David -
Any update on how you guys are feeling about PMEMD? Want it to remain
part
of the product in the 8 timeframe? Any specific requests? Current status
is that it "sort of" makes amber competitive with NAMD, with the "sort of"
being an issue of NAMD having better high-end performance on the psc tsc
with quadrics interconnect. I have not heard more from the namd folks,
but
in the polking around I have done since release, I am now pretty convinced
that any additional performance they get is associated with their use of
low-level quadrics i/o, which allows them to efficiently use both rails.
I
actually don't think it is that important; we are at least matching their
scaling for all practical purposes, and anyone that stays with amber does
not do it for a little bit of extra speed. Also, any advantage they have
is
specific to the one machine, and realistically speaking, targetting
specific
machines with low level optimizations requires lots of resources (they
have
about 16 guys, looking at the pubs). I have noticed that they have not
gotten back to me on their sp4 performance, where they have had a bit of
grief doing low level optimizations. Anyway, you should let me know
what's
up. I am moving ahead to playing with new ways to handle electrostatics,
which may or may not work out. We are also in limbo out here on NIH
funding, so I could be doing something completely different in a couple of
months.
Regards - Bob
----- Original Message -----
From: "David A. Case" <case.scripps.edu>
To: <amber-developers.scripps.edu>
Sent: Thursday, November 13, 2003 7:30 PM
Subject: amber-developers: updated "to-do" list for amber8
> Folks:
>
> Below, I have reproduced the amber8/TODO file from last June, and added
> comments about how well the ideas were carried through, and what my
current
> expectations are for the amber8 release. Please let the list know if
you
> see any place where my expectations are incorrect.
>
> I think we should be pretty happy: a very large fraction of the things
we
> had hoped for last summer look like they will actually take place. A
few
> things (X-ray refinement, supporting TINKER and CHARMM force fields)
probably
> won't make it, but there has neverthess been substantial progress in
those
> areas.
>
> Several places say "need docs": this means both documentation, but also
> examples and test suites. Please don't forget that telling others how
to
> use these programs is a key part of the development process.
>
> My current plan is to freeze the code part of CVS on Jan. 1, 2004.
After
> that, any revisions will have to come through me first. I will leave
> the documentation part open until Jan. 20, 2004. Please try to meet
these
> deadlines, and remember that there is a significant lead time in
actually
> producing the physical manuals.
>
> Thanks for all of your work. I am pretty much relying upon each of you
> to take responsibility for "your" things that are not yet done. (I will
> send out a few additional, more specific e-mails about a small number of
> open issues.) Also, please encourage people in your groups to be using
> and testing amber8 as much as possible.
>
> ..regards...dac
>
>
==========================================================================
==
===
>
> "To-do" + wish list for Amber8 (5 June 03)
> Bracketed Comments ("[]") are Nov. 03 estimates of what will actually
get
done
>
>
==========================================================================
==
===
>
> 1. Free energies:
> a. add support for umbrella sampling and WHAM analysis [done, needs
docs]
> b. re-incorporate OWFEG [will not get done]
> c. replica-exchange [Carlos, TEC groups working on this]
> d. deprecate gibbs -- get any essential functionality into sander
[done]
>
> 2 Implicit solvation models:
> a. constant pH simulation capability [done, needs docs]
> b. test Langevin simulations [code in, tests underway]
> c. put in a PB solver/reaction field [done, needs docs]
> d. GB-LES capability [done, needs docs]
>
> 3. Dynamics integrators:
> a. Add Nose-Hoover chains [will not get done]
> b. replace erfc splitting with one closer to 1 at small distances
> (for improved MTS) [Tom D. will consider; probably will not get
done]
>
> 4. NMR/X-ray refinement:
> a. Add in structure factors and their derivatives as restraints
> [not ready for amber8]
> b. put restraints into NMR*/ARIA-like format [will not get done]
>
> 5. LEAP/Antechamber:
> a. keep ResId around, make available to analysis programs [may get
done,
> Scott is looking at this]
> b. Update to gaff/2; check vs. many molecules [done, needs docs]
>
> 6. Allow users to "script" a sander run, rather than just doing one
thing
> a. incorporate Joe's memory database [will not get done]
> b. create a driver with access to (most of) the internal data
structures
> [will not get done]
>
> 7. Speed and parallel efficiency:
> a. Spatial decomposition for parallel MD [done, needs more testing]
> b. reduce the amount of redundant calculation in sander TI calcs
> [probably will not get done]
> c. Use a non-bonded list for GB [will not get done]
> d. Tunings for intel processors; reduction of memory usage/cache
misses
> [done]
>
> 8. New/revised force fields:
> a. Carlos' GA revisions [don't know where this stands]
> b. Yong's new ff03 force field [done, needs docs]
> c. Develop a lipid force field for membrane simulations
> [will not get done]
> d. Integrate more with glycam [should get done, needs docs]
> e. TINKER force fields; better amber polarizable potentials
> [will not get done, but significant progress has been made]
> f. add CHARMM-style out-of-plane torsion; teach LEaP how to output
the
> proper prmtop file *OR* have a lower-level psf_to_prmtop
converter
> [probably will not get done]
>
> 9 . General:
> a. remove rgroup(); replace with atom selection strings from ptraj
> [partially there, but needs significant work; may get done]
> b. combine ptraj/carnal/anal/nmanal [partially done, needs docs]
> c. better integration of sander and roar (esp. QM/MM mixing)
> [PSU group to get this done]
> d. more cleanly divide progs into GPL/non-GPL (or other open source)
> [not done yet, but will probably get done]
>
> 10. Publications:
> a. gaff paper; antechamber paper [gaff paper done, antechamber
still
> in progress]
> b. review for Adv. Protein Chem. (in press) [done]
>
> 11. Documentation
> a. make all tutorials really work; update and add [ha! ha!, but
maybe
> part of this will get done]
> b. put more "science" in the Users' Manual [partially done, needs
much
> more]
> c. updates to the FAQ web pages [see comments to 11a]
>
>
==========================================================================
==
===
>
> Potential significant new features for Amber 8:
>
> spatial decomposition; better parallel peformance [yes]
> X-ray refinement capability(?) [no]
> constant pH simulations [yes]
> GB/LES, better(?) GB [yes]
> Support for new force fields: ff03, glycam, CHARMM, AMOEBA, .... [yes]
> Junmei's Sybyl/GUI [??]
> QM/MM with AM1/PM3 [yes, I think]
> replica exchange, multisander [yes]
> new clustering, other trajectory analsysis routines [yes]
>
>
> --
>
> ==================================================================
> David A. Case | e-mail: case.scripps.edu
> Dept. of Molecular Biology, TPC15 | fax: +1-858-784-8896
> The Scripps Research Institute | phone: +1-858-784-9768
> 10550 N. Torrey Pines Rd. | home page:
> La Jolla CA 92037 USA | http://www.scripps.edu/case
> ==================================================================
>
>
Received on Wed Apr 05 2006 - 23:50:07 PDT
