High-Performance Computing

What is high-performance computing?

Are your models taking a couple days to run?  Are the data you're employing in your research too large to manage on your own lab equipment?  Your work might benefit from the large-scale compute clusters available through University IT's Research Computing group.

Not sure where to begin? 

To learn more about these services and how they can be employed for social science research, you can contactjquan42 [at] stanford.edu (subject: High%20Performance%20Computing%20Inquiry) ( IRiSS-info[at]stanford[dot]edu).

High Performance Computing at Stanford

ITS has created a tiered system of shared compute environments, known collectively as FarmShare, for use by Stanford-affiliated researchers.  The primary HPC clusters are FarmShare and Sherlock.  

FarmShare

FarmShare is available to all SUNet ID holders and is intended for students to be able to test and refine models without the large expenditures involved with HPC computing in the cloud. More information is available at the FarmShare wiki.

Sherlock

Sherlock is reserved for faculty-sponsored projects and therefore is a more limited resource.  Sherlock servers have two high-RAM nodes as well as GPU clusters which can be very useful to R and other statistics users. More information is available at the Sherlock wiki.

More information

Additional information about these environments, technical specifications, how to connect to and use them, and more can be found at Stanford University IT services.