Citing CHTC Resources
In a Publication
This research was performed using the compute resources and assistance of the UW-Madison
Center For High Throughput Computing (CHTC) in the Department of Computer Sciences. The CHTC is
supported by UW-Madison, the Advanced Computing Initiative, the Wisconsin Alumni Research Foundation, the Wisconsin Institutes
for Discovery, and the National Science Foundation, and
is an active member
of the Open Science Grid, which is supported by the National Science Foundation and the U.S.
Department of Energy's Office of Science.
(Last updated Oct 5, 2014)
For a Grant Proposal
(Feel free to modify the below text, use only certain paragraphs, or contact us for more input or customizable letters of support.)
The University of Wisconsin-Madison (UW-Madison) campus is an excellent match for meeting
the computational needs of this project. Existing UW-Madison technology infrastructure supported
by the CHTC can be readily leveraged, including CPU capacity, network connectivity, storage
availability, and middleware connectivity. But perhaps most important, the UW-Madison has significant
staff experience and core competency in deploying, managing, and using computational technology.
The UW-Madison's Advanced Computing Initiative (ACI) has invested in the CHTC as the primary
provider of shared large-scale computing infrastructure to campus researchers, and all standard
CHTC services are provided free-of-charge to campus researchers.
For high-throughput computing (HTC) capability, UW-Madison maintains many compute clusters
across campus, which are managed via software developed by the UW-Madison's HTCondor Project
distributed computing research group; therefore, these clusters are linked together to share
resources via widely adopted distributed computing technologies. Together these clusters
represent roughly 30,000 CPU cores in support of research. Between 1/1/2015 and 12/31/2015, the
CHTC provided over 300 million CPU hours of computing work. Temporary file space for
large individual files can support up to hundreds of terabytes of total working data.
For single computing runs needing significant memory on a single server, beyond the typical value
of 128 GB, the CHTC maintains
two multi-core servers, one with 1 TB of memory and one with 2 TB of memory. Should these
resources not be sufficient for your project, the CHTC can also engage computing resources from the Open Science
Grid (OSG), also at no cost. OSG is an expanding, NSF-funded alliance of more than 120 universities, national
laboratories, scientific collaborations, and software developers, and CHTC users have automatic access to OSG's
considerable computing and storage resources. Individual users of our high thoughput computing (HTC)
system, including Open Science Grid capacity, can frequently obtain in excess of 200,000 CPU hours per day.
For high performance computing (HPC) capability, the CHTC also maintains
a shared-use cluster of approximately 7000 tightly
coupled cores, with expansion room to increase cluster compute capacity as campus needs
grow and as research groups contribute for dedicated access to a number of cores. Compute nodes have
16 or 20 cores, each, and 64 or 128 GB RAM, with access to a shared file system
and resources managed via the open-source software, SLURM. This cluster, deployed in June 2013
and expanded significantly in summer 2014, delivered nearly 40 million CPU hours between 1/1/2015 and 12/31/2015. The current configuration of
this cluster is the result of intimate collaborations with UW-Madison's Advanced Computing
Initiative (ACI) and fellow computing centers at other campuses through the NSF-funded
Between the ACI, CHTC, and aforementioned HTCondor Project, UW-Madison is home to over 20
full-time staff with a proven track record of making compute middleware work for scientists.
Far beyond just being familiar with the deployment and use of such software, UW staff has been
intimately involved in its design and implementation. Furthermore, the CHTC provides consulting
for the development of robust HTC and HPC research methods, support for grant proposal
development, and a variety of formal and informal training opportunities for users of CHTC resources.
(Other campus resources you may wish to write about:)
The UW-Madison network is currently comprised of a 10GB backbone with 10GB connections to heavy-use
buildings and departments, and 1GB connections to the rest. Redundancy is built into every
network node. An equitable funding model assures that network resources are kept current;
upgrades to 20GB connections and beyond are already being planned. WiscWaves, our high-speed
optical network connection to Chicago, provides researchers with 10GB dedicated research
networks (lamdas). For example, the UW-Madison Department of Physics uses a dedicated lambda to the
High Energy Physics Large Hadron Collider project. UW-Madison has been fundamental to the
establishment of the Broadband Optical Research Education And Science network (BOREAS). This
Regional Optical Network (RON) connects to the CIC OmniPoP in Chicago, providing a high-speed
gateway to various research networks, including Internet2, National Lambda Rail (NLR), ESNet
and other global research networks.
To collaborate with off-campus collaborators, and when local UW-Madison compute resources are not
sufficient, grid middleware connectivity becomes essential. In this area, UW-Madison is an
active participant in NSF XSEDE activities, is the lead institution for the NSF/DOE Open
Science Grid, and runs a Tier-2 computing center for the international CMS LHC experiment.
(Last updated Jul 10, 2015)