Re: [AMBER] CUDA MPI issues

From: Biplab Ghosh <ghosh.biplab.gmail.com>
Date: Thu, 26 May 2016 11:30:53 +0530

Dear Dr. Ross Walker and Amber experts,

Thank you very much for your suggestion. The issues are as follows:

1. pmemd.cuda.MPI is not working in my system:
   ==================================

I recompiled my amber14 using (earlier I had compiled with gnu compiler)-
CUDA Version 7.5.18
Intel parallel studio XE 2016 update2
openmpi-1.10.2

Compilation went smooth but some of the tests failed. I have attached the
test logs and diffs (file: 2016-05-24_16-51-23.*). To test individual GPU, I
ran the code

https://dl.dropboxusercontent.com/u/708185/GPU_Validation_Test.tar.gz

The results are also attached.

Then I was trying to run the basic tutorial:
http://ambermd.org/tutorials/basic/tutorial0/ using the command:

mpirun -np 2 pmemd.cuda.MPI -O -i 01_Min.in -o 01_Min.out -p prmtop -c
inpcrd -r 01_Min.rst -inf 01_Min.mdinfo

The errors are attached in the file: pmemd.cuda.MPI.err

MPI library for parallel GPU (MVAPICH2 v1.8 or later / MPICH2 v1.4p1 or
later recommended) has not been installed. Instead, I am using openmpi.

Please help me to rum in parallel gpu.

My second problem is as follows:

2. p2p not working between the GPU 0,1
   ===========================

The system was purchased from supermicro and details are available
at the following URL. The vendor says, the board supports p2p.


http://www.supermicro.com/products/system/4U/7048/SYS-7048A-T.cfm
http://www.supermicro.com/products/motherboard/Xeon/C600/X10DAi.cfm

*SUPERMICRO GPU SuperWorkstation 7048A-T*

Component

Prod. Code

Qty

Chassis

CSE-743TQ-1200B-SQ

1

Mainboard

Super X10DAi

1

Processor

Intel 8C E5-2640V3 2.6G

2

RAM

16 GB DDR4-2133 ECC REG

4

HDD

4TB SATA Ent HDDs 3.5" (Toshiba)

3

SSD

512GB SSD Sansung EVO (MZ-7KE512BW)

1

DVDROM

DVD RW

1

Heatsink

P0050AP4

2

rail kit

CSE-PT26L-B

1

Kb/Mouse

USB 3-button optical scroll mouse / Std.kb

2

Tray

2.5" Tray

1

GPU

Nvidia GTX Titan X (12GB)

2


However, when I run gpuP2PCheck, I am getting the following outputs:

CUDA_VISIBLE_DEVICES="0,1"
CUDA-capable device count: 2
   GPU0 "GeForce GTX TITAN X"
   GPU1 "GeForce GTX TITAN X"

Two way peer access between:
   GPU0 and GPU1: NO

For debugging purpose, I have attached the output of "nvidia-smi -q"
and also the following information may be useful:

gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-4), CentOS 7
GA compatible controller: NVIDIA Corporation GM200 [GeForce GTX TITAN X]
(rev a1)

nvidia-settings gives:
----------------------------
Graphics Processor- GeForce GTX TITAN X
GPU UUID- GPU-7e9e08a8-9b27-828b-7fb1-b5d7e8f6866d
CUDA Cores- 3072
VBIOS Version- 84.00.45.00.03
Total Memory- 12288 MB
Memory Interface- 384-bit
Bus Type- PCI Express x16 Gen3
Bus ID- PCI:2:0:0
PCI Device ID- 0x17c2
PCI Vendor ID- 0x10de
IRQ- 69
PCIe Generation- Gen3
Maximum PCIe Link Width- x16
Maximum PCIe Link Speed- 8.0 GT/s


I would really appreciate any help to resolve these issues.

With my regards,
Biplab.

