2006 User Survey Results

Software

Legend:

Software Satisfaction - by Score

7=Very satisfied, 6=Mostly satisfied, 5=Somewhat satisfied, 4=Neutral, 3=Somewhat dissatisfied, 2=Mostly dissatisfied, 1=Very dissatisfied

Software Satisfaction - by Platform

7=Very satisfied, 6=Mostly satisfied, 5=Somewhat satisfied, 4=Neutral, 3=Somewhat dissatisfied, 2=Mostly dissatisfied, 1=Very dissatisfied

Comments about Software:   27 responses

General (Cross Platform) Software Comments:   9 responses

It would be great to have more up-to-date versions of quantum chemistry packages running at NERSC.

It would be great to add more support for highly parallel molecular dynamics code, most notably NAMD by Klaus Schulten's group.

I think the main difficulty that I run into is not having an up to date version of Python available on all the machines. I would like to see versions: 2.3.6, 2.4.4, 2.5. These are the latest stable version of Python for each of the major releases.
The other things that I would *really* like is to have more modern MPI implementations - specifically ones that support the MPI-2 spec.

The GNU autotools for building software (autoconf, automake, etc) are frequently out of date which necessitates installing your own version to build some piece of software. Given that these are so commonly used, they should be kept up to date.

Compilers are the bane of our existence. NERSC is no worse than any other site and probably slightly better. The failure of compilers to be ansi compliant is not something I expect to be fixed any time in the near future. Indeed the reverse seems more likely. Perhaps NERSC with DOE behind it could be a leader. Certainly NSF is unlikely to. Otherwise how could they seriously propose Petascale computing.

Debugging with Totalview still seems more painful than necessary --- in parallel mode it is not fun at all (although I have not used it much in parallel mode in FY2006)

The software on all the computers is excellent!

Software resources at NERSC are excellent especially for research in mathematical and physical sciences. NERSC makes special efforts from time to time to upgrade the software and users are advised about the upgrades,etc. in sufficient details. NERSC deserves thanks from the users for their efforts to provide most recent upgrades.

Comments by Bassi Users:   3 responses

CHARMM performance on bassi is worse than that on jacquard, though bassi charges more than jacquard. I don't know whether it is bassi's problem or CHARMM's problem.

I wish there were a way to run some jobs for more than 24 hours on a small number of processors.

I would like need it to be available on Bassi.

Comments by DaVinci Users:   4 responses

On Davinci, some Fortran library is needed.

I don't use DaVinci. The /project directory has been flakey. I don't want to move data around. I would like to see Trilinos installed.

The only software I use a lot on NERSC machines is MATLAB. Being able to run multiple MATLABs simultaneously on DaVinci, and fairly quickly, has been a huge help to my research program. If there is any way to run MATLAB code (regular code, not parallelized in any way) faster I would like to know about it. Overall I am very satisfied with the resource as it has allowed me to do computations that I would not otherwise have been able to do.

... So far I have been enjoying the various options of visualization softwares (mostly AVS and IDL) available in DaVinci. However, one of the major simulation code I have been recently using, the NIMROD code, has been designed to have its data output format work mostly with the visualization package Tecplot. Tecplot is a commonly used commercial visualization package that is well known for its easy accessibility and short learning curve. Unfortunately it is not available on DaVinci. I requested the consideration of installation of Tecplot on DaVinci about a year ago, based on not only the need from my own project, but also from the more important fact that the installation of Tecplot will benefit a large pool of NERSC users who are also users of the NIMROD code, which is one of the two major fusion MHD codes supported by DOE Office of Fusion Science. Yet my request is still under "evaluation" after nearly a year. I would like to take the opportunity of this annual survey to reinstate my request and concern about this request.

Comments by HSS Users:   1 response

The HPSS interface options are shockingly bad. e.g. Kamland had to resort to writing hsi wrappers to achieve reasonable performance. The SNfactory wrote a perl module to have a standard interface within perl scripts rather than spawning and parsing hsi and htar calls one by one. Even the interactive interface of hsi doesn't have basic command line editing. htar failures sometimes don't return error codes and leave 0 sized files in the destination locations. NERSC should provide C, C++, perl, and python libraries for HPSS access in addition to htar, hsi, etc. The HPSS hardware seems great, but the ability to access it is terrible.

Comments by Jacquard Users:   6 responses

Need latest version of NWChem to be installed in Jacquard.

Porting some code to the pathscale compiler has been problematic on jacquard.

The PathScale compilers on Jacquard, particularly the Fortran one, have been quite awkward/awful for us to use. I keep on running into a variety of problems compiling and running our codes on Jacquard with PathScale, which are absent in other machines such as Bassi and Seaborg. Similar experience also applies to the mvapich library on Jacquard. ...

Even if it's heretical, please put Intel compilers on Jacquard. Some software does not support PathScale-specific options and will not compile. I have found Intel compilers, even on AMD machines, to be the most reliable and high-performing compilers for all of my programs.

Interactive debugging of large parallel jobs remotely is difficult even with Totalview, due to network lags and the opacity of the PETSc library. It is impossible on machines such as Jacquard on which there can be long delays in the launch of "interactive" jobs.

I haven't used the system much since the cmb module was upgraded, but my initial impression previously was that support for many quite standard libraries was not immediately apparent. (fftw3, gsl, boost, ATLAS, CBLAS, LAPACK). It is particularly surprising that the AMD math library interface is not compatible with the standard CBLAS interface, but that's obviously not your fault. I think the net result, though, is there are many many copies of these libraries floating around, which users have individually compiled themselves so their existing code would work. On the whole, though, the development support is still excellent.

Comments by PDSF Users:   2 responses

We recently switched operating systems on PDSF and are now using Scientific Linux 3.02. Unfortunately, the default installation is fairly bare bones, lacking any kind of graphics software to look at PDF, PNG, GIF etc.

The "RedHat 8 for STAR" CHOS environment on PDSF lacks many small yet highly useful tools, like Midnight Commander for instance. As a result it is necessary to switch to the default profile, which slow things down and makes it impossible to use STAR-specific components together with such tools.

Comments by Seaborg Users:   4 responses

I understand the reason why NERSC has to remove imsl from seaborg. But, I am not happy about this action.

The OS of Seaborg seems a bit clunky. For example, users can't use the up arrow to get most recent commands, and doesn't do automatic completion.

As earlier, the scope of our project is served well by the current setup at NERSC.

On seaborg, compiling C++ with optimization is very slow.