[openib-general] Trying to compile mvapich RHEL4U3 for ib.

[Roger Heflin]

Sayantan Sur wrote:
> Hello Roger,
> 
> I'm just CC-ing this to openib-general for the community.
> 
> Thanks for giving us access. I have verified that the
> `ibv_get_device_list' verb is indeed *missing* from the OpenIB install.
> I'm afraid that given this Redhat rpm, it is difficult to get mvapich to
> work (without patching it).
> 
> As Roland and others have indicated, perhaps the best way is for you to
> upgrade to atleast the 1.0 branch. That should be the most stable OpenIB
> release yet.
> 
> https://openib.org/svn/gen2/branches/1.0/src/userspace/
> 
> You should be able to keep the kernel stuff intact and just upgrade the
> user level support (management, libibverbs, libmthca). You may skip
> upgrading management, however it'll be best to upgrade it too, lest you
> face any OpenSM issues.
> 
> Thanks,
> Sayantan.


I now have the machines running RHEL4U3 + kernel.org 2.6.16.5 + the
Openib 1.0 userspace, given that the RPM spec files did work for the
openib tools that made things pretty simple, and have a resonable
set of rpms and tar files to execute the kernel+userspace update.

I have succeeded in getting OpenMPI to compile and execute HPL under
raw IB, and so far I am getting reasonable results and no corruption

Mvapich compiles but appears to not have made the mpirun version for
Infiniband, and yells about that when attempting to start HPL, I have
not yet looked at that in detail to see what the nature of the failure
is.

                               Roger