Re: [gmx-users] BGQ compilation with verlet kernels: #include
file "kernel_impl.h" not found.
Side note: On a Blue Gene/Q machine this particular version of Gromacs
is 2.5x times faster than the regular one. I really hope thw BGQ
accelerated kernels will go into the main branch soon.
Best,
Jernej
On Wed, Sep 18, 2013 at 11:40 PM, <gmx-users-requ...@gromacs.org> wrote:
> Send gmx-users mailing list submissions to
> gmx-users@gromacs.org
>
> To subscribe or unsubscribe via the World Wide Web, visit
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> or, via email, send a message with subject or body 'help' to
> gmx-users-requ...@gromacs.org
>
> You can reach the person managing the list at
> gmx-users-ow...@gromacs.org
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of gmx-users digest..."
>
>
> Today's Topics:
>
> 1. Re: periodic molecule (Valentina)
> 2. Re: Re: periodic molecule (Justin Lemkul)
> 3. Re: periodic molecule (Valentina)
> 4. Re: modify nsteps in an existing tpr file (Mark Abraham)
> 5. Re: periodic molecule (Valentina)
> 6. Re: BGQ compilation with verlet kernels: #include file
> "kernel_impl.h" not found. (Mark Abraham)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Wed, 18 Sep 2013 06:57:11 -0700 (PDT)
> From: Valentina <valentina.erast...@durham.ac.uk>
> Subject: [gmx-users] Re: periodic molecule
> To: gmx-users@gromacs.org
> Message-ID: <1379512631663-5011288.p...@n6.nabble.com>
> Content-Type: text/plain; charset=us-ascii
>
> Bonds are harmonic. I can test it by restraining them? For purpose of testing
> I can even make the topology bond to be 2.06, as in the input file.
>
> My system shouldn't be any different from a mice sheet, for instance. Both
> have Metal - O bonds that continue through the sheet. What do you mean by
> "the original geometry" - obv it shall be slightly changing through the MD.
>
> Shall I be on Gromacs Dev sub-forum to get more insights on how the
> periodic_molecule works? Or may be could I ask person who implemented that
> part? I would really like to get it working in Gromacs, rather than having
> to change the engine.
>
> Thank you,
> V
>
> --
> View this message in context:
> http://gromacs.5086.x6.nabble.com/periodic-molecule-tp5011270p5011288.html
> Sent from the GROMACS Users Forum mailing list archive at Nabble.com.
>
>
> ------------------------------
>
> Message: 2
> Date: Wed, 18 Sep 2013 10:01:39 -0400
> From: Justin Lemkul <jalem...@vt.edu>
> Subject: Re: [gmx-users] Re: periodic molecule
> To: Discussion list for GROMACS users <gmx-users@gromacs.org>
> Message-ID: <5239b243.4050...@vt.edu>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
>
>
> On 9/18/13 9:57 AM, Valentina wrote:
>> Bonds are harmonic. I can test it by restraining them? For purpose of testing
>> I can even make the topology bond to be 2.06, as in the input file.
>>
>
> Constraints would probably fail, given the relatively large change in
> geometry.
> You can try, but I'd suspect it will crash.
>
>> My system shouldn't be any different from a mice sheet, for instance. Both
>> have Metal - O bonds that continue through the sheet. What do you mean by
>> "the original geometry" - obv it shall be slightly changing through the MD.
>>
>
> Original geometry = what you are providing for coordinates.
>
>> Shall I be on Gromacs Dev sub-forum to get more insights on how the
>> periodic_molecule works? Or may be could I ask person who implemented that
>> part? I would really like to get it working in Gromacs, rather than having
>> to change the engine.
>>
>
> If you want to talk about code, the development list is the right place for
> the
> discussion, but I know that periodic_molecules should work, given previous
> (solved) issues on the list and the fact that I made a simple CNT work at one
> point. Have a look through the code and if there are any obvious (or
> non-obvious!) issues, please report back via gmx-developers.
>
> -Justin
>
> --
> ==================================================
>
> Justin A. Lemkul, Ph.D.
> Postdoctoral Fellow
>
> Department of Pharmaceutical Sciences
> School of Pharmacy
> Health Sciences Facility II, Room 601
> University of Maryland, Baltimore
> 20 Penn St.