--
"Simplicity in life allows you to focus on what's important"
On 24 May 2016 at 00:30, <amber-request.ambermd.org> wrote:
> Send AMBER mailing list submissions to
>         amber.ambermd.org
>
> To subscribe or unsubscribe via the World Wide Web, visit
>         http://lists.ambermd.org/mailman/listinfo/amber
> or, via email, send a message with subject or body 'help' to
>         amber-request.ambermd.org
>
> You can reach the person managing the list at
>         amber-owner.ambermd.org
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of AMBER digest..."
>
>
> AMBER Mailing List Digest
>
> Today's Topics:
>
>    1. Re: QM/MM Interface (Nisler, Collin R.)
>    2. optimised corrections to ff03 (Neha Gandhi)
>    3. Re: optimised corrections to ff03 (David Cerutti)
>    4. Re: QM/MM Interface (Nisler, Collin R.)
>    5. nativecontacts (Neha Gandhi)
>    6. Re: obtain or build parameters for non-natural residues
>       (Karl Kirschner)
>    7. Re: nativecontacts (Abelak, Kavin)
>    8. CUDA MPI issues (Biplab Ghosh)
>    9. Re: CUDA MPI issues (Bill Ross)
>   10. Re: MMPBSA.py issue with complex prmtop file (Jason Swails)
>   11. Re: MMPBSA- topology files preparation (Jason Swails)
>   12. Re: CUDA MPI issues (Jason Swails)
>   13. Re: obtain or build parameters for non-natural residues
>       (David Cerutti)
>   14. Re: Problems with GTX titan-X GPU (Ross Walker)
>   15. Re: CUDA MPI issues (Ross Walker)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Sun, 22 May 2016 23:46:39 +0000
> From: "Nisler, Collin R." <nisler.1.buckeyemail.osu.edu>
> Subject: Re: [AMBER] QM/MM Interface
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID:
>         <
> BN1PR01MB3119DC827CE63D82B131BE9984D0.BN1PR01MB311.prod.exchangelabs.com>
>
> Content-Type: text/plain; charset="us-ascii"
>
> When I let Orca finish, I do get a .engrad file that is not empty, and
> orca terminates normally. I tried using your files, and I end up getting
> the same result as before, an empty .engrad file. When I try to run amber
> on the command line, I get the attached error. Could this be relevant at
> all?
>
> ________________________________________
> From: Stanislav Simko <s.simko.uu.nl>
> Sent: Sunday, May 22, 2016 4:31:58 AM
> To: AMBER Mailing List
> Subject: Re: [AMBER] QM/MM Interface
>
> try to let finish the orca calculation - it shouldn't take very long
> (depending on your QM region) to see if you obtain nonzero .engrad
> file. This is critical part since Amber can not continue without this
> file and it seems to me that there's smth wrong with orca calculation
> since the file is created (i.e. orca most probably ran but crashed) but
> it's empty. If that tells you nothing then try to run the attached
> calculation (you need to prepare .prmtop file yourself though, it's too
> big to attach) - i just tested it and it should run fine - the tpl file
> is prepared from your .inp (i also attach you the submit script for SGE
> batch).
> best
> stano
> --
>
> Stanislav Simko.
> PhD applicant
> Inorganic Chemistry and Catalysis
> -------------- next part --------------
> A non-text attachment was scrubbed...
> Name: Capture.PNG
> Type: image/png
> Size: 12344 bytes
> Desc: Capture.PNG
> Url :
> http://lists.ambermd.org/mailman/private/amber/attachments/20160522/73beba3c/attachment-0001.png
>
> ------------------------------
>
> Message: 2
> Date: Mon, 23 May 2016 11:09:04 +1000
> From: Neha Gandhi <n.gandhiau.gmail.com>
> Subject: [AMBER] optimised corrections to ff03
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID:
>         <
> CA+Hq8HzHqNkxPZsH7AiOe2WUPOUDeDdkh4DpOG0wntuG5jODfw.mail.gmail.com>
> Content-Type: text/plain; charset=UTF-8
>
> Hi List,
>
> I was reading papers on ff03* and ff03*. These force fields are available
> for download for gromacs package, however I would like to run it in amber.
> I would l like to know which files are required to be modified in amber to
> include a cosine correction to ? . i.e. *k*? (kcal/mol)
>
> and ?? (deg)
>
>
>
> Many thanks,
> Neha
>
> -
> Regards,
> Dr. Neha S. Gandhi,
> Vice Chancellor's Research Fellow,
> Queensland University of Technology,
> 2 George Street, Brisbane, QLD 4000
> Australia
> LinkedIn
> Research Gate
>
>
> ------------------------------
>
> Message: 3
> Date: Sun, 22 May 2016 22:21:07 -0400
> From: David Cerutti <dscerutti.gmail.com>
> Subject: Re: [AMBER] optimised corrections to ff03
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID:
>         <
> CAEmzWj0N-djoUrFQEZj+WS2dZ_U0iESe7BL0DCSRmu-qMXJusQ.mail.gmail.com>
> Content-Type: text/plain; charset=UTF-8
>
> Hi Neha,
>
> It seems that the folder ${AMBERHOME}/dat/leap/parm/ is the place to
> start.  In it you'll find frcmod.ff03, which contains Yong's dihedral
> parameters.  However, that's only one of at least two files you'd need, and
> the only other one that I see in there is frcmod.ff03ua.  You'll also need
> all_amino03.lib for the charges and atom types, in
> ${AMBERHOME}/dat/leap/lib/, but the most important file is a leaprc, such
> as "leaprc.ff03." That leaprc file is in ${AMBERHOME}/dat/leap/cmd/oldff/,
> which indicates that ff03 is no longer in active development, but there's
> nothing wrong with using it.  If you need to write a frcmod from scratch,
> the format of those things can be found at
>
> http://ambermd.org/formats.html#frcmod
>
> My advice would be to get in touch with the writers of ff03*, and get a
