1 |
On Fri, 2007-12-14 at 12:13 +0000, Sébastien Fabbro wrote: |
2 |
> -----BEGIN PGP SIGNED MESSAGE----- |
3 |
> Hash: SHA1 |
4 |
> |
5 |
> On 14/12/07 10:24, Matthias Langer wrote: |
6 |
> >> On 02:10 Thu 13 Dec , Justin Bronder (jsbronder) wrote: |
7 |
> >>> 1.1 sys-cluster/openmpi/openmpi-1.2.4.ebuild |
8 |
> >>> |
9 |
> >>> file : http://sources.gentoo.org/viewcvs.py/gentoo-x86/sys-cluster/openmpi/openmpi-1.2.4.ebuild?rev=1.1&view=markup |
10 |
> >>> plain: http://sources.gentoo.org/viewcvs.py/gentoo-x86/sys-cluster/openmpi/openmpi-1.2.4.ebuild?rev=1.1&content-type=text/plain |
11 |
> > |
12 |
> > Why have you removed the "ifc" USE flag (as well as a few others), that |
13 |
> > was present in the ebuilds that can still be found at bug 166787? |
14 |
> |
15 |
> For icc and f90-typesafe, I advised him to do so on #gentoo-science. |
16 |
|
17 |
Well, too bad that gfortran generates code that is often not even half |
18 |
as fast as the code generated by ifc. Thus, there is a *very valid* |
19 |
reason to prefer icc over gfortran when it comes to performance critical |
20 |
software, which programs that use MPI almost always are. However, the |
21 |
fortran eclass prefers gfortran over ifc even if both are installed |
22 |
(which is fine by me as long as it can be overridden) and the |
23 |
fortran-compiler-wrapper "mpif90/opal_wrapper" (which is a binary for |
24 |
some reason) wrapps gfortran then. |
25 |
|
26 |
For sure, something like |
27 |
F77="ifort" FC="ifort" FFLAGS="-O3 -xO" emerge -av openmpi |
28 |
still does the trick, but if i want that "f90-typesave" stuff back, this |
29 |
line would have to be certainly a bit longer... |
30 |
|
31 |
Maybe someone can explain to me what positive side effects the removal |
32 |
of the ifc USE flag has - and why this flag is generally discouraged. |
33 |
|
34 |
> |
35 |
> ifc/icc use flags should be avoided (see bug #97929). |
36 |
> Disabling f90-typesafe did not make much sense as a use flag once you |
37 |
> have the fortran one enabled, but why --with-mpi-f90-size=medium" |
38 |
> and "--with-f90-max-array-dim=4" disappeared I don't know. |
39 |
|
40 |
The reason it disappeared is that it makes gfortran horribly slow when |
41 |
compiling against mpi. This is not the case with ifc, and therefore the |
42 |
old ebuild in bugzilla emitted a bold warning when emerging with |
43 |
USE="-ifc f90-typesafe" but kept quiet if USE="ifc f90-typesave". Thus |
44 |
it *did make sense* to control it with a USE flag, at least with the |
45 |
"ifc" USE flag being around also. |
46 |
|
47 |
> For the gm and gx flags, I don't know, and there is still a nocxx one. |
48 |
|
49 |
I'm not qualified to talk about these flags as I've never used them. |
50 |
|
51 |
Matthias |