March 2012 - petsc-dev - mailman3.cels.anl.gov

Checking for PETSc Cuda Support
by Dave Nystrom 19 Mar '12

19 Mar '12

Is there a way to ask PETSc if it has cuda support compiled in from a Fortran code? Thanks, Dave

1 0

GAMG hiding MG -help
by Mark F. Adams 19 Mar '12

19 Mar '12

I just found that GAMG seems to be hiding MG's parameters when -help is used. I have this code in my PCSetUp_GAMG: /* should be called in PCSetFromOptions_GAMG(), but cannot be called prior to PCMGSetLevels() */ ierr = PetscObjectOptionsBegin((PetscObject)pc);CHKERRQ(ierr); ierr = PCSetFromOptions_MG( pc ); CHKERRQ(ierr); ierr = PetscOptionsEnd();CHKERRQ(ierr); if (mg->galerkin != 2) SETERRQ(wcomm,PETSC_ERR_USER,"GAMG does Galerkin manually so the -pc_mg_galerkin option must not be used."); Which seems funky, but I think I cloned this or had some reason to put this here. And I set: pc->ops->setfromoptions = PCSetFromOptions_GAMG; in PCCreate_GAMG. Any advice on how I should fix this? Mark

1 0

Re: [petsc-dev] undefined reference to `SNES_LineSearch'
by Peter Brune 19 Mar '12

19 Mar '12

Should be fixed. - Peter On Mon, Mar 19, 2012 at 10:56 AM, Peter Brune <prbrune(a)gmail.com> wrote: > Sorry about that; I'm removing a lot of stuff. I found where that didn't > get removed, and am about to deal with it. > > - Peter > > > On Mon, Mar 19, 2012 at 10:48 AM, Brad Aagaard <baagaard(a)usgs.gov> wrote: > >> Peter, >> >> I get SNES_LineSearch as an undefined symbol when I do "make test" with >> the petsc-dev I just pulled. My guess is that your name change might not >> have been entirely consistent. >> >> Is it SNESLineSearch or SNES_LineSearch? >> >> Brad >> > >

1 0

Re: [petsc-dev] Reminder: *every* symbol with extern linkage *must* be namespaced
by Jed Brown 19 Mar '12

19 Mar '12

On Mon, Mar 19, 2012 at 10:52, Matthew Knepley <knepley(a)gmail.com> wrote: > Note that every // comment needs to be fixed. There are still a number of >> Wunused-but-set-variable >> > > Yes, I was waiting on the comments. That will be a ton of mindless work. > $ perl -pi -e 's,//(.*)$,/* $1 */,' src/dm/impls/complex/ctetgen.c $ perl -pi -e 's,//,,' src/dm/impls/complex/ctetgen.c # remove the funny end-of-line // pushed http://petsc.cs.iit.edu/petsc/petsc-dev/rev/d9e53d1d4cb4

1 0

Re: [petsc-dev] undefined reference to `SNES_LineSearch'
by Peter Brune 19 Mar '12

19 Mar '12

Sorry about that; I'm removing a lot of stuff. I found where that didn't get removed, and am about to deal with it. - Peter On Mon, Mar 19, 2012 at 10:48 AM, Brad Aagaard <baagaard(a)usgs.gov> wrote: > Peter, > > I get SNES_LineSearch as an undefined symbol when I do "make test" with > the petsc-dev I just pulled. My guess is that your name change might not > have been entirely consistent. > > Is it SNESLineSearch or SNES_LineSearch? > > Brad >

1 0

Re: [petsc-dev] Reminder: *every* symbol with extern linkage *must* be namespaced
by Sean Farley 19 Mar '12

19 Mar '12

> > It's a normal int, not a PetscInt, so at least it will always fit in that > pointer variable. > > You can silence the warning by changing the line from > > leftarray[0] = (point) i; // The array length. > > to > > leftarray[0] = (point) (PETSC_UINTPTR_T)i; // The array length. > What I don't understand, is that with all this work done (thanks Matt!) why is it bundled with PETSc and not as a forked project? I mean, ctetgen seems like it would have value for other projects and most of the work is just scrubbing out the C++ (which, again, Matt just did the bulk of!). Why not just make an interface to it? This would have the added bonus of solving the license quirks (and, yes, Matt, there is a difference between having a fork / linking vs. *bundling* the code).

