MARCC Allocation Request (for PIs only)

For UMD requests:

The UMD HPCC Allocations and Advisory Committee, which already manages allocation requests for Deepthought and Deepthought2, will manage Bluecrab allocation requests in the same manner.
Please follow this link:
http://www.glue.umd.edu/hpcc/getting-access.html#grants

For all other requests, please use this information and the form below to request an allocation from your Deans. If you need help please contact marcc-help@marcc.jhu.edu

As a trial period to determine usage, allocations for the initial 3 months will be made by the appropriate Deans’ office. Please note that although MARCC is a very powerful HPC resource, it will not immediately replace HHPC, XSEDE centers or any other local HPC system.

An allocation is the amount of CPU-hours assigned to Principal Investigators (PIs). Allocations are measured using “Service Units (SUs)”, the use of one core for an hour of wall time. Allocations are awarded for a term on 12 months but the actual allocation of time will be done in quarters. PIs are strongly encouraged to use their allocation on a regular basis.

You can read more about the allocation policy here.

Once a PI has received an allocation from his/her school, users will be able to request accounts subject to PI’s approval. Accounts cannot be shared with other users but data can be shared within groups. Individual accounts will have these directories and allocations:

  • 50 GB for HOME directory per user, backed up
  • 1 TB per group on shared lustre (scratch/work)
  • Extra storage upon request, fees may apply beyond 10 TB
  • Regular compute nodes have 128 GB RAM

If the form has been successfully submitted, you will receive a confirmation email.

Allocation Request Form for PI's

Fields marked with an * are required

Please select one
Note: A standard compute node has 24 cores and 128 GB RAM: equivalent to about 50,000 SU's  over  90 consecutive days. A service unit (SU) is defined as the use of one core for one hour (walltime).

 

Note: Deans will take this into consideration

Note: A GPU node has 2 Nvidia K80 GPUs. These are dedicated nodes therefore your allocation will be charged for the use of 24 cores.

Note: A large memory node has 48 cores and 1024 GB RAM. Your allocation will be charged for the use of 48 cores.

Note: There may be a fee.

Note: There may be a fee