> Baltimore, MD 21201
>
> jalem...@outerbanks.umaryland.edu | (410) 706-7441
>
> ==================================================
>
>
> ------------------------------
>
> Message: 3
> Date: Wed, 18 Sep 2013 07:10:39 -0700 (PDT)
> From: Valentina <valentina.erast...@durham.ac.uk>
> Subject: [gmx-users] Re: periodic molecule
> To: gmx-users@gromacs.org
> Message-ID: <1379513439478-5011290.p...@n6.nabble.com>
> Content-Type: text/plain; charset=us-ascii
>
> On restraining bonds:
>
> The atom moves across the basic unit, i.e. it is not recognising that it
> should bond via periodic_molecule
>
> Error: The charge group starting at atom 2833 moved than the distance
> allowed by the domain decomposition (1.615483) in direction X
>
> In any case (restraining or not all-bonds) when I am doing grompp, I get a
> warning -
> WARNING 1 [file LDH21/em.mdp]:
> The sum of the two largest charge group radii (1.062153) is larger than
> rlist (1.000000)
>
> I decided to take it as a healthy one, as according to
> http://www.gromacs.org/Documentation/Errors#The_sum_of_the_two_largest_charge_group_radii_(X)_is_larger_than.c2.a0rlist_-_rvdw.2frcoulomb
>
> this is because of
> "Molecules are broken across periodic boundaries, which is not a problem in
> a periodic system. In this case, the sum of the two largest charge groups
> will correspond to a value of twice the box vector along which the molecule
> is broken."
>
>
>
> --
> View this message in context:
> http://gromacs.5086.x6.nabble.com/periodic-molecule-tp5011270p5011290.html
> Sent from the GROMACS Users Forum mailing list archive at Nabble.com.
>
>
> ------------------------------
>
> Message: 4
> Date: Wed, 18 Sep 2013 16:14:11 +0200
> From: Mark Abraham <mark.j.abra...@gmail.com>
> Subject: Re: [gmx-users] modify nsteps in an existing tpr file
> To: Discussion list for GROMACS users <gmx-users@gromacs.org>
> Message-ID:
> <camnumatyfnidgrec3wumqmfkjht64ulxyccd42zvftn1q-f...@mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
>
> That -om mechanism has been broken for about a decade, unfortunately.
>
> You will need to include the file, or post a link a file, not attach
> it, if you want users of this list to see it.
>
> gmxcheck to compare your new and old .tpr files is useful to see what
> you might need in the new .mdp file to reproduce the first one. Note
> that grompp -c yourold.tpr is the best way to get the same starting
> coordinates.
>
> Mark
>
> On Wed, Sep 18, 2013 at 3:48 PM, Guanglei Cui
> <amber.mail.arch...@gmail.com> wrote:
>> gmxdump -om writes out a mdp file based on the tpr, but that is not read by
>> grompp. I tried to change or comment out mdp options that are not
>> recognized by grompp. It is attached here. The simulation soon crashes with
>> LINCS errors after 25 steps, while the original tpr runs properly. I'm not
>> sure what's missing here.
>>
>>
>> On Tue, Sep 17, 2013 at 6:12 PM, Mark Abraham
>> <mark.j.abra...@gmail.com>wrote:
>>
>>> No. Theoretically useful, but not implemented.
>>>
>>> Mark
>>>
>>> On Tue, Sep 17, 2013 at 4:45 PM, Guanglei Cui
>>> <amber.mail.arch...@gmail.com> wrote:
>>> > Thanks. Is it possible to dump the parameters in the tpr file to a mdp
>>> file?
>>> >
>>> >
>>> > On Tue, Sep 17, 2013 at 3:20 PM, Mark Abraham <mark.j.abra...@gmail.com
>>> >wrote:
>>> >
>>> >> mdrun -nsteps in 4.6 overrides the number of steps in the .tpr
>>> >>
>>> >> Mark
>>> >>
>>> >> On Tue, Sep 17, 2013 at 8:55 PM, Guanglei Cui
>>> >> <amber.mail.arch...@gmail.com> wrote:
>>> >> > Dear GMX users,
>>> >> >
>>> >> > I'm new to Gromacs. So apologies if this question is too simple.
>>> >> >
>>> >> > I downloaded top/tpr files from the supplementary material of a
>>> published
>>> >> > paper. The nsteps set in the tpr file is 100ns. I wish to do a small
>>> test
>>> >> > run. Is there any way I can modify that? I've tried to create a mdp
>>> file
>>> >> > that best matches the parameters found through gmxdump, but it gives
>>> me a
>>> >> > lot of LINCS error. I can upload the mdp file and gmxdump file if you
>>> are
>>> >> > kind to help. Thanks in advance.