> hold of some structures that you can take single point energies of (use the
> simplest energy minimization file with one step, run sander, and compare
> the results to the benchmark values the ff03* guys can hopefully provide).
> This will be an important step to verifying that you have the correct
> parameters, since it looks like you'll need to piece together the ff mod
> yourself.
>
> After you get that, your goal is to make a leaprc.ff03* file in the form of
> leaprc.ff03.  Critical elements of that file are references to the
> appropriate sub-files (all_amino03.lib, frcmod.ff03*) and a proper listing
> of the relevant atom types (you can check some of the current protein force
> fields, but notice how there is a divergence of atom types--check to make
> sure that you know what everything in ff03* is called).  Once you've got
> frcmod.ff03* and leaprc.ff03* (perhaps call it leaprc.protein.ff03* to
> stick with the new conventions), feel free to contribute--we should keep up
> with the various mods.
>
> Dave
>
>
> On Sun, May 22, 2016 at 9:09 PM, Neha Gandhi <n.gandhiau.gmail.com> wrote:
>
> > Hi List,
> >
> > I was reading papers on ff03* and ff03*. These force fields are available
> > for download for gromacs package, however I would like to run it in
> amber.
> > I would l like to know which files are required to be modified in amber
> to
> > include a cosine correction to ? . i.e. *k*? (kcal/mol)
> >
> > and ?? (deg)
> >
> >
> >
> > Many thanks,
> > Neha
> >
> > -
> > Regards,
> > Dr. Neha S. Gandhi,
> > Vice Chancellor's Research Fellow,
> > Queensland University of Technology,
> > 2 George Street, Brisbane, QLD 4000
> > Australia
> > LinkedIn
> > Research Gate
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
>
>
> ------------------------------
>
> Message: 4
> Date: Mon, 23 May 2016 06:27:13 +0000
> From: "Nisler, Collin R." <nisler.1.buckeyemail.osu.edu>
> Subject: Re: [AMBER] QM/MM Interface
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID:
>         <
> BN1PR01MB3119A6E4830D1CA292AFFC3984E0.BN1PR01MB311.prod.exchangelabs.com>
>
> Content-Type: text/plain; charset="us-ascii"
>
> Please disregard my last email, I finally, at long last, discovered the
> issue. Your initial suggestion that it might be calling the wrong orca was
> in fact correct, the wrong orca was earlier in my path than the orca I
> wanted. The solution was to add this line (in csh) in my submission script:
>
> setenv PATH $HOME/local/src/orca_3.0.1/orca_3_0_1_linux_x86-64:$PATH
>
> where orca_3_0_1_linux_x86-64 is the directory where orca is located.
> Thanks again so much for your help, I very much appreciate the time you
> invested.
>
> Collin
>
>
>
> ________________________________________
> From: Stanislav Simko <s.simko.uu.nl>
> Sent: Sunday, May 22, 2016 4:31:58 AM
> To: AMBER Mailing List
> Subject: Re: [AMBER] QM/MM Interface
>
> try to let finish the orca calculation - it shouldn't take very long
> (depending on your QM region) to see if you obtain nonzero .engrad
> file. This is critical part since Amber can not continue without this
> file and it seems to me that there's smth wrong with orca calculation
> since the file is created (i.e. orca most probably ran but crashed) but
> it's empty. If that tells you nothing then try to run the attached
> calculation (you need to prepare .prmtop file yourself though, it's too
> big to attach) - i just tested it and it should run fine - the tpl file
> is prepared from your .inp (i also attach you the submit script for SGE
> batch).
> best
> stano
> --
>
> Stanislav Simko.
> PhD applicant
> Inorganic Chemistry and Catalysis
>
>
>
> ------------------------------
>
> Message: 5
> Date: Mon, 23 May 2016 17:25:36 +1000
> From: Neha Gandhi <n.gandhiau.gmail.com>
> Subject: [AMBER] nativecontacts
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID:
>         <CA+Hq8Hw58W2byokbcpzNAG=
> CcgdAbw2NXLEEDfwx6hLR6t1t_g.mail.gmail.com>
> Content-Type: text/plain; charset=UTF-8
>
> I have 60000 frames in total and I am using following command to calculate
> contacts (AMBER14 and ambertools15). WIth Ambertools16, it prints fraction
> as 0 with the same command.
>
> nativecontacts :LIG !:POP writecontacts contacts.out resout resout.out
> distance 10 includesolvent byresidue map mapout resmap.gnu series seriesout
> sout.out reference
>
> For some residues I can contact fraction > 100. Is it really possible.e.g.
> output below
>
> #Res1       #Res2  TotalFrac   Contacts
> 442           206    582.479        813
>
> Am I doing this correctly?
>
> --
> Regards,
> Dr. Neha S. Gandhi,
> Vice Chancellor's Research Fellow,
> Queensland University of Technology,
> 2 George Street, Brisbane, QLD 4000
> Australia
> LinkedIn
> Research Gate
>
>
> ------------------------------
>
> Message: 6
> Date: Mon, 23 May 2016 09:31:24 +0200
> From: Karl Kirschner <k.n.kirschner.gmail.com>
> Subject: Re: [AMBER] obtain or build parameters for non-natural
>         residues
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID:
>         <CAF=
> D-bzeZQ1Kr4ek2tFBeEgSt5V7y_cLBWisVyPP8s3WzJLOrg.mail.gmail.com>
> Content-Type: text/plain; charset=UTF-8
>
> Hello Aline,
>
>   The model development of nonstandard residues is a difficult thing to do.
> The main issues that one has to consider include a) determining partial
> atomic charges that are balanced with your desired force field, b) are
