Serial, MPI, OpenMP, CUDA, MPI/OpenMP, and CUDA/MPI all build with
Intel16/Intel MPI 5/CUDA 8 (haven't yet run tests). However, the
cpptraj target was broken (having to do with the double colon rule)
such that only the first build of cpptraj would work (e.g. you would
get the serial cpptraj but not OpenMP cpptraj if you subsequently
reconfigured and re-built). It is now fixed.
-Dan
On Thu, Mar 23, 2017 at 3:47 PM, Ruxi Qi <ruxiq.uci.edu> wrote:
> Thank you all! That's most likely due to the grep mistake in our
> Makefile. Sorry for my carelessness check.
>
> And Dan, could you also test on your intel platform? My test passed just
> with some ifort/icpc warning. Thanks!
>
> Best,
>
> Ruxi
>
>
> On 3/23/17 10:17, Daniel Roe wrote:
>> Dave's fix appears to have worked for me for serial, openmp, mpi,
>> openmp/mpi, cuda, and cuda/mpi (at least with gnu). Serial tests have
>> just finished and all pass. I'm going to keep testing with other
>> compilers etc.
>>
>> -Dan
>>
>> On Thu, Mar 23, 2017 at 1:11 PM, Ruxi Qi <ruxiq.uci.edu> wrote:
>>> Hi All,
>>>
>>> I am really sorry about the break. I should have pushed to other branch
>>> instead of master. What we were trying to do is to remove fftw stuff
>>> inside pbsa due to the license incompatibility concern. Looks like the
>>> build hanging is a dependency issue. I am trying to fix this; if there
>>> is no good at the end of today, we will revert this push and only keep
>>> license and no-harm changes. Sorry.
>>>
>>> Best,
>>>
>>> Ruxi
>>>
>>>
>>> On 3/23/17 6:20, David Case wrote:
>>>> On Wed, Mar 22, 2017, Hai Nguyen wrote:
>>>>> I tried serial build with my mac (serira 10.12.3), native clang (8.0.0),
>>>>> gfortran
>>>>> I don't see the output of "make install" for pbsa. This process hangs
>>>>> forever (I've been waiting for few minutes and
>>>>> need to interrupt that).
>>>> The recent updates to the Makefile seemed to have removed references to
>>>> PARALLEL_TOKEN (and all other references to parallel builds: was this the
>>>> intent)?
>>>>
>>>> The upshot is that configured_serial hangs at the grep step. I've added
>>>> back in the reference to PARALLEL_TOKEN, so that least all builds won't hang.
>>>>
>>>> I'm not sure why we are attempting such major changes to pbsa at this late
>>>> stage in the release cycle. But it is more than ever *imperative* to run
>>>> tests before making commits to master. Please do not push something that
>>>> you know doesn't work, and ask others to test. And don't commit things
>>>> without testing the serial and parallel builds, as well as cuda.
>>>>
>>>> ...thx...dac
>>>>
>>>>
>>>> _______________________________________________
>>>> AMBER-Developers mailing list
>>>> AMBER-Developers.ambermd.org
>>>> http://lists.ambermd.org/mailman/listinfo/amber-developers
>>>
>>> _______________________________________________
>>> AMBER-Developers mailing list
>>> AMBER-Developers.ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber-developers
>>
>>
>
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
--
-------------------------
Daniel R. Roe
Laboratory of Computational Biology
National Institutes of Health, NHLBI
5635 Fishers Ln, Rm T900
Rockville MD, 20852
https://www.lobos.nih.gov/lcb
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Fri Mar 24 2017 - 09:00:02 PDT