>>> >> >
>>> >> > Best regards,
>>> >> > --
>>> >> > Guanglei Cui
>>> >> > --
>>> >> > gmx-users mailing list gmx-users@gromacs.org
>>> >> > http://lists.gromacs.org/mailman/listinfo/gmx-users
>>> >> > * Please search the archive at
>>> >> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
>>> >> > * Please don't post (un)subscribe requests to the list. Use the
>>> >> > www interface or send it to gmx-users-requ...@gromacs.org.
>>> >> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>> >> --
>>> >> gmx-users mailing list gmx-users@gromacs.org
>>> >> http://lists.gromacs.org/mailman/listinfo/gmx-users
>>> >> * Please search the archive at
>>> >> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
>>> >> * Please don't post (un)subscribe requests to the list. Use the
>>> >> www interface or send it to gmx-users-requ...@gromacs.org.
>>> >> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>> >>
>>> >
>>> >
>>> >
>>> > --
>>> > Guanglei Cui
>>> > --
>>> > gmx-users mailing list gmx-users@gromacs.org
>>> > http://lists.gromacs.org/mailman/listinfo/gmx-users
>>> > * Please search the archive at
>>> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
>>> > * Please don't post (un)subscribe requests to the list. Use the
>>> > www interface or send it to gmx-users-requ...@gromacs.org.
>>> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>> --
>>> gmx-users mailing list gmx-users@gromacs.org
>>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>>> * Please search the archive at
>>> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
>>> * Please don't post (un)subscribe requests to the list. Use the
>>> www interface or send it to gmx-users-requ...@gromacs.org.
>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>
>>
>>
>>
>> --
>> Guanglei Cui
>>
>> --
>> gmx-users mailing list gmx-users@gromacs.org
>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>> * Please search the archive at
>> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
>> * Please don't post (un)subscribe requests to the list. Use the
>> www interface or send it to gmx-users-requ...@gromacs.org.
>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
>
> ------------------------------
>
> Message: 5
> Date: Wed, 18 Sep 2013 07:47:50 -0700 (PDT)
> From: Valentina <valentina.erast...@durham.ac.uk>
> Subject: [gmx-users] Re: periodic molecule
> To: gmx-users@gromacs.org
> Message-ID: <1379515670323-5011291.p...@n6.nabble.com>
> Content-Type: text/plain; charset=us-ascii
>
> Update
>
> https://docs.google.com/file/d/0B6Qm4snFANimcmxYVnBOSkNCd2s/edit?usp=sharing
>
> Seems I had to declare the atoms in .rtp with + and - when they are across
> periodic boundary.
> The new movie is showing that atoms do move, but not across the box. Yes, I
> would have to play a little ore with parameters for the system (ongoing
> work), but the major problem with PBC is sorted.
>
> I am giving below my new (working) molecule.rtp file, that is different from
> the earlier one - may be once this it may be helpful for someone;)
>
> Thanks!