1 0

Re: [petsc-dev] Reminder: *every* symbol with extern linkage *must* be namespaced
by Matthew Knepley 19 Mar '12

19 Mar '12

On Mon, Mar 19, 2012 at 10:23 AM, Jed Brown <jedbrown(a)mcs.anl.gov> wrote: > On Mon, Mar 19, 2012 at 09:59, Matthew Knepley <knepley(a)gmail.com> wrote: > >> ctetgen.c:10050:20: warning: cast to pointer from integer of different >>> size [-Wint-to-pointer-cast] >>> ctetgen.c:10070:21: warning: cast to pointer from integer of different >>> size [-Wint-to-pointer-cast] >>> >> >> I can't change this, since this is what the package does. I guess they >> get away with it now. >> > > It's a normal int, not a PetscInt, so at least it will always fit in that > pointer variable. > > You can silence the warning by changing the line from > > leftarray[0] = (point) i; // The array length. > > to > > leftarray[0] = (point) (PETSC_UINTPTR_T)i; // The array length. > Oh, I see now. Fixed. > Note that every // comment needs to be fixed. There are still a number of > Wunused-but-set-variable > Yes, I was waiting on the comments. That will be a ton of mindless work. I know about the unused stuff. I was reluctant to remove a bunch of these statement so that I could remain statement compatible with TetGen for debugging purposes. I can comment them out. > /home/jed/petsc/src/dm/impls/complex/ctetgen.c: In function > ‘TetGenMeshBTreeSort’: > /home/jed/petsc/src/dm/impls/complex/ctetgen.c:10050:20: warning: cast to > pointer from integer of different size [-Wint-to-pointer-cast] > /home/jed/petsc/src/dm/impls/complex/ctetgen.c:10070:21: warning: cast to > pointer from integer of different size [-Wint-to-pointer-cast] > /home/jed/petsc/src/dm/impls/complex/ctetgen.c: In function > ‘TetGenMeshFindDirection2’: > /home/jed/petsc/src/dm/impls/complex/ctetgen.c:13153:49: warning: variable > ‘cop’ set but not used [-Wunused-but-set-variable] > /home/jed/petsc/src/dm/impls/complex/ctetgen.c: In function > ‘TetGenMeshConstrainedFacets’: > /home/jed/petsc/src/dm/impls/complex/ctetgen.c:15042:22: warning: variable > ‘delaunayflag’ set but not used [-Wunused-but-set-variable] > /home/jed/petsc/src/dm/impls/complex/ctetgen.c: In function > ‘TetGenMeshConstrainedFacets2’: > /home/jed/petsc/src/dm/impls/complex/ctetgen.c:15340:22: warning: variable > ‘delaunayflag’ set but not used [-Wunused-but-set-variable] > /home/jed/petsc/src/dm/impls/complex/ctetgen.c: In function > ‘TetGenMeshRepairEncSubs’: > /home/jed/petsc/src/dm/impls/complex/ctetgen.c:18617:8: warning: variable > ‘oldptnum’ set but not used [-Wunused-but-set-variable] > /home/jed/petsc/src/dm/impls/complex/ctetgen.c: In function > ‘TetGenMeshOptimize’: > /home/jed/petsc/src/dm/impls/complex/ctetgen.c:19734:18: warning: variable > ‘maxdihed’ set but not used [-Wunused-but-set-variable] > /home/jed/petsc/src/dm/impls/complex/ctetgen.c: At top level: > /home/jed/petsc/src/dm/impls/complex/ctetgen.c:2026:23: warning: > ‘ListHasItem’ defined but not used [-Wunused-function] > > > And the complex/int64 build is broken: > > /home/jed/petsc/src/dm/impls/complex/complex.c:4047:36: error: assigning > to 'PetscReal' (aka 'double') from incompatible type 'PetscScalar' (aka > 'complex<double>'); > in->pointlist[idx*dim + d] = array[off+d]; > ^ ~~~~~~~~~~~~ > /home/jed/petsc/src/dm/impls/complex/complex.c:4049:14: error: no matching > function for call to 'DMComplexGetLabelValue' > ierr = DMComplexGetLabelValue(boundary, "marker", v, > &in->pointmarkerlist[idx]);CHKERRQ(ierr); > ^~~~~~~~~~~~~~~~~~~~~~ > /home/jed/petsc/src/dm/impls/complex/complex.c:1821:16: note: candidate > function not viable: no known conversion from 'int *' to 'PetscInt *' (aka > 'long long *') for 4th argument; > PetscErrorCode DMComplexGetLabelValue(DM dm, const char name[], PetscInt > point, PetscInt *value) > ^ > /home/jed/petsc/src/dm/impls/complex/complex.c:4082:14: error: no matching > function for call to 'DMComplexGetLabelValue' > ierr = DMComplexGetLabelValue(boundary, "marker", f, > &in->facetmarkerlist[idx]);CHKERRQ(ierr); > ^~~~~~~~~~~~~~~~~~~~~~ > /home/jed/petsc/src/dm/impls/complex/complex.c:1821:16: note: candidate > function not viable: no known conversion from 'int *' to 'PetscInt *' (aka > 'long long *') for 4th argument; > PetscErrorCode DMComplexGetLabelValue(DM dm, const char name[], PetscInt > point, PetscInt *value) > ^ > /home/jed/petsc/src/dm/impls/complex/complex.c:4300:36: error: assigning > to 'PetscReal' (aka 'double') from incompatible type 'PetscScalar' (aka > 'complex<double>'); > in->pointlist[idx*dim + d] = array[off+d]; > ^ ~~~~~~~~~~~~ > /home/jed/petsc/src/dm/impls/complex/complex.c:4302:14: error: no matching > function for call to 'DMComplexGetLabelValue' > ierr = DMComplexGetLabelValue(dm, "marker", v, > &in->pointmarkerlist[idx]);CHKERRQ(ierr); > ^~~~~~~~~~~~~~~~~~~~~~ > /home/jed/petsc/src/dm/impls/complex/complex.c:1821:16: note: candidate > function not viable: no known conversion from 'int *' to 'PetscInt *' (aka > 'long long *') for 4th argument; > PetscErrorCode DMComplexGetLabelValue(DM dm, const char name[], PetscInt > point, PetscInt *value) > ^ > 5 errors generated. > make[2]: *** [CMakeFiles/petsc.dir/src/dm/impls/complex/complex.c.o] Error > 1 > make[2]: *** Waiting for unfinished jobs.... > /home/jed/petsc/src/dm/impls/complex/ctetgen.c:14202:26: warning: if > statement has empty body [-Wempty-body] > if (!crossface.tet); {SETERRQ(PETSC_COMM_SELF, PETSC_ERR_PLIB, "Not > handled yet");} > ^ > /home/jed/petsc/src/dm/impls/complex/ctetgen.c:2026:23: warning: unused > function 'ListHasItem' [-Wunused-function] > static PetscErrorCode ListHasItem(List *l, void *checkitem, int *idx) > I will fix this somehow. Matt -- What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead. -- Norbert Wiener