> there atom types that already exist in desired force field that can be used
> to describe your molecule, or will you need to create new atom types c)
> choosing the Lennard-Jones parameters for new atom types, d) parameterizing
> any missing bond, angle, and torsion parameters using target experimental
> and/or quantum data. Amber's antechamber program is a good place to start,
> which can provide AM1-BCC charges easily and approximate missing
> parameters. However, to truly optimize bonded parameter one needs to use
> target experimental and/or QM data, which can be very time consuming since
> you might need to generate QM potential energy curves of various internal
> coordinates. The R.E.D. server is a good place for determining more robust
> charges that are balanced with a given force field. The other thing to
> consider is if these nonstandard residues will be used as part of the
> primary sequence of a protein, or if they will be used a nonbonded ligands.
> In the former case, one needs to make sure the charges and parameters are
> balanced with the protein force field. In the later case, one has the
> option of using the GAFF force field for modeling the ligand component.
>
> Bests,
> Karl
>
> On Fri, May 20, 2016 at 9:45 AM, ALINE THOMAS <
> aline.thomas.univ-grenoble-alpes.fr> wrote:
>
> > Dear Amber users and developers
> >
> > I am looking for .lib or .frcmod files for amber for two non-natural
> > residues
> > as gamma-carboxylic glutamic acid (GLA) and beta-hydroxy-aspartate
> > residues (BHA),
> > could not find anything on the archive, or from any website (PTM, NCAA,
> > Bryce lab...)
> >
> > would you have any helping advice for obtaining them
> > or for building them de novo ? never done that,
> >
> > Thank you very much for your help
> >
> > Aline
> >
> >
> > --
> > Aline Thomas, PhD
> >
> > D?partement de Pharmacochimie Mol?culaire
> > Universit? Joseph Fourier BP53
> > 470, rue de la chimie
> > 38041 Grenoble cedex 9
> > T?l : +33 (0) 4.76.63.53.26
> > Fax : +33 (0) 4.76.63.52.98
> > http://dpm.ujf-grenoble.fr/
> >
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
>
>
>
> --
> Karl. N. Kirschner, Ph.D.
> Research Associate
> Bonn-Rhein-Sieg University of Applied Sciences
> Grantham-Allee 20, 54757 Sankt Augustin, Germany
>
>
> ------------------------------
>
> Message: 7
> Date: Mon, 23 May 2016 09:40:02 +0000
> From: "Abelak, Kavin" <kabelak.rvc.ac.uk>
> Subject: Re: [AMBER] nativecontacts
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID: <0FEECC9A-3BCC-45F4-9C8A-B3619501E733.rvc.ac.uk>
> Content-Type: text/plain; charset="us-ascii"
>
> Hi Neha,
>
> As I understand it, if there is more than one native contact between
> residue pairs (i.e., contacts between more than one atom pair) you will
> likely get fractions > 1. See p583 of the AmberTools15 manual.
>
> HTH,
> Kavin
>
>
>
>
>
>
> On 23 May 2016, at 08:25, Neha Gandhi <n.gandhiau.gmail.com<mailto:
> n.gandhiau.gmail.com>> wrote:
>
> I have 60000 frames in total and I am using following command to calculate
> contacts (AMBER14 and ambertools15). WIth Ambertools16, it prints fraction
> as 0 with the same command.
>
> nativecontacts :LIG !:POP writecontacts contacts.out resout resout.out
> distance 10 includesolvent byresidue map mapout resmap.gnu series seriesout
> sout.out reference
>
> For some residues I can contact fraction > 100. Is it really possible.e.g.
> output below
>
> #Res1       #Res2  TotalFrac   Contacts
> 442           206    582.479        813
>
> Am I doing this correctly?
>
> --
> Regards,
> Dr. Neha S. Gandhi,
> Vice Chancellor's Research Fellow,
> Queensland University of Technology,
> 2 George Street, Brisbane, QLD 4000
> Australia
> LinkedIn
> Research Gate
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org<mailto:AMBER.ambermd.org>
> http://lists.ambermd.org/mailman/listinfo/amber
>
>
> [RVC Logo - link to RVC Website]<http://www.rvc.ac.uk>    [Twitter icon -
> link to RVC (Official) Twitter] <http://twitter.com/RoyalVetCollege>
>  [Facebook icon - link to RVC (Official) Facebook] <
> http://www.facebook.com/theRVC>     [YouTube icon - link to RVC YouTube] <
> http://www.youtube.com/user/RoyalVetsLondon?feature=mhee>     [Pinterest
> icon - link to RVC Pinterest] <http://pinterest.com/royalvetcollege/>
>  [Instagram icon - link to RVC Instagram] <
> http://instagram.com/royalvetcollege>
>
> This message, together with any attachments, is intended for the stated
> addressee(s) only and may contain privileged or confidential information.
> Any views or opinions presented are solely those of the author and do not
> necessarily represent those of the Royal Veterinary College (RVC). If you
> are not the intended recipient, please notify the sender and be advised
> that you have received this message in error and that any use,
> dissemination, forwarding, printing, or copying is strictly prohibited.
> Unless stated expressly in this email, this email does not create, form
> part of, or vary any contractual or unilateral obligation. Email
> communication cannot be guaranteed to be secure or error free as
> information could be intercepted, corrupted, amended, lost, destroyed,
> incomplete or contain viruses. Therefore, we do not accept liability for
> any such matters or their consequences. Communication with us by email will
> be taken as acceptance of the risks inherent in doing so.
