[petsc-users] Can I compile PETSC --with-mpi=0 when I use hypre ???

Satish Balay balay at mcs.anl.gov
Thu Apr 26 21:09:31 CDT 2012


getting this working properly might require work - but as a hack - you
can edit config/PETSc/packages/hypre.py and change:

    self.deps       = [self.mpi,self.blasLapack]  
to
    self.deps       = [self.blasLapack]  

> Somehow my mac does not like mpi, so I would like to have no-mpi version for testing in my mac.

Don't see why mpi would be a problem - everyone over here with mac
uses mpi. If you send us logs to petsc-mait with the mpi problems [try
--download-mpich] - we can suggest fixes.

Satish


On Thu, 26 Apr 2012, Barry Smith wrote:

> 
>   Please send to petsc-maint at mcs.anl.gov HOW it "complains" that it needs mpi for hypre. (Send ALL error messages, configure.log make.log)
> 
>   We've never tested this configuration, there may be some flags that need to be changed to get the build going; and we don't know if the build will work. But we can get started on helping you test it with a complete error message to petsc-maint
> 
>    Barry
> 
> On Apr 26, 2012, at 8:52 PM, Kyungjoo Kim wrote:
> 
> > Hello,
> > 
> > 
> > While compiling PETSC without mpi, it complains that he needs mpi for hypre.
> > But I already install hypre also without mpi (hypre give me a option that I can install without mpi).
> > 
> > So, without mpi, I can compile PETSC but it removes all hypre interfaces, too. 
> > 
> > is there any detouring way ? Somehow my mac does not like mpi, so I would like to have no-mpi version for testing in my mac.
> > 
> > 
> > 
> 
> 



More information about the petsc-users mailing list