentation is available beneath /usr/local/cuda/doc.

We will see how this works in a BG. If things go smoothly, we will install another Tesla C870 and a couple of GeForce 8800 GT cards in other BGs for GPU code development.

BG2 also has a Tesla C870 installed.

This equipment was donated to us by NVIDIA under their Professor Partnership Program.

The behavior of the departmental Linux printing subsystem was changed as we upgraded to OpenSuSE 10.2. The lpq and lprm commands now deal with print queues on your local machine. This is fairly useless: the client printer software is very quick to send your print job to the system we use as a print server, and an lpq will almost always report an empty local queue (correct, but uninteresting).

We have implemented alternative commands, rlpq, and rlprm, to deal with this problem. The rlpq command will report the state of the print queue on the server; rlprm will let you remove your print jobs from the server.

Examples of use:

rlpq

checks status of your default printer on the print server that this will give the jobid of all jobs in the queue)

rlpq -Pxxx

checks status of the printer named xxx on the print server

rlprm jobid

removes your print job with specified jobid from your default queue on the print server

rlprm -a

removes all of your print jobs from your default queue on the print server

rlprm -Pxxx jobid

removes your print job with specified jobid from the queue for the printer named xxx on the print server

Matlab R2007b has been installed on all departmental systems. A simple "matlab" should invoke it on the command line on Linux systems; on Macs , matlab is in a folder in /Applications.

The follwoing sub-products are available:

• Simulink
• Bioinformatics Toolbox
• Curve Fitting Toolbox
• Control System Toolbox
• Data Acquisition Toolbox
• Distributed Computing Toolbox
• Signal Processing Blockset
• Instrument Control Toolbox
• Image Processing Toolbox
• SimMechanics
• Optimization Toolbox
• PDE Toolbox
• RF Toolbox
• Stateflow Coder
• Signal Processing Toolbox
• Symbolic Math Toolbox
• Simscape
• Statistics Toolbox
• Wavelet Toolbox

The Linux boxes named bg1, bg2, bg3, bg4 and bg5 are full-fledged departmental Linux systems.

bg1, bg2 and bg3 have two dual core 3.8GHz EM64T Xeon processors and 8GB RAM.

bg4 and bg5 have 3.6GHz EM64T processors and 4GB RAM.

• Please don't run background jobs on lab systems any more.
• Don't do routine computing on the bg[12345] systems.
• If you are doing a lot of i/o, allocate files in the /scratch directory on the bg system you are using. Files in bg[12345]:/scratch are deleted if not accessed for 5 days, and there are no backups of bg[12345]:/scratch.
• The bg[12345] systems are fully networked. If you want to run MPI jobs on them, you may. Network connections between the bg's are full-duplex switched Gigabit Ethernet.

/home/scratch has grown in size by a factor of 20. It is now a 1.4TB Raid-5 box. Although we do not back up /home/scratch to tape, data should be safe as long as there is no coincident failure of 2 or more drives of the RAID.

There is no quota on /home/scratch, so please be considerate of the other people in the department when using this disk. To see how much space you are using on this file system, do the following:

cd /home/scratch/USERID
du -sh .

where /home/scratch/USERID is your directory on this file system.