>
>
> ------------------------------
>
> Message: 8
> Date: Mon, 23 May 2016 16:15:08 +0530
> From: Biplab Ghosh <ghosh.biplab.gmail.com>
> Subject: [AMBER] CUDA MPI issues
> To: amber.ambermd.org
> Message-ID:
>         <CAGsFkLq66CMe8pvKoa45a_o=
> sPSP-jrUF0UO22V_RqK2kBK7hw.mail.gmail.com>
> Content-Type: text/plain; charset=UTF-8
>
> Dear Amber Experts,
>
> I am trying to run amber14 using parallel GPUs.
> I have 2 "GeForce GTX TITAN X" cards installed
> in my workstation and having cuda-7.5 libs.
> Individual GPUs are performing but when I run
> pmemd.cuda.MPI, it gives me the following error:
>
> application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
>
> I then referred to the Amber website to check why GPU communication
> is failing. I downloaded  "check_p2p.tar.bz2" program from the amber site
> and getting the following output upon running.
>
> [biplab.proline check_p2p]$ ./gpuP2PCheck
> CUDA_VISIBLE_DEVICES is unset.
> CUDA-capable device count: 2
>    GPU0 "GeForce GTX TITAN X"
>    GPU1 "GeForce GTX TITAN X"
>
> Two way peer access between:
>    GPU0 and GPU1: NO
>
>
> Can anyone help me on how to configure my system, so that both
> GPU can work in parallel.
>
> Many thanks and regards
>
> Biplab Ghosh
> Bhabha Atomic Research Center, Mumbai
> India.
>
> --
> "Simplicity in life allows you to focus on what's important"
>
>
> ------------------------------
>
> Message: 9
> Date: Mon, 23 May 2016 04:01:05 -0700
> From: Bill Ross <ross.cgl.ucsf.edu>
> Subject: Re: [AMBER] CUDA MPI issues
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID: <911ea656-fa9a-d257-fd17-bb8aba84e37a.cgl.ucsf.edu>
> Content-Type: text/plain; charset=windows-1252; format=flowed
>
>  From "CUDA_VISIBLE_DEVICES is unset" it appears you haven't set an
> environment variable that may be needed.
>
> E.g.
>
>    http://www.acceleware.com/blog/cudavisibledevices-masking-gpus
>
> Bill
>
> On 5/23/16 3:45 AM, Biplab Ghosh wrote:
> > Dear Amber Experts,
> >
> > I am trying to run amber14 using parallel GPUs.
> > I have 2 "GeForce GTX TITAN X" cards installed
> > in my workstation and having cuda-7.5 libs.
> > Individual GPUs are performing but when I run
> > pmemd.cuda.MPI, it gives me the following error:
> >
> > application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
> >
> > I then referred to the Amber website to check why GPU communication
> > is failing. I downloaded  "check_p2p.tar.bz2" program from the amber site
> > and getting the following output upon running.
> >
> > [biplab.proline check_p2p]$ ./gpuP2PCheck
> > CUDA_VISIBLE_DEVICES is unset.
> > CUDA-capable device count: 2
> >     GPU0 "GeForce GTX TITAN X"
> >     GPU1 "GeForce GTX TITAN X"
> >
> > Two way peer access between:
> >     GPU0 and GPU1: NO
> >
> >
> > Can anyone help me on how to configure my system, so that both
> > GPU can work in parallel.
> >
> > Many thanks and regards
> >
> > Biplab Ghosh
> > Bhabha Atomic Research Center, Mumbai
> > India.
> >
> > --
> > "Simplicity in life allows you to focus on what's important"
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
>
>
>
>
> ------------------------------
>
> Message: 10
> Date: Mon, 23 May 2016 08:53:50 -0400
> From: Jason Swails <jason.swails.gmail.com>
> Subject: Re: [AMBER] MMPBSA.py issue with complex prmtop file
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID:
>         <
> CAEk9e3pNpKOxawZm-WWhdQ0waF6dM-8bTg16FFY0mxQzyAzutg.mail.gmail.com>
> Content-Type: text/plain; charset=UTF-8
>
> On Fri, May 20, 2016 at 6:23 PM, Kalenkiewicz, Andrew (NIH/NICHD) [F] <
> andrew.kalenkiewicz.nih.gov> wrote:
>
> > Dear Amber Users,
> >
> > I'm trying to run MMPBSA.py with residue decomposition, but the job keeps
> > failing with the following output:
> >
> > Loading and checking parameter files for compatibility...
> > sander found! Using /usr/local/apps/amber/amber14/bin/sander
> > cpptraj found! Using /usr/local/apps/amber/amber14/bin/cpptraj
> > Preparing trajectories for simulation...
> > 1000 frames were processed by cpptraj for use in calculation.
> >
> > Running calculations on normal system...
> >
> > Beginning GB calculations with /usr/local/apps/amber/amber14/bin/sander
> >   calculating complex contribution...
> >   calculating receptor contribution...
> >   File "/usr/local/apps/amber/amber14/bin/MMPBSA.py.MPI", line 104, in
> > <module>
> >     app.run_mmpbsa()
> >   File
> >
> "/usr/local/apps/amber/amber14/lib/python2.6/site-packages/MMPBSA_mods/main.py",
> > line 218, in run_mmpbsa
> >     self.calc_list.run(rank, self.stdout)
> >   File
> >
> "/usr/local/apps/amber/amber14/lib/python2.6/site-packages/MMPBSA_mods/calculation.py",
> > line 82, in run
> >     calc.run(rank, stdout=stdout, stderr=stderr)
> >   File
> >
> "/usr/local/apps/amber/amber14/lib/python2.6/site-packages/MMPBSA_mods/calculation.py",
> > line 431, in run
> >     self.prmtop) + '\n\t'.join(error_list) + '\n')
> > CalcError: /usr/local/apps/amber/amber14/bin/sander failed with prmtop
> > complex.prmtop!
> >
> >
> > Error occured on rank 127.
> > Exiting. All files have been retained.
> >
> --------------------------------------------------------------------------
> > MPI_ABORT was invoked on rank 127 in communicator MPI_COMM_WORLD
> > with errorcode 1.
> >
> > NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes.
> > You may or may not see output from other processes, depending on
