Computing Support
Computing Support
Computing Support Home Accounts and Passwords Software Clusters E-mail Scanners and Printers Web Services Education Other Services Administrative FAQs
Clusters

There are two clusters available to the Math Dept users:

The qwe cluster in Wean Hall room 6120 contains the computers qwe1-qwe8.math.cmu.edu that work with the math accounts. A key can be obtained from the Math main office. Please do not power down or reboot these computers since there are other people using them remotely. These computers are intended for general purpose use. They offer easy access to:

    • most scientific and office applications,
    • printing and scanning,
    • a departmental email address username@math.cmu.edu which can be filtered and managed locally by the user,
    • a personal website http://www.math.cmu.edu/~username/ showing the files in the public directory on the cluster.

The parallel cluster is a shared resource with the Physics Department and is not appropriate for large or very long simulations. Please use it mainly for development and small test jobs.
If your code contains large amounts of I/O (read/write to/from files), please create a directory for you inside /data/ which is located on the local drive, and place the input/output files there, as opposed to your home directory which is on a network drive. Otherwise, yous code and the entire cluster will run much slower.
The parallel cluster contains the following computers:

    • fp80-fp87.math.cmu.edu for general purpose computing,
    • fpg1.math.cmu.edu contains two Nvidia GeForce GTX 1080 Ti (3584 cores each) GPUs, fpg2.math.cmu.edu contains one Nvidia GeForce GTX 1080 and one Nvidia GeForce GTX 1080 Ti (3584 cores each) GPUs, fpg3.math.cmu.edu contains two Nvidia GeForce RTX 2080 Ti (4352 cores) GPUs, fpg4.math.cmu.edu contains one Nvidia GeForce RTX 3070 GPU (5888 cores) and 576GB of RAM all can be used for GPU computing using CUDA, and
    • fp48.math.cmu.edu has 80 cores and 256GB of RAM, and is appropriate for large memory jobs.
    • fp64.math.cmu.edu has 64 cores and 128GB of RAM, and is appropriate for parallel programming development.

The parallel cluster can be accessed remotely via ssh, using the math accounts. To run processes while disconnected from the account, please use start them through "byobu", "tmux" or "nohup".