Long ago, when everyone in MRSEC lived in the Research Institutes building at 5640 S. Ellis Ave., room L-114 of the "Low-Temp" area of the building was a public workstation lab that allowed researchers and students to have access to Sun and SGI UNIX workstations using their JFI login ID's, which were based on NIS rather than LDAP at the time.

As time went on, the Suns were more and more replaced by ordinary PC's which were faster and cheaper, and Solaris and IRIX gave way to Linux. Of those three Linux machines in the lab, the one known as "joule" became the one everybody stayed logged in on, checked their mail on, coded scripts on, and worked on LaTeX documents. When they weren't actually in the Lab, they would be connected to joule from elsewhere on or off campus via SSH login, remotely. Joule, fresnel, and alder all were fairly similar to eachother, and all mounted the same shared disk from jfihome, the home directory server, so in that respect, there wasn't really much reason for prefering joule over one of the others. Still, joule seemed to serve as the de facto shell server and electronic congregating place for MRSEC users, partially out of the impression that many seemed to have then (and even now sometimes) that their files were actually on joule, and that joule was "the server." (Partially, this map is intended to help users understand the relationship between the machines better to avoid a misunderstanding like this.)

When the new Gorden Center (GCIS) building was completed, room E124 became the new MRSEC Computer Lab, and it was quickly termed by everyone the "Fishbowl" because it had glass pane doors and walls on three sides, and it was located in the middle of JFI's space in the building where everyone could watch you work in there all day. (Somewhat not surprisingly, most of our usage on the Fishbowl computers is via remote network connection rather than desktop login these days.) Enough people now know it by the "Fishbowl" name that there are now few who even remember it as the MRSEC Lab, or know that it is still paid for by MRSEC.

One unfortunate tradition that is very much alive however from the days when we were in RI Low-Temp is the use of the three machines for long-running computation and simulations. There was a time when doing this made sense, since at one time, getting access to a Sun Sparcstation was more computational power than most people dreamt about. The lab computers were used for research simulations because they were, more or less, the best machines in the house.

As PC's overtook Sun hardware in cost effectiveness and gained computational power as well, they became effective computational machines too, except that before long, we also had computational servers in the server room for that. Some professors even had their own research clusters, made of dozens of such machines ganged together. The worst part is that when users were logged in on the Fishbowl computers running simulations, it consumed memory and CPU power, and made the desktop sluggish for anyone else to use at the same time. Many people complained of trying to use the Linux machines in the lab and finding that they were bogged down by someone's computational project running in the background.

We now have a computational server named pondermatic available to users wanting to run simulations and other programs which may potentially saturate a machine's capacity, rather than running them on the Fishbowl computers. Also, depending on which professor a user is affiliated with, many of our faculty members have computational servers and research clusters of their own, which are more capable than pondermatic, and more dedicated to only the users of their respective groups and their projects.

There will of course still be times when pondermatic is full of users, or the Fishbowl machines are, for some other reason, the most expedient place to run a computational job in a given moment. In those cases, we at least ask that users run their programs with this command:

where "COMMAND" is whatever the program command was going to be otherwise. In other words, if you want to launch a Matlab session, and you know there will be intense long-term computation from this (for hours, days, etc.), you should launch it with 'nice -n +8 matlab &', not just 'matlab &'. This will help immensely with keeping the machine interactive and usable for desktop users who sit down in the lab while your computational project is running, because it basically tells the system not to let your job hog the whole machine.

Please see our Frequently Asked Questions section for more information about our desktop Linux environment.