1 |
Hi Christoph, |
2 |
|
3 |
thanks for the detailed answer. I have made a few changes to the ebuild, |
4 |
implementing what you suggested. The only thing missing right now is a |
5 |
better way to deal with the Makefile. I am working on that. In the meantime |
6 |
I have created a pull request. |
7 |
|
8 |
Thanks again, |
9 |
|
10 |
nick |
11 |
|
12 |
|
13 |
|
14 |
On Sun, May 12, 2013 at 5:08 PM, Christoph Junghans <ottxor@g.o>wrote: |
15 |
|
16 |
> Hi Nick, |
17 |
> |
18 |
> thanks for your contribution. |
19 |
> |
20 |
> I think there is no general answer to your question. It depends if it |
21 |
> is a disadvantage of running the mpi-version with 1 MPI process over |
22 |
> running the serial version. I imagine that LAMMPS enable some special |
23 |
> optimization if mpi is disabled as it can assume that everything |
24 |
> happens locally and no mpi message need to be send. On the other hand, |
25 |
> if the build of LAMMPS takes hours, one might prefer building it only |
26 |
> once. Do LAMMPS offer other palatalizations like OpenMP or Threads to |
27 |
> use multiple cores? |
28 |
> |
29 |
> From a brief look at our patch I see the following issues: |
30 |
> - I am not sold on this Makefile patch and patching of the patch |
31 |
> approach, seems a bit fragile. It might be better to use the default |
32 |
> gcc Makefile and patch that. I prefer adding CC,CFLAGS etc, on emake |
33 |
> comandline, you could create a function to avoid writing it on every |
34 |
> emake call. (my_emake() { emake "$@" CC= CFLAGS=; } ) |
35 |
> - sed calls need to secured with "|| die" |
36 |
> - DEPEND should be static, DEPEND="mpi? ( vitrual/mpi )" should do it |
37 |
> - virtual/fortran is missing in DEPEND or use fortran-2.eclass and |
38 |
> FORTRAN_NEEDED="fortran" instead. |
39 |
> - lammps-gzip enables normal gzip support, so the use flags should be |
40 |
> called gzip. |
41 |
> |
42 |
> If you are unsure about some technical details in an ebuild, create a |
43 |
> pull request on github and somebody will give you advice in the |
44 |
> review. |
45 |
> |
46 |
> Otherwise good job, looking forward to more commits from you, |
47 |
> |
48 |
> Christoph |
49 |
> |
50 |
> 2013/5/12 Nicolas Bock <nicolasbock@×××××.com>: |
51 |
> > Hi, |
52 |
> > |
53 |
> > I recently added a lammps ebuild to the science overlay. It so far only |
54 |
> > builds the serial version, and has very limited support for build time |
55 |
> > options. I now added the mpi use flag and changed the ebuild so that it |
56 |
> > builds the mpi version. I have attached a patch against master that does |
57 |
> > this. I would like to get some feedback on those changes. Are they |
58 |
> > acceptable in this form? What should I have done differently? How would |
59 |
> I go |
60 |
> > about writing an ebuild that builds both, the serial _and_ the mpi |
61 |
> version |
62 |
> > in case the mpi use flag is set? |
63 |
> > |
64 |
> > Thanks already, |
65 |
> > |
66 |
> > nick |
67 |
> > |
68 |
> |
69 |
> |
70 |
> |
71 |
> -- |
72 |
> Christoph Junghans |
73 |
> http://dev.gentoo.org/~ottxor/ |
74 |
> |
75 |
> |