On Aug 9, 2005, at 8:04 AM, Jeff Squyres wrote:
On Aug 9, 2005, at 7:24 AM, Sridhar Chirravuri wrote:
I have disabled building OpenIB and to do so I have touched
.ompi_ignore. This should not be a problem for MVAPI.
If the Open IB headers / libraries are not located in compiler-known
locations, then you shouldn't need to .ompi_ignore the tree (i.e.,
configure won't find the Open IB headers / libraries, and will
therefore automatically skip those components).
Again, it is our intention that users will neither know about nor have
to touch files in the distribution -- they only need use appropriate
options to "configure" and then "make".
I'm not sure if we have explicit options to disable a component in
configure -- Brian, can you comment here?
But of course - we can cook your breakfast as well ;).
You can explicitly disable a component using the --enable-mca-no-
build=NAME. I know the syntax is a bit wonky, but Autoconf doesn't
make this easy :(. Since you want to disable Open IB, you might do
something like:
./configure [rest of options] --enable-mca-no-build=bml-openib
Does something bad happen if you try to build with the Open IB
component enabled? If so, could you let us know what happens? Galen
and I are trying to make sure we have all the BTL configure scripts
tightened up before we release to the world. If it's just something
about your test environment, no biggie - just want to make sure we
don't have a real problem.
Since you only are interested in the MVAPI port and therefore not at
all interested in the TEG PML (the older generation of communication
interface), you can build with:
./configure [rest of options] --enable-mca-no-build=bml-openib,pml-
teg,pml-uniq
which will cause the older generation PMLs to not be built. I think
this is probably going to make life easiest on you.
Is there any thing that I am missing while building btl mvapi? Also,
is anyone built for mvapi and tested this OMPI stack. Please let me
know.
Galen Shipman and Tim Woodall are doing all the IB work.
I think that most of your problems will go away once you start using
the OB1 PML and therefore actually start using the MVAPI interface.
I know that both Galen and I have been running a slightly modified
copy of the Intel test suite against the MVAPI driver on a fairly
regular basis. There are one or two minor datatype issues that still
need to be worked out, but we should pass all the point to point tests.
Hope this helps,
Brian
--
Brian Barrett
Open MPI developer
http://www.open-mpi.org/