> > exactly when Open MPI kills them.
> >
> --------------------------------------------------------------------------
> >   File "/usr/local/apps/amber/amber14/bin/MMPBSA.py.MPI", line 104, in
> > <module>
> >     app.run_mmpbsa()
> >   File
> >
> "/usr/local/apps/amber/amber14/lib/python2.6/site-packages/MMPBSA_mods/main.py",
> > line 218, in run_mmpbsa
> >     self.calc_list.run(rank, self.stdout)
> >   File
> >
> "/usr/local/apps/amber/amber14/lib/python2.6/site-packages/MMPBSA_mods/calculation.py",
> > line 82, in run
> >     calc.run(rank, stdout=stdout, stderr=stderr)
> >   File
> >
> "/usr/local/apps/amber/amber14/lib/python2.6/site-packages/MMPBSA_mods/calculation.py",
> > line 431, in run
> >     self.prmtop) + '\n\t'.join(error_list) + '\n')
> > CalcError: /usr/local/apps/amber/amber14/bin/sander failed with prmtop
> > complex.prmtop!
> >
> >
> > Error occured on rank 125.
> > Exiting. All files have been retained.
> > [cn0935:52965] 1 more process has sent help message help-mpi-api.txt /
> > mpi-abort
> > [cn0935:52965] Set MCA parameter "orte_base_help_aggregate" to 0 to see
> > all help / error messages
> >
> > My complex, receptor, and ligand files were generated with ante-MMPBSA.py
> > and look fine as far as I can tell. I double checked the strip_mask and
> > other likely problems. What other reasons could there be for this error
> > message?
> >
>
> ?Check the _MMPBSA_complex_gb.mdout.127 file for any error messages that
> have been printed (this is where the real error message is contained --
> MMPBSA.py doesn't know what went wrong, it only knows that something did go
> wrong).
>
> That said, 128 a *lot* of threads for an MMPBSA.py job.  I would suggest
> switching to 1 CPU with only a couple frames to make debugging easier.
>
> HTH,
> Jason
>
> --
> Jason M. Swails
>
>
> ------------------------------
>
> Message: 11
> Date: Mon, 23 May 2016 08:57:44 -0400
> From: Jason Swails <jason.swails.gmail.com>
> Subject: Re: [AMBER] MMPBSA- topology files preparation
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID:
>         <
> CAEk9e3q+0KT7x+89Tboj8mCkhwsCLv3WdCbVuMtfbA96aC7yDQ.mail.gmail.com>
> Content-Type: text/plain; charset=UTF-8
>
> On Sat, May 21, 2016 at 10:44 PM, Kat G <katwin86.gmail.com> wrote:
>
> > Hi all,
> >
> > I am interested in MMPBSA estimation for the interaction of protein and
> > ligand. My system is a homo-dimer, and each monomer interacts with the
> same
> > ligand at Mg2+ site. By experiments, we know that only in dimer form and
> > with the presence of Mg2+ cofactor, ligands can be bound to the dimer
> > structure. Then my question is how should I prepare topology files for
> > "receptor" and "ligand" in this case for meaningful calculation.
> >
> > 1. "Ligand" is ligand 1, and "receptor" is a dimer, 2 Mg2+ ions,  and the
> > other ligand. Then "complex" is the whole system, except WAT and
> > counter-ions.
> >
> > 2. "Ligand" is both similar ligands, and "receptor" is a dimer with Mg2+
> > ions.
> >
> > 3. "Ligand" is ligand 1, and "receptor" is just one corresponding monomer
> > and Mg2+ ion. Then, complex is just only 1 monomer interacting with 1
> > ligand.
> >
> > Could it be one of these above cases? I think of the first case but still
> > not be sure. Could you please give me any suggestion.
> >
>
> ?They are all valid choices depending on what you are trying to learn.
>
> HTH,
> Jason
>
> --
> Jason M. Swails
>
>
> ------------------------------
>
> Message: 12
> Date: Mon, 23 May 2016 09:22:20 -0400
> From: Jason Swails <jason.swails.gmail.com>
> Subject: Re: [AMBER] CUDA MPI issues
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID:
>         <
> CAEk9e3reVJejFUdsDdfnLGTVPDeVCOM1mpBGVMdQfWg_j8cB2g.mail.gmail.com>
> Content-Type: text/plain; charset=UTF-8
>
> On Mon, May 23, 2016 at 6:45 AM, Biplab Ghosh <ghosh.biplab.gmail.com>
> wrote:
>
> > Dear Amber Experts,
> >
> > I am trying to run amber14 using parallel GPUs.
> > I have 2 "GeForce GTX TITAN X" cards installed
> > in my workstation and having cuda-7.5 libs.
> > Individual GPUs are performing but when I run
> > pmemd.cuda.MPI, it gives me the following error:
> >
> > application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
> >
> > I then referred to the Amber website to check why GPU communication
> > is failing. I downloaded  "check_p2p.tar.bz2" program from the amber site
> > and getting the following output upon running.
> >
> > [biplab.proline check_p2p]$ ./gpuP2PCheck
> > CUDA_VISIBLE_DEVICES is unset.
> > CUDA-capable device count: 2
> >    GPU0 "GeForce GTX TITAN X"
> >    GPU1 "GeForce GTX TITAN X"
> >
> > Two way peer access between:
> >    GPU0 and GPU1: NO
> >
> >
> > Can anyone help me on how to configure my system, so that both
> > GPU can work in parallel.
> >
>
> ?This does not mean that GPU0 and GPU1 cannot work in parallel, it simply
> means that they cannot communicate via the P2P (peer-to-peer) protocol.  So
> they won't be able to work together *efficiently*.  This support comes from
> the motherboard, and you need to plug the cards into two PCIe slots that
> support this communication.  Some motherboards don't support this
> communication at all, to my knowledge.  If that is the case, there is
> nothing you can do to fix P2P connectivity in your computer.
>
> ?For the error message, you should check your output file for possible