1 0

undefined reference to `SNES_LineSearch'
by Brad Aagaard 19 Mar '12

19 Mar '12

Peter, I get SNES_LineSearch as an undefined symbol when I do "make test" with the petsc-dev I just pulled. My guess is that your name change might not have been entirely consistent. Is it SNESLineSearch or SNES_LineSearch? Brad

1 0

Re: [petsc-dev] Reminder: *every* symbol with extern linkage *must* be namespaced
by Jed Brown 19 Mar '12

19 Mar '12

On Mon, Mar 19, 2012 at 09:59, Matthew Knepley <knepley(a)gmail.com> wrote: > ctetgen.c:10050:20: warning: cast to pointer from integer of different >> size [-Wint-to-pointer-cast] >> ctetgen.c:10070:21: warning: cast to pointer from integer of different >> size [-Wint-to-pointer-cast] >> > > I can't change this, since this is what the package does. I guess they get > away with it now. > It's a normal int, not a PetscInt, so at least it will always fit in that pointer variable. You can silence the warning by changing the line from leftarray[0] = (point) i; // The array length. to leftarray[0] = (point) (PETSC_UINTPTR_T)i; // The array length. Note that every // comment needs to be fixed. There are still a number of Wunused-but-set-variable /home/jed/petsc/src/dm/impls/complex/ctetgen.c: In function ‘TetGenMeshBTreeSort’: /home/jed/petsc/src/dm/impls/complex/ctetgen.c:10050:20: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast] /home/jed/petsc/src/dm/impls/complex/ctetgen.c:10070:21: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast] /home/jed/petsc/src/dm/impls/complex/ctetgen.c: In function ‘TetGenMeshFindDirection2’: /home/jed/petsc/src/dm/impls/complex/ctetgen.c:13153:49: warning: variable ‘cop’ set but not used [-Wunused-but-set-variable] /home/jed/petsc/src/dm/impls/complex/ctetgen.c: In function ‘TetGenMeshConstrainedFacets’: /home/jed/petsc/src/dm/impls/complex/ctetgen.c:15042:22: warning: variable ‘delaunayflag’ set but not used [-Wunused-but-set-variable] /home/jed/petsc/src/dm/impls/complex/ctetgen.c: In function ‘TetGenMeshConstrainedFacets2’: /home/jed/petsc/src/dm/impls/complex/ctetgen.c:15340:22: warning: variable ‘delaunayflag’ set but not used [-Wunused-but-set-variable] /home/jed/petsc/src/dm/impls/complex/ctetgen.c: In function ‘TetGenMeshRepairEncSubs’: /home/jed/petsc/src/dm/impls/complex/ctetgen.c:18617:8: warning: variable ‘oldptnum’ set but not used [-Wunused-but-set-variable] /home/jed/petsc/src/dm/impls/complex/ctetgen.c: In function ‘TetGenMeshOptimize’: /home/jed/petsc/src/dm/impls/complex/ctetgen.c:19734:18: warning: variable ‘maxdihed’ set but not used [-Wunused-but-set-variable] /home/jed/petsc/src/dm/impls/complex/ctetgen.c: At top level: /home/jed/petsc/src/dm/impls/complex/ctetgen.c:2026:23: warning: ‘ListHasItem’ defined but not used [-Wunused-function] And the complex/int64 build is broken: /home/jed/petsc/src/dm/impls/complex/complex.c:4047:36: error: assigning to 'PetscReal' (aka 'double') from incompatible type 'PetscScalar' (aka 'complex<double>'); in->pointlist[idx*dim + d] = array[off+d]; ^ ~~~~~~~~~~~~ /home/jed/petsc/src/dm/impls/complex/complex.c:4049:14: error: no