Archive for the 'High Performance Computing' Category
July 22nd, 2009 by Francisco Lopez
-
Greenplanet
Physical Sciences, with support from IAT-NACS, has assembled a high-performance computing cluster for climate modeling and other computational-intensive research.
Called “Greenplanet,” the cluster comprises nodes purchased by faculty in Earth Systems Sciences (ESS), Chemistry, and Physics, and it is expected that Math faculty will also participate. At this time, Greenplanet includes almost 900 CPUs and is still growing.
IAT provides secure, climate-controlled space in the Academic Data Center, system administration services as a team with Physical Sciences IT staff, and consultation on code parallelization and optimization.
According to Assistant Professor Keith Moore of ESS, Greenplanet is “a flexible cluster, suitable for massively parallel complex computations (such as climate simulations), and for smaller-scale use on a single node as a workstation.”
A typical node features 8 64-bit Intel CPUs. Greenplanet features the Load Sharing Facility (LSF) for job management and the Lustre caching file system for extremely high-performance access to the large datasets typical of climate modeling. Two message passing techniques are available for parallel code: OpenMP for communication between CPUs on a node, and MPI for communication between CPUs on different nodes. Greenplanet also has the high-performance Infiniband interlink between nodes for high-speed communications. There is extensive instrumentation available for tuning jobs to optimal execution speed and use of all available computational capacity in the cluster.
Software includes the Climate Systems Modeling package, parallel Matlab, and quantum chemistry packages such as Gaussian and Turbomole.
February 23rd, 2009 by Francisco Lopez
Computer Cluster
Last year, Broadcom graciously donated over 400 compute servers to UC Irvine. While the majority of the servers were distributed to campus researchers, NACS and the Bren School of Information and Computer Sciences have collaborated to bring a new general-purpose campus computing solution to researchers and graduate students at no charge.
Initially, the Broadcom Distributed Unified Cluster (BDUC) is comprised of 80 nodes: 40 nodes with 32-bit Intel processors and 40 nodes with 64-bit AMD processors. Broadcom is expected to donate newer servers over time, allowing nodes to be upgraded. NACS and ICS plan to further expand the cluster as well, subject to available staff and Data Center resources.
BDUC includes standard open-source compilers, debuggers, and libraries; in addition, the MATLAB Distributed Computing Engine (DCE) will soon be available. In the near future, BDUC will offer priority queues for research groups that provide financial support or hardware to the cluster.
BDUC is now available to all faculty, staff, and graduate using your UCInetID and password. To request an account, send an e-mail to bduc-request@uci.edu. A new user how-to guide is available on the NACS website http://www.nacs.uci.edu/computing/bduc/newuser.html.
January 23rd, 2009 by Harry Mangalam
-
Moving Bulk Data
Data transfer is a routine activity for most faculty, whether it’s sharing research data with colleagues, downloading research databases, or backing up vital data. When the volume of data you’re transferring is in the tens or hundreds of megabytes, any tool can get the job done. When you have gigabytes, or tens of gigabytes of data to move, more strategy is called for.
The tool and strategy you should use depends on the kind of data you have, the size of the data, whether you need to do the transfer once or repeatedly, and the computer and tools you’re most comfortable with. Some ideas are outlined below, but NACS’s Research Computing Support maintains a detailed discussion with links to sites from which you can get data transfer tools.
Two basic strategies exist which can reduce the actual volume of data you need to transfer: compression and synchronization. Unless your data is already in a compressed form (say, MP3 files), compression can save a great deal of time and network capacity. Many transfer tools can even do on-the-fly compression. If your files contain sensitive information, you may wish to consider encrypting the data you’re transferring, although this imposes a small time penalty.
The second strategy, particularly when you’re regularly moving the same data, is to use a synchronization tool that recognizes that only part of your data is new and needs to be transferred. This can be particularly convenient if you have an entire directory tree you wish to send over the network.
A final technique which might apply in some cases is to make the best possible use of the network, either by setting up multiple parallel data-transfer streams, or even creating a special-purpose GridFTP node. RCS staff can help you analyze your data transfer needs, choose a method, and set up your system.
RCS staff will also coordinate with NACS Network Engineers to ensure they are aware of research data transfer needs in various campus locations. This will help inform future network upgrade plans. In addition, in a few cases, it may be possible to upgrade network connections to higher speed to support critical research requirements.
November 10th, 2008 by Jessica Yu
CENIC Network
The Corporation for Education Network Initiatives in California (CENIC — http://www.cenic.org/) and UC have been discussing a possible new network infrastructure. The intent is to facilitate ad-hoc, point-to-point, gigabit research network connections among UC campuses and other institutions (including Stanford and USC) connected to CENIC’s High Performance Research (HPR) Network.
This new infrastructure would parallel the existing production network links and could provide two distinctive services: dedicated, low-latency bandwidth to researchers’ labs for special applications, and optical connections for network protocol development or similar activities.
A Zotmail to all faculty recently went out to identify those with needs in this area. Faculty input is sought to guide NACS on how to proceed and at what priority relative to other network needs. For more information, please join the discussion mailing list high-speed-networking@uci.edu .
Meanwhile, if you transfer research data sets over the network, and the speed of doing so is impeding your work, we want to hear from you. Please contact NACS at x42222 or email nacs@uci.edu .
May 14th, 2004 by Dana Roode
NACS hosts and manages the Middle Performance Computing (MPC) “Beowulf” Cluster on behalf of campus researchers who need substantial computational power.
MPC comprises private nodes and shared nodes, including a part-time shared cluster using NACS PC lab systems. MPC systems feature a mix of architectures to provide high computational throughput.
A feature of the MPC service is the opportunity for researchers to join their own systems to the cluster. In exchange for system administration, housing, and 24/7 oversight provided by NACS, researchers allow 25% of their systems to be configured as part of a campus computational resource. (The remaining 75% is configured to be a cluster, or “queue”, dedicated to the owner.) Contributors may, of course, make use of systems designated for campus use.
Contributors also become voting members of the MPC Advisory Board. The purpose of the MPCAB is to advise NACS on the governance, policies, procedures, and technical aspects of the MPC cluster.
Researchers may request accounts on MPC (and other NACS resources) online. Any future changes that impact MPC users will be posted on the MPC website.
NACS also hosts the GradEA Beowulf Cluster for the exclusive use of UCI graduate students.
MPC web site:
http://www.nacs.uci.edu/computing/mpc/
MPC account requests:
http://www.nacs.uci.edu/rcs/resources.html
GradEA web site:
http://www.nacs.uci.edu/computing/gradea/
April 24th, 2003 by Dana Roode
NACS provides and supports various computing resources and services for the UCI community. One noteworthy computing resource, GradEA, has been developed for the exclusive use of UCI graduate students. The intent is to provide graduate students with access to high-speed CPUs, large-data storage capacity, and advanced software.
GradEA consists of 11 dual-CPU Intel Xeon 2.0 GHz nodes, running the Red Hat Linux 8.0 operating system. The most recent hardware upgrade to GradEA occurred in January of this year and more are planned for the coming year. Software available on GradEA includes: Mathematica 4.2, MATLAB 6.5, IDL 5.4, SAS, IMSL, S+/R, the Portland Group compiler suite, MPICH (aka MPI) and Open PBS.
Programs on GradEA can be run in a single, or multi-CPU mode; the cluster network is interconnected by Gigabit ethernet. Users also have access to a disk-storage workspace of 700GB (Gigabytes).
By default, all graduate students have accounts on the GradEA cluster; try logging into gradea.uci.edu with your UCInetID and password. Further information is available at: http://www.nacs.uci.edu/computing/gradea