> errors.  And/or run in serial (or on the CPU) if it's still not clear what
> the error is.
>
> HTH,
> Jason\
> --
> Jason M. Swails
>
>
> ------------------------------
>
> Message: 13
> Date: Mon, 23 May 2016 10:12:23 -0400
> From: David Cerutti <dscerutti.gmail.com>
> Subject: Re: [AMBER] obtain or build parameters for non-natural
>         residues
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID:
>         <
> CAEmzWj0qzi+FLFN9Ez14GOb1H9haFaoTwvRy322jXGGVw4QXVg.mail.gmail.com>
> Content-Type: text/plain; charset=UTF-8
>
> Hi Aline,
>
> If you want to fight the good fight and make parameters de-novo, I can walk
> you through a pipeline for this.  It's something I'm (very) actively
> developing, but we're just getting ready to submit the manuscript for the
> second-generation protein force field.  The charges will be relatively
> simple to create, and for the cases you've identified I would imagine that
> only retyping CB and CG atoms, respectively, would be necessary, which
> greatly limits the number of torsion parameters that would need to be
> fitted.  There are three protein force fields currently distributed with
> Amber: ff14SB, ff15ipq, and FB15.  The first and third use the Cornell
> charges first minted in 1995, which are very much like what you'd get if
> you were to use the R.E.D. server that Karl mentioned.  The charges on
> ff15ipq are of a different breed, using a protocol I developed with some
> IBM people in 2010.
>
> Whichever charge set you use, you'll need some new torsions.  I would
> recommend you inherit backbone torsions from whatever force field you
> choose to model your nonstandard amino acids after, and when you do that
> it's also a good idea to inherit the charges on backbone atoms.  I would
> imagine that R.E.D. server lets you peg charges to pre-defined values, but
> if not I know exactly how to do this in my AmberTools program.
>
> So, the workflow I'd see here is that you pick one of ff14SB (standard
> Amber protein model), ff15ipq (hot off the presses, complete rederivation
> of charges and bonded parameters with an updated physics model), or fb15
> (ff14SB charges with rederivation of bonded parameters).  Then, make your
> library files for the amino acids as they will appear in the protein
> (main-chain or terminal groups... just need to know about how they connect
> to other residues), refit charges appropriate to whichever protein force
> field you chose, differentiate the atom types starting at the point along
> the side chain that these amino acids become "nonstandard," and finally
> make some conformations and do the torsion fitting.
>
> Dave
>
>
> On Mon, May 23, 2016 at 3:31 AM, Karl Kirschner <k.n.kirschner.gmail.com>
> wrote:
>
> > Hello Aline,
> >
> >   The model development of nonstandard residues is a difficult thing to
> do.
> > The main issues that one has to consider include a) determining partial
> > atomic charges that are balanced with your desired force field, b) are
> > there atom types that already exist in desired force field that can be
> used
> > to describe your molecule, or will you need to create new atom types c)
> > choosing the Lennard-Jones parameters for new atom types, d)
> parameterizing
> > any missing bond, angle, and torsion parameters using target experimental
> > and/or quantum data. Amber's antechamber program is a good place to
> start,
> > which can provide AM1-BCC charges easily and approximate missing
> > parameters. However, to truly optimize bonded parameter one needs to use
> > target experimental and/or QM data, which can be very time consuming
> since
> > you might need to generate QM potential energy curves of various internal
> > coordinates. The R.E.D. server is a good place for determining more
> robust
> > charges that are balanced with a given force field. The other thing to
> > consider is if these nonstandard residues will be used as part of the
> > primary sequence of a protein, or if they will be used a nonbonded
> ligands.
> > In the former case, one needs to make sure the charges and parameters are
> > balanced with the protein force field. In the later case, one has the
> > option of using the GAFF force field for modeling the ligand component.
> >
> > Bests,
> > Karl
> >
> > On Fri, May 20, 2016 at 9:45 AM, ALINE THOMAS <
> > aline.thomas.univ-grenoble-alpes.fr> wrote:
> >
> > > Dear Amber users and developers
> > >
> > > I am looking for .lib or .frcmod files for amber for two non-natural
> > > residues
> > > as gamma-carboxylic glutamic acid (GLA) and beta-hydroxy-aspartate
> > > residues (BHA),
> > > could not find anything on the archive, or from any website (PTM, NCAA,
> > > Bryce lab...)
> > >
> > > would you have any helping advice for obtaining them
> > > or for building them de novo ? never done that,
> > >
> > > Thank you very much for your help
> > >
> > > Aline
> > >
> > >
> > > --
> > > Aline Thomas, PhD
> > >
> > > D?partement de Pharmacochimie Mol?culaire
> > > Universit? Joseph Fourier BP53
> > > 470, rue de la chimie
> > > 38041 Grenoble cedex 9
> > > T?l : +33 (0) 4.76.63.53.26
> > > Fax : +33 (0) 4.76.63.52.98
> > > http://dpm.ujf-grenoble.fr/
> > >
> > > _______________________________________________
> > > AMBER mailing list
> > > AMBER.ambermd.org
> > > http://lists.ambermd.org/mailman/listinfo/amber
> > >
> >
> >
> >
> > --
> > Karl. N. Kirschner, Ph.D.
> > Research Associate
