Welcome to HPCC News page. Posts here will bring attention to new improvements or changes to the HPCC as well as new opportunities and sessions for training. Also available is a calendar of events located underneath HPCC News on the left sidebar.
Introduction to High Performance Computing at OSC
We are planning to offer our popular introductory workshop on OSC resources just once this semester. This two-and-a-half hour workshop will provide an introduction to OSC resources and how to access them, and is intended for new users or researchers who may not already have accounts but are interested in learning about OSC resources. The workshop will be held at 9:30 a.m. on Thursday, Jan. 23 in BALE Theater at the Ohio Supercomputer Center, 1224 Kinnear Road in Columbus.
Introduction to Linux/UNIX workshop
OSC will be offering a workshop for users new to Linux environments, to help you more effectively use our HPC resources. Topics include logging in and out, directory structure, wildcards, paths, file permissions, editors, and more. This workshop will be held on Thursday, Jan. 30 in BALE Theater at OSC and runs from 1:30-4 p.m.
XSEDE HPC Monthly Workshop - Big Data
XSEDE and the Pittsburgh Supercomputing Center will be presenting a one-day "Big Data" workshop onFeb. 4, from 11 a.m. to 5 p.m. Users from around the country will be able to participate in this workshop by registering to attend at their nearest partner site. OSC will be one of those sites. Please visit the XSEDE Portal for more information on this class and to register. You can elect to attend the closest locations in Pittsburgh (Pittsburgh Supercomputing Center) or in Columbus (OSC). XSEDE account is free.
XSEDE14 Call for Participation
The XSEDE14 conference will be in Atlanta, Ga., on July 13-18
, and XSEDE has issued a Call for Participation, including papers in several tracks and a poster session. The first dates to keep in mind are March 15
, when tutorial proposals and paper submissions are due, and April 15
, when poster, BoF, and Visualization Showcase submissions are due, as well as student travel grant requests. For more details, please see the CFP.Blue Waters Graduate Fellowships
Doctoral students who need access to one of the most powerful supercomputers in the world to help complete their academic research might have that opportunity through a new graduate fellowship program managed by the Ohio Supercomputer Center through NCSA at the University of Illinois. Application deadline: Feb. 3, 2014
. For more information, visit: https://bluewaters.ncsa.illinois.edu/fellowships
We have now reopened the clusters, both RHEL5 (arc-login) and RHEL6
(hpclogin). We have also transferred more nodes to RHEL6 cluster, so
now the ratio of the available processors is about 10:90 between RHEL5
and RHEL6, which is why we would strongly encourage you to use the RHEL6
for your daily operations. Currently the RHEL5 cluster is aimed only
for users who still have issues porting their code to the RHEL6 cluster.
Let us know if you miss any software on the RHEL6 cluster and we will get it installed for you.
are also in the process of installing Matlab 2013b, but it is not
running yet, so please continue to use version 2011b or 2012b.
These are the remaining nodes currently available on the RHEL5 cluster:
7 comp nodes (12 procs, 47 gb memory)
32 quad nodes (8 procs, 15 gb memory)
Please note that the scheduler will begin holding jobs that cannot complete before 9am on Jan 8. Please use the command "remaining' (type on the prompt) to check the remaining time before the downtime and adjust your walltime accordingly.
We have scheduled downtime for all the CWRU clusters on Wednesday, January 8 from 9am
. The scheduled downtime is for one day, but we will let you know when the cluster is available again by email, probably earlier than that.
136 hrs 18 mins
"Our Glenn cluster has a MUCH faster throughput for all sizes of jobs. Basically zero wait time. Not all software is as up to date (RHEL-5 vs RHEL-6 on Oakley) but if this is an issue for users, we take the requests and module up whatever is needed.
I HIGHLY recommend running on Glenn for larger jobs I just tested a 256 core job. 1 min queue time.
On the minus side, the Glenn cluster has older hardware, older GPUs, and max memory sizes are smaller than Oakley nodes. Although both clusters maintain the 4GB/core that is the OSC standard. (both have special large memory nodes) Both clusters have all the same storage services. (lustre, gpfs, home directory, etc)"
We have reopened the RHEL6 cluster. Thank you for your patience.
We have finish upgrading the storage/switch OS and also the job scheduler. RHEL5 cluster that is accessed through arc-login.case.edu is now open. We have also now reopened the RHEL6 cluster (accessed through hpclogin, hpcviz and hpctransfer). Thank you for your patience.
Dr. Nathan Morris from EpiBio department at CWRU is giving a talk on "Using GPUs within the R Environment". Please see the details below and mark your calendar if you are interested.
Where: EpiBio Dept (E401)
When: Sept 25, 2013 at 12:00 - 1:00pm
Abstract: Leverage GPU Computing Power for High Dimensional Data Analysis in R
Within the context of high dimensional data analysis, biostatisticians and bioinformaticists are often held back by very real computational limits. Recently it has been shown that graphical processing units (GPUs) have tremendous potential to dramatically push these computational limits. GPUs, as originally developed for high performance computer graphics, are massively parallel devices which excel at performing simple tasks repeatedly. Numerous GPU accelerated software applications in bioinformatics have been written such as sequence alignment algorithms, molecular dynamics simulations and docking algorithms. However, programming for the GPU involves considerable specialized knowledge and is beyond the reach of many biostatisticians. In this talk I will discuss the options for putting GPU power into the hands of biostatisticians within the R computing environment, focusing specifically on our own R package “gmatrix.” I will give several examples of real world applications. Some downsides and cautionary notes about GPU computing will be discussed as well.
Dr. Nathan Morris is a Sr. Instructor in the Department of Epidemiology and Biostatistics at Case Western Reserve University. He is a biostatistician / genetic epidemiologist with a strong background in applied methodological research. Much of his work has focused on modeling complex causal relationships and measurement error in family data using the structural equation modeling (SEM) approach. Dr. Morris is currently funded to produce SEM software to analyze data from a family based study of Mycobacterium tuberculosis infections, and he is also involved in several other large epidemiological studies.
Dr. Morris is also interested in developing methods for RNA-seq expression data. He has written several R packages implementing his methodological work, and he recently released an R package (https://github.com/njm18/gmatrix/
) which exposes the massively parallel computing power of the graphical processing unit within the R environment.
HPC will undergo maintenance on Oct 9, 9am until Oct 11, 9am. These maintenance days will shut down both
clusters as we upgrade our storage system/switch OS, and migrate some of
the nodes from the RHEL5 to the RHEL6 clusters.