>
>
> [ bondedtypes ]
> ; bonds angles dihedrals impropers
> 1 1 0 0
>
> [ B21 ] ;# ldh21
> [ atoms ]
> ; atomname atomtype charge chargegroup
> o1 oc18 -0.875 3
> h1 hoy 0.350 3
> o2 oc17 -0.875 3
> h2 hoy 0.350 3
> o3 oc20 -0.875 3
> h3 hoy 0.350 3
> o4 oc21 -0.875 1
> h4 hoy 0.350 1
> o5 oc16 -0.875 1
> h5 hoy 0.350 1
> o6 oc19 -0.875 1
> h6 hoy 0.350 1
> mg1 mg 1.40 0
> al1 ac3 1.35 1
> mg2 mg 1.40 0
> o7 oc16 -0.875 2
> h7 hoy 0.350 2
> o8 oc19 -0.875 2
> h8 hoy 0.350 2
> o9 oc18 -0.875 1
> h9 hoy 0.350 1
> o10 oc17 -0.875 1
> h10 hoy 0.350 1
> o11 oc20 -0.875 1
> h11 hoy 0.350 1
> o12 oc21 -0.875 2
> h12 hoy 0.350 2
> mg3 mg 1.40 0
> mg4 mg 1.40 0
> al2 ac3 1.35 2
> o13 oc20 -0.875 2
> h13 hoy 0.350 2
> o14 oc21 -0.875 3
> h14 hoy 0.350 3
> o15 oc16 -0.875 3
> h15 hoy 0.350 3
> o16 oc19 -0.875 3
> h16 hoy 0.350 3
> o17 oc18 -0.875 2
> h17 hoy 0.350 2
> o18 oc17 -0.875 2
> h18 hoy 0.350 2
> al3 ac3 1.35 3
> mg5 mg 1.40 0
> mg6 mg 1.40 0
> [ bonds ]
> ; atom1 atom2 parametersindex
> o1 h1 b_oc-h
> o2 h2 b_oc-h
> o3 h3 b_oc-h
> o4 h4 b_oc-h
> o5 h5 b_oc-h
> o6 h6 b_oc-h
> o7 h7 b_oc-h
> o8 h8 b_oc-h
> o9 h9 b_oc-h
> o10 h10 b_oc-h
> o11 h11 b_oc-h
> o12 h12 b_oc-h
> o13 h13 b_oc-h
> o14 h14 b_oc-h
> o15 h15 b_oc-h
> o16 h16 b_oc-h
> o17 h17 b_oc-h
> o18 h18 b_oc-h
> al3 +o1 b_ac-oc
> al3 +o2 b_ac-oc
> al3 +o3 b_ac-oc
> o4 al1 b_ac-oc
> o5 al1 b_ac-oc
> o6 al1 b_ac-oc
> al2 -o7 b_ac-oc
> al2 -o8 b_ac-oc
> o9 al1 b_ac-oc
> o10 al1 b_ac-oc
> o11 al1 b_ac-oc
> o12 al2 b_ac-oc
> al2 -o13 b_ac-oc
> o14 al3 b_ac-oc
> o15 al3 b_ac-oc
> o16 al3 b_ac-oc
> o17 al2 b_ac-oc
> o18 al2 b_ac-oc
> [ angles ]
> ; ai aj ak gromos type
> al3 o14 h14 a_ac-oc-h
> al3 o15 h15 a_ac-oc-h
> al3 +o1 +h1 a_ac-oc-h
> al3 o16 h16 a_ac-oc-h
> al3 +o2 +h2 a_ac-oc-h
> al3 +o3 +h3 a_ac-oc-h
> al1 o4 h4 a_ac-oc-h
> al1 o5 h5 a_ac-oc-h
> al1 o9 h9 a_ac-oc-h
> al1 o6 h6 a_ac-oc-h
> al1 o10 h10 a_ac-oc-h
> al1 o11 h11 a_ac-oc-h
> al2 o12 h12 a_ac-oc-h
> al2 -o7 -h7 a_ac-oc-h
> al2 o17 h17 a_ac-oc-h
> al2 -o8 -h8 a_ac-oc-h
> al2 o18 h18 a_ac-oc-h
> al2 -o13 -h13 a_ac-oc-h
> o5 al1 o10 a_oc-ac-oc_1
> -o7 al2 o18 a_oc-ac-oc_1
> o15 al3 +o2 a_oc-ac-oc_1
> o9 al1 o6 a_oc-ac-oc_1
> o17 al2 -o8 a_oc-ac-oc_1
> +o1 al3 o16 a_oc-ac-oc_1
> o11 al1 o4 a_oc-ac-oc_1
> -o13 al2 o12 a_oc-ac-oc_1
> +o3 al3 o14 a_oc-ac-oc_1
> o5 al1 o9 a_oc-ac-oc_2
> -o7 al2 o17 a_oc-ac-oc_2
> o15 al3 +o1 a_oc-ac-oc_2
> o5 al1 o4 a_oc-ac-oc_2
> -o7 al2 o12 a_oc-ac-oc_2
> o15 al3 o14 a_oc-ac-oc_2
> o10 al1 o6 a_oc-ac-oc_2
> o18 al2 -o8 a_oc-ac-oc_2
> +o2 al3 o16 a_oc-ac-oc_2
> o10 al1 o11 a_oc-ac-oc_2
> o18 al2 -o13 a_oc-ac-oc_2
> +o2 al3 +o3 a_oc-ac-oc_2
> o9 al1 o4 a_oc-ac-oc_2
> o17 al2 o12 a_oc-ac-oc_2
> +o1 al3 o14 a_oc-ac-oc_2
> o6 al1 o11 a_oc-ac-oc_2
> -o8 al2 -o13 a_oc-ac-oc_2
> o16 al3 +o3 a_oc-ac-oc_2
> o5 al1 o6 a_oc-ac-oc_3
> -o7 al2 -o8 a_oc-ac-oc_3
> o15 al3 o16 a_oc-ac-oc_3
> o5 al1 o11 a_oc-ac-oc_3
> -o7 al2 -o13 a_oc-ac-oc_3
> o15 al3 +o3 a_oc-ac-oc_3
> o10 al1 o9 a_oc-ac-oc_3
> o18 al2 o17 a_oc-ac-oc_3
> +o2 al3 +o1 a_oc-ac-oc_3
