NERSC 1998 User Survey Results

These are the complete responses to this question, with a few minor edits.

NERSC is best, better than; favorable evaluation:

• NERSC is the best I know of. My other experiences are the LANL Origins and Avalon.

• Best overall. LANL ACL, LLNL LC, EPA NESC

• NERSC compares very favorably to other T3E sites I have used.

• Very competitive. Excellent computer support and CS expertise. The environments have lots of scripts and programs with all the things one would like to have and never gets around to write them. Other centers: NASA Goddard, San Diego Supercomputing center, CERFACS- France, CNUSC- France, Argonne National Laboratory

• NERSC has the best commitment to maintaining a usable production facility of the centers I have used. Some of them forget that we scientists are less interested in producing code which is a thing of beauty in the eye of the computer scientist, and more interested in number crunching to get (and store) answers. I.e. we are interested in scientific computing, not computer science. Other centers I have used include NCSA, PSC, SCRI, NPAC.

• NERSC in general has more computing power than some of the NSF supported centers(for example, Cornell theory center).

• You are number 1. I would rather not list the other centers used in this comparison;-)

• Nersc is far superior to the center I generally use which is the computers of The Hong Kong Polytechnic University Computing Services

• Very Good. NCAR and LANL Open System.

• NERSC compares favorably to the Cornell Theory Center

• UCSD- Much more help from consultants at nersc. Nersc is much more user friendly.

• [center name deleted] - what a joke! You guys have them beat hands down...

• quite favorably for my needs as compared to PSC or the North Carolina SC.

• Have used ORNL's CCS and Sandia NM. NERSC provides much better support.

• I rank NERSC higher than SDSC

• Better than most. Universities of Texas and Oklahoma, SDSC.

• It is much more efficient than the other CRAYS (J90 and T3E) I have ever used which are located at CIEMAT (SPAIN)

• I think I would rate it as somewhat better compared to the Center for Computational Sciences at Oak Ridge National Lab.

• NERSC has done a better job than ACL at LANL even though the latter is getting better.

• A little better; Minnesota Super Computer Center

• Very favorably. I have also used the NSF supercomputer site at [center name deleted] (which I didn't like) and the Minn. Super Computer institute.

• In comparison with GFDL, NERSC has more and better resources and much better user support. In general, NERSC is more "user friendly".

• Compared to CERN computing center in 1990, I would rate NERSC more favorably

• Emerson Center (Emory). DOE machines are faster than the Emerson Center RISC cluster and less crowded.

• San Diego - NERSC is better

• Compared to the Brookhaven MDS, HPSS at NERSC is 5000X better. I have also used the data archives at CERN. Data access is slower there, but it is possible to access data without using ftp.

• NERSC compares favorably; San Diego

• NERSC could fire 3/4 of its staff at random and still provide vastly better support than [center name deleted].

• I have also used machines at the Pittsburgh Supercomputer Center. I have found your web manuals to be very easy to use and understand. Technical support has been prompt and friendly.

• NERSC compares well with other centers less oriented to general ER usage. I have used LANL, LLNL, and ORNL in earlier MPP centers (as well as ANL).

• Does very well compared to NCSA, for example. You are more responsive and technically capable. Your consultants are truly experts.

• NCSA - NERSC is more reliable and easier to use. Finnish Supercomputing center CSC - about equally good as NERSC

• Much better than CCD at BNL. On a par with CN division at CERN (which really fills a different role, but I am comparing service more than function.

• Generally comparable to SDSC, GSFC, and the late PSC.

• As good as others (Princeton Plasma Physics Laboratory, Brookhaven National Laboratory).

• Good. Better support than Argonne. About the same as LLNL

• Top third. Argonne ACRF, SNL, LANL, NCSA, Pittsburgh Supercomputing Center, Cornell Supercomputing Center, Minnesota Supercomputer Institute, Michigan Computing Center.

• Not bad.

• You responded faster than other centers but were not as accommodating as OAC at UCLA for some things.

• I have also used the CEWES computer center, but I am not as familiar with their services. I rate both NERSC and CEWES very highly.

• Best overall in user support (account management, consulting, etc.) Not as good at simply providing huge chunks of time (compared to ORNL(DOE), SDSC and Illinois (NSF)

• Good in general consulting services, but bad in usability of the machines. Bad queue structure and charging system on the C90.

• [Hmmm, this may get me in trouble!] The other center I use is the ACL at Los Alamos where I have done the bulk of my computing to date. The T3E at NERSC is hands-down a better machine than the O2K cluster at the ACL so that clearly is a big advantage right now. In addition, the ACL is not a ``service'' center, so one works in a minimal support environment. Being smaller, however, the ACL is much more flexible than NERSC, for example, if one needs a lot of computer time just before a big meeting, they can schedule it. Also, with fewer users, turnaround times are not as long as at NERSC.

• In the last 10 years I have interacted mostly with the CERN computing center, which is slightly less bureaucratic and more responsive to the needs of large collaborations, but to be fair, it is dedicated to supplying the kind of computing that I require.

• to be honest I do all my code development at OSC ... if I would have to rely on nersc, I would not have been able to achieve anything reasonable.
  from that point of view: nersc does not support new ideas.

• service level declined since move from LLNL

• I have used the Edinburgh EPCC center. The day to day operating system support staff at EPCC had a better idea of how to fix problems etc.

• EPCC had very fast network connections for ftp and remote login.

• I have experience with computing centers at GSI (Germany) where I worked before. The IBM mainframe was removed and replaced by 2 SP2 systems. In addition we had a rs6000 AIX workstation farm and a mass storage system. That was found to be ideal for our work.
  In addition labs like CERN (European high energy physics lab), DESY (German lab), BNL and FERMIlab here in the US provide the high energy physics community with mass storage and large workstation farms. The community has recognized this as the way to go. On those workstation farms a batch system like LSF is provided to submit jobs.

• NERSC is the only center I have used for many years.

• I have had no such experience.

Back to Survey Summary Page