Computing

The School of Mathematics provides computer facilities for faculty, staff, visitors, post-doctoral researchers, graduate students, and guests.  If you need assistance, please look at our knowledge base of frequently asked questions or email support@math.gatech.edu.

Workstations: We support a variety of desktop and laptop hardware. We have over 175 computers in offices and computing labs. The PCs mostly run Linux and are typically at least 2 Ghz CPUs with 1 Gb of RAM. Many of these are based on 64 bit AMD processors. We also have dozens of Sunray thin clients running on a Sun v40z server with 4 AMD Opteron processors and a total of 8 Gb of RAM.

Labs: We support three computer labs for students. There are two general purpose labs with a total of approximately 35 seats. These labs include complete systems as well as scanners and printers. They are available to both graduate and undergraduate students. The systems include several dual AMD 64 bit Opteron workstations with 4 Gb of RAM, and one dual Opteron with 8 Gb RAM. The remainder of the systems in the labs are dual core dual CPU AMD 64 bit Opteron Linux PCs (4 cores with 4Gb of RAM), with a few Windows and Apple systems, too. The third "lab" is a mobile cart with 28 laptops. This cart can be rolled into any classroom and the laptops are all configured (with Linux) to connect to the campus-wide wireless network. The cart has been used in a variety of settings, including a final exam in numerical analysis (with network access disabled of course)!

File Server: All user data and many software packages are stored on a central file server and shared using NFS (Unix/Linux/Mac OS clients) or SMB (Windows clients). The file server is a failover cluster of dual Opterons running Linux and attached to Gigabit ethernet. The storage capacity is several terabytes of hot-swappable SATA disks in a RAID 5 configuration. All user data (files, email, etc.) and critcial system configuration data is backed up nightly on an iSCSI disk array and later streamed to LTO3 tape stored in an autoloader.

Compute Servers: Large computationally intensive jobs can be run on a variety of multiprocessor shared memory systems with significant resources. These include many dual AMD 64 bit Opteron workstations with 4 Gb or 8 Gb of RAM, and/or a quad AMD Opteron server with 8 Gb of RAM.

Distributed Computing Cluster:The School of Mathematics has a dedicated cluster of PCs for high performance jobs that can take advantage of a distributed parallel architecture. We currently have a combination of nodes. Each node in the cluster is at least a dual AMD 1.2 GHz system with at least 2 Gb RAM. There have been several additional nodes added by individual researchers over the past few years. These extra nodes include a dual dual-core AMD Opteron 2214 based system with 2 Gb RAM and an NVIDIA 8800GTX video card for GPU processing. We have also added a special high memory node which is based on two Opteron 2220 CPUs and has 16 Gb of RAM. Because we operate the cluster as a shared resource, these are also available to anyone in the School who has a valid computational need. Of course, investigators working on a current project which funded the resource have top priority for that resource.

Other Servers: We have several other servers, split between Sun equipment running Solaris and AMD based systems running Linux. These function as our web server, mail server, database server, etc. In addition, we have smaller systems that provide DNS, print services, backup, logging, remote conosle access, authentication, directory services, and automated installs. We have a very advanced installation environment (based on Solaris jumpstart, Linux kickstart, and Windows Unattended Installs) that allows a fully automated installation of operating systems and applications for Linux, Solaris, and Windows.

Network Printers: Several network printers are located throughout the School, and they are available from any workstation. These include duplex printing and color laser printing as well as a digital copier that can be accessed directly from the network and can automatically send PDF files. Most offices also have local desktop printers, and many include duplex capability.