matching function for call to 'DMComplexGetLabelValue' ierr = DMComplexGetLabelValue(boundary, "marker", v, &in->pointmarkerlist[idx]);CHKERRQ(ierr); ^~~~~~~~~~~~~~~~~~~~~~ /home/jed/petsc/src/dm/impls/complex/complex.c:1821:16: note: candidate function not viable: no known conversion from 'int *' to 'PetscInt *' (aka 'long long *') for 4th argument; PetscErrorCode DMComplexGetLabelValue(DM dm, const char name[], PetscInt point, PetscInt *value) ^ /home/jed/petsc/src/dm/impls/complex/complex.c:4082:14: error: no matching function for call to 'DMComplexGetLabelValue' ierr = DMComplexGetLabelValue(boundary, "marker", f, &in->facetmarkerlist[idx]);CHKERRQ(ierr); ^~~~~~~~~~~~~~~~~~~~~~ /home/jed/petsc/src/dm/impls/complex/complex.c:1821:16: note: candidate function not viable: no known conversion from 'int *' to 'PetscInt *' (aka 'long long *') for 4th argument; PetscErrorCode DMComplexGetLabelValue(DM dm, const char name[], PetscInt point, PetscInt *value) ^ /home/jed/petsc/src/dm/impls/complex/complex.c:4300:36: error: assigning to 'PetscReal' (aka 'double') from incompatible type 'PetscScalar' (aka 'complex<double>'); in->pointlist[idx*dim + d] = array[off+d]; ^ ~~~~~~~~~~~~ /home/jed/petsc/src/dm/impls/complex/complex.c:4302:14: error: no matching function for call to 'DMComplexGetLabelValue' ierr = DMComplexGetLabelValue(dm, "marker", v, &in->pointmarkerlist[idx]);CHKERRQ(ierr); ^~~~~~~~~~~~~~~~~~~~~~ /home/jed/petsc/src/dm/impls/complex/complex.c:1821:16: note: candidate function not viable: no known conversion from 'int *' to 'PetscInt *' (aka 'long long *') for 4th argument; PetscErrorCode DMComplexGetLabelValue(DM dm, const char name[], PetscInt point, PetscInt *value) ^ 5 errors generated. make[2]: *** [CMakeFiles/petsc.dir/src/dm/impls/complex/complex.c.o] Error 1 make[2]: *** Waiting for unfinished jobs.... /home/jed/petsc/src/dm/impls/complex/ctetgen.c:14202:26: warning: if statement has empty body [-Wempty-body] if (!crossface.tet); {SETERRQ(PETSC_COMM_SELF, PETSC_ERR_PLIB, "Not handled yet");} ^ /home/jed/petsc/src/dm/impls/complex/ctetgen.c:2026:23: warning: unused function 'ListHasItem' [-Wunused-function] static PetscErrorCode ListHasItem(List *l, void *checkitem, int *idx) ^

1 0

Re: [petsc-dev] Reminder: *every* symbol with extern linkage *must* be namespaced
by Jed Brown 19 Mar '12

19 Mar '12

On Mon, Mar 19, 2012 at 10:05, Matthew Knepley <knepley(a)gmail.com> wrote: > Matt has significantly modified tetgen (removing its superfluous use of >> C++) so all these symbols are different. >> > > AND you could have tested your assumption of a problem by trying it with > --download-tetgen, which > works because this is how I tested it. The --download-ctetgen option will > work shortly. > > What is this with criticizing what I do before even finding out what I > actually did? > Oh come on, how is Satish supposed to know how you intend to do things? He just made the standard argument against bundling. PETSc normally avoids bundling (unlike a lot of packages) and what you are doing with ctetgen is unconventional, so it was natural to point it out.

1 0