> > Bonn-Rhein-Sieg University of Applied Sciences
> > Grantham-Allee 20, 54757 Sankt Augustin, Germany
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
>
>
> ------------------------------
>
> Message: 14
> Date: Mon, 23 May 2016 08:35:23 -0700
> From: Ross Walker <ross.rosswalker.co.uk>
> Subject: Re: [AMBER] Problems with GTX titan-X GPU
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID: <5B329370-200D-401A-B1AF-7952C3619EF1.rosswalker.co.uk>
> Content-Type: text/plain; charset=us-ascii
>
> Hi Elvis,
>
> Please try running the following validation suite:
>
> https://dl.dropboxusercontent.com/u/708185/GPU_Validation_Test.tar.gz
>
> untar it and edit run_test.x and set gpu_count to the number of GPUs in
> your system - likely 1 I am guessing. Change test_count to 10 and
> run_large_test to true. Then do:
>
> nohup ./run_test.x >& run_test.log &
>
> It will take around 24 hours to run - you can check if it is still running
> with nvidia-smi. After it is complete please send the contents of the
> GPU_0.log, GPU.large_0.log and run_test.log to the list and we should be
> able to help you debug things and determine if it is a bad GPU or something
> else.
>
> All the best
> Ross
>
> > On May 21, 2016, at 03:19, Elvis Martis <elvis.martis.bcp.edu.in> wrote:
> >
> > Hello,
> > I have a strange problem running my simulations using  Amber14 on GTX
> > Titan-X. Jobs run for several minutes and then die without any error
> > message, hence I am unable to figure out what is the problem. I keep
> > monitoring the GPU temperature while a job is running and it never
> crosses
> > 82C.
> > This GPU card was obtained as gift from NVIDIA as seeding grant.
> > Some information on my system.
> > Amber version- Amber14 and AmberTools15 <all updates applied>
> > CenOS release- centos-release-6-7.el6.centos.12.3.x86_64
> > CUDA version- CUDA-7.5 (Driver version 352.79) <see attachments for
> > complete system log generated using "nvidia-smi -q" nvidia.log and
> > nvidia-smi.txt>
> > CPU- Intel(R) Core(TM) i7-4790 CPU . 3.60GHz <see attached file cpu.log>
> >
> > Please let me know if any more information is needed to help me solve
> this
> > problem.
> > Thanks in advance for your Help.
> > Regards
> >
> > *Elvis Martis*   PhD Student, Bombay College of Pharmacy
> > [image: photo] Website: http://www.elvismartis.in
> > <http://www.elvismartis.in/>
> > Address: Kalina, Santa Cruz [E]
> >
> > <https://in.linkedin.com/in/elvisadrianmartis>
> >
> <nvidia-smi.txt><nvidia.log><cpu.log>_______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
>
>
>
>
> ------------------------------
>
> Message: 15
> Date: Mon, 23 May 2016 08:47:36 -0700
> From: Ross Walker <ross.rosswalker.co.uk>
> Subject: Re: [AMBER] CUDA MPI issues
> To: AMBER Mailing List <amber.ambermd.org>
> Message-ID: <6A3AD813-8F26-426E-874C-99432938353F.rosswalker.co.uk>
> Content-Type: text/plain; charset=us-ascii
>
> Hi Biplab,
>
> How are you actually running the calculation across two GPUs? Are you
> using mpirun? As in
>
> mpirun -np 2 $AMBERHOME/bin/pmemd.cuda.MPI -O -i ...
>
> It's unclear from your description if you are doing this or not. I'd
> suggest reading through the instructions on http://ambermd.org/gpus/ to
> make sure you are runnings things correctly. If you are then we can do some
> further debugging - e.g. take a look in the mdout file etc. - Do the test
> cases work in parallel, does CPU parallel work etc?
>
> With regards to P2P support - this is required if you want to see speedup
> across 2 GPUs. Who manufacturers your system? In the first instance you
> should go back to them and request details on how to reconfigure the
> hardware to support P2P connectivity. If it is a home built system then
> please send the motherboard spec (make and model) and we can try and figure
> out how the GPUs should be installed to support this.
>
> All the best
> Ross
>
> > On May 23, 2016, at 03:45, Biplab Ghosh <ghosh.biplab.gmail.com> wrote:
> >
> > Dear Amber Experts,
> >
> > I am trying to run amber14 using parallel GPUs.
> > I have 2 "GeForce GTX TITAN X" cards installed
> > in my workstation and having cuda-7.5 libs.
> > Individual GPUs are performing but when I run
> > pmemd.cuda.MPI, it gives me the following error:
> >
> > application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
> >
> > I then referred to the Amber website to check why GPU communication
> > is failing. I downloaded  "check_p2p.tar.bz2" program from the amber site
> > and getting the following output upon running.
> >
> > [biplab.proline check_p2p]$ ./gpuP2PCheck
> > CUDA_VISIBLE_DEVICES is unset.
> > CUDA-capable device count: 2
> >   GPU0 "GeForce GTX TITAN X"
> >   GPU1 "GeForce GTX TITAN X"
> >
> > Two way peer access between:
> >   GPU0 and GPU1: NO
> >
> >
> > Can anyone help me on how to configure my system, so that both
> > GPU can work in parallel.
> >
> > Many thanks and regards
> >
> > Biplab Ghosh
> > Bhabha Atomic Research Center, Mumbai
> > India.
> >
> > --
> > "Simplicity in life allows you to focus on what's important"
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
>
>
>
>
> ------------------------------
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
>
> End of AMBER Digest, Vol 1585, Issue 1
> **************************************
>











_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber


Received on Wed May 25 2016 - 23:30:02 PDT
Custom Search