[petsc-users] Errors in installing PETSc

[Satish Balay]

On Tue, 13 Sep 2016, Kong (Non-US), Fande wrote:

> On Tue, Sep 13, 2016 at 12:05 PM, Satish Balay <balay at mcs.anl.gov> wrote:
> 
> > On Tue, 13 Sep 2016, Kong (Non-US), Fande wrote:
> >
> > > > However - as Matt refered to - its best to use latest petsc-3.7
> > > > release. Does MOOSE require 3.6?
> > > >
> > > >
> > > I think MOOSE works fine with petsc-3.7 as long as you do not use
> > > superlu_dist.  The superlu_dist has bugs in the latest petsc-3.7.
> >
> >
> Thanks, Satish,
> 
> 
> > Hm - I don't see any on reports on the superlu_dist bug tracker [or
> > perhaps I missed some of the e-mail tarffic on petsc lists on this
> > issue]
> >
> > https://urldefense.proofpoint.com/v2/url?u=https-3A__github.
> > com_xiaoyeli_superlu-5Fdist_issues&d=CwIBAg&c=
> > 54IZrppPQZKX9mLzcGdPfFD1hxrcB__aEkJFOKJFd00&r=DUUt3SRGI0_
> > JgtNaS3udV68GRkgV4ts7XKfj2opmiCY&m=sbGb6NjNAkTFB_rboPxuOUZE_
> > e0JXSaKebcodwE2J_s&s=p-u5ET_flkwmIQlbzL2ahWQkYGmzrxNtQ3MfbaI63rg&e=
> >
> >
> The bug is related with the algorithm in superlu_dist. It is a runtime bug
> not an compiling  issue.  I am thinking  how to reproduce this issue on the
> petsc side.

superlu_dist has git history from apporximately version 4.1 [the
version thats used in petsc-3.6] - so you you can get a working and
buggy build with the oldest and latest git snapshots - then you might
be able to do 'git bisect' to narrow down to the change thats causing
this issue.

Satish