> o10 al1 o4 a_oc-ac-oc_3
> o18 al2 o12 a_oc-ac-oc_3
> +o2 al3 o14 a_oc-ac-oc_3
> o9 al1 o11 a_oc-ac-oc_3
> o17 al2 -o13 a_oc-ac-oc_3
> +o1 al3 +o3 a_oc-ac-oc_3
> o6 al1 o4 a_oc-ac-oc_3
> -o8 al2 o12 a_oc-ac-oc_3
> o16 al3 o14 a_oc-ac-oc_3
>
> --
> View this message in context:
> http://gromacs.5086.x6.nabble.com/periodic-molecule-tp5011270p5011291.html
> Sent from the GROMACS Users Forum mailing list archive at Nabble.com.
>
>
> ------------------------------
>
> Message: 6
> Date: Wed, 18 Sep 2013 17:39:50 +0200
> From: Mark Abraham <mark.j.abra...@gmail.com>
> Subject: Re: [gmx-users] BGQ compilation with verlet kernels: #include
> file "kernel_impl.h" not found.
> To: Discussion list for GROMACS users <gmx-users@gromacs.org>
> Message-ID:
> <CAMNuMAQDn1FWEiJJds6v7AZQZ1_x4fLy-suK=4dbmw5ovt4...@mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
>
> Thanks for the follow up.
>
> The take-home lesson is that building for BlueGene/Q is unlike
> building for the usual homogenous x86 cluster. You still need an MPI
> and non-MPI build, but the latter should be targeted at the front end
> (Linux on PowerPC, usually), unless/until GROMACS tools acquire MPI
> functionality useful on a BlueGene/Q scale.
>
> Mark
>
> On Wed, Sep 18, 2013 at 2:07 AM, Christopher Neale
> <chris.ne...@mail.utoronto.ca> wrote:
>> Indeed, it works just fine when I compile with mpi. I never thought to check
>> that. My usual procedure is
>> to compile the whole package without mpi and then to compile mdrun with mpi.
>> Thanks for the help Mark.
>>
>> Here is the compilation script that worked for me.
>>
>> module purge
>> module load vacpp/12.1 xlf/14.1 mpich2/xl
>> module load cmake/2.8.8
>> module load fftw/3.3.2
>>
>> export FFTW_LOCATION=/scinet/bgq/Libraries/fftw-3.3.2
>>
>> cmake ../source/ \
>> -DCMAKE_TOOLCHAIN_FILE=BlueGeneQ-static-XL-C \
>> -DCMAKE_PREFIX_PATH=$FFTW_LOCATION \
>> -DCMAKE_INSTALL_PREFIX=$(pwd) \
>> -DGMX_X11=OFF \
>> -DGMX_MPI=ON \
>> -DGMX_PREFER_STATIC_LIBS=ON
>>
>> make -j 16
>> make install
>>
>> --
>> gmx-users mailing list gmx-users@gromacs.org
>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>> * Please search the archive at
>> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
>> * Please don't post (un)subscribe requests to the list. Use the
>> www interface or send it to gmx-users-requ...@gromacs.org.
>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
>
> ------------------------------
>
> --
> gmx-users mailing list
> gmx-users@gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
>
> End of gmx-users Digest, Vol 113, Issue 81
> ******************************************
--
Windows: Re-Boot, Linux: Be-Root.
--
gmx-users mailing list gmx-users@gromacs.org
http://lists.gromacs.org/mailman/listinfo/gmx-users
* Please search the archive at
http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
* Please don't post (un)subscribe requests to the list. Use the
www interface or send it to gmx-users-requ...@gromacs.org.
* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
