NERSC 1998 User Survey Results
Question 10. If you were in charge of NERSC what changes would you implement?
These are the complete responses to this question, with a few minor edits. Some responses have
been repeated (to group them under multiple categories).
Provide more cycles / more machines:
- I would make sure that a machine with the speed/memory of the A-machine
were in place before taking down that reliable resource.
- Make getting a high performance PVP machine a top priority.
- Keep the C90!
- A vector machine with tons of memory devoted entirely to HUGE jobs!
- Decrease number of users or increase capacity. Machines are often very crowded and slow.
- More emphasis on SMP architectures. I fear the MPP approach
may be dead in a few years and is not relevant to many computing
projects in DOE.
- I would give myself a CRAY for my personal use.... just kidding!
Better software support:
- I would try to install and maintain more end-user oriented tools like PSPASES, pPCx,
optimization tools, etc.
- Put more effort into system software development -- the old NERSC model (from
Livermore days) was better.
Better / different batch management:
- Longer queues on the T3Es.
- I would increase the queue time limit on the Marie Curie to 10 hours.
- Definitely increase the speed of remote access and expand the batch system.
- do something about easier development work on t3e
- Work on the queue structure and charging system on the C90.
- See above. [account and queuing policies that allow multiple users on the same project to be
able to work in a coherent fashion]
Better/different allocations process:
- Decrease number of users or increase capacity. Machines are often very crowded and slow.
- Permit scientists with scientifically sound projects, but not
currently funded by DOE, to use NERSC.
- change the allocation process and make it more flexible
- Things are pretty good right now, so no major changes (well,
maybe, rationalize allocations process somehow, although this seems a difficult task).
- See 7. and 8. [increase allocations and don't divvy them into monthly allotments]
Improve documentation/training:
- Continue support of average/beginning users as is. Increase support
of computational science research.
- Improve quality of written documentation, eg., hire an editor who
does *not* know anything about computers; this way he/she will
focus on the completeness/correctness/logical consistency
of the language rather than on the contents; also eliminate acronyms
as much as reasonably possible.
More storage, better storage interfaces, better file management, new file
services:
- Change migration and disk space management system!
- Give high priority to improving/replacing HPSS.
Better accounting/charging/account management procedures:
- Stricter adherence to allocation accounting
- Try to adapt the resource allocation of user accounts better to
user needs. This could happen by talking to account users before
setting up the account.
- Managing allocations seemed a bit clunky to me (setcub commands).
- See 7. and 8. [increase allocations and don't divvy them into monthly allotments]
- See above. [account and queuing policies that allow multiple users on the same project to be
able to work in a coherent fashion]
- Work on the queue structure and charging system on the C90.
Keep users better informed; better interactions with users:
- More user oriented, less NERSC oriented.
- improve consultant quality
- Give all the consultants a raise. I really do appreciate those people.
Improve the T3E / other T3E issues:
- SSH on the T3E's for everyone.
- Consider keeping Pierre separate.
- I would not even think to merge pierre and mcurie. My guess it will lower down overall
efficiency and realability.
Faster networks / better response time:
- Definitely increase the speed of remote access and expand the batch system.
Provide a workstation farm; better PDSF support:
- More commodity computing development such as the PDSF linux farm
- I would support and EXTEND PDSF.
Visualization support/software:
- Readily available 4-D graphics and animation.
- Add to research computing facilities to compliment the production
facility. Explore innovative use of non-numerical hardware and
software, e.g. visualization environments.
Increase research / scientific computing support:
- Continue support of average/beginning users as is. Increase support
of computational science research.
- Add to research computing facilities to compliment the production
facility. Explore innovative use of non-numerical hardware and
software, e.g. visualization environments.
No changes needed / no opinion:
- If it ain't broke, don't fix it. NERSC rules!
- Can't come out with any specific suggestions at this time yet.
- Would not change much.
- I would give myself a raise. Kidding again!
- Sorry, I am not.
- I have been working at NERSC for only 6 months or so, and it
is to early for me to formulate an answer at this time.
- give Monica a job
- No strong opinions.
- I don't know.
- n/a
Back to Survey Summary Page