July 2016 - petsc-dev - mailman3.cels.anl.gov

Re: [petsc-dev] array ownership of MatCreateMPIAIJWithSplitArrays and MatCreateSeqAIJWithArrays
by Stefano Zampini 24 Jul '16

24 Jul '16

Barry, I decided to use your PetscContainer solution and not change the API. Thanks Stefano On Jul 23, 2016, at 12:58 PM, Barry Smith <bsmith(a)mcs.anl.gov> wrote: > >> On Jul 23, 2016, at 2:36 PM, Stefano Zampini <stefano.zampini(a)gmail.com> wrote: >> >> I’m interfacing a FEM library with PETSc. >> >> They already use Hypre and so they have a valid distributed CSR format (except for the diagonal entry of the diagonal block :-)) >> >> For efficiency reasons, I was thinking of using MatCreateMPIAIJWithSplitArrays and MatCreateSeqAIJWithArrays; >> However, If using these functions as they are now I will need to keep track of the allocated arrays, even with referencing objects, and this could be annoying and cumbersome. >> >> A trivial fix would be to add an extra PetscCopyMode parameter to these routines, so that PETSc can take ownership of the arrays. >> Thoughts? Any objection? > > Sounds reasonable except that PETSc uses PetscFree() which may not be a direct call to free() so that if the user used malloc() to obtain the space (which is more likely than PetscMalloc()) bad stuff will happen. One could introduce PETSC_OWN_RAWPOINTER which tells PETSc it should free the space with a regular free() instead of PetscFree() but then code get's more complicated since you'll need to keep a flag indicating this and call the appropriate free() based on it. Of course if you have control over the FEM library source you could change it to use PetscMalloc() to obtain the pointers. > > An alternative is for you to create a PetscContainer for each array pointer, provide a PetscContainerSetUserDestroy() that calls free on the pointer and then use PetscObjectCompose() to attach the containers to the matrix, now call PetscContainerDestroy() on all the containers. When the matrix is destroyed it will go through all the composed objects destroying each of them which will free up your space. No need to change PETSc source or APIs. > > Barry > > > > > >> >

1 0

Re: [petsc-dev] array ownership of MatCreateMPIAIJWithSplitArrays and MatCreateSeqAIJWithArrays
by Barry Smith 23 Jul '16

23 Jul '16

> On Jul 23, 2016, at 2:36 PM, Stefano Zampini <stefano.zampini(a)gmail.com> wrote: > > I’m interfacing a FEM library with PETSc. > > They already use Hypre and so they have a valid distributed CSR format (except for the diagonal entry of the diagonal block :-)) > > For efficiency reasons, I was thinking of using MatCreateMPIAIJWithSplitArrays and MatCreateSeqAIJWithArrays; > However, If using these functions as they are now I will need to keep track of the allocated arrays, even with referencing objects, and this could be annoying and cumbersome. > > A trivial fix would be to add an extra PetscCopyMode parameter to these routines, so that PETSc can take ownership of the arrays. > Thoughts? Any objection? Sounds reasonable except that PETSc uses PetscFree() which may not be a direct call to free() so that if the user used malloc() to obtain the space (which is more likely than PetscMalloc()) bad stuff will happen. One could introduce PETSC_OWN_RAWPOINTER which tells PETSc it should free the space with a regular free() instead of PetscFree() but then code get's more complicated since you'll need to keep a flag indicating this and call the appropriate free() based on it. Of course if you have control over the FEM library source you could change it to use PetscMalloc() to obtain the pointers. An alternative is for you to create a PetscContainer for each array pointer, provide a PetscContainerSetUserDestroy() that calls free on the pointer and then use PetscObjectCompose() to attach the containers to the matrix, now call PetscContainerDestroy() on all the containers. When the matrix is destroyed it will go through all the composed objects destroying each of them which will free up your space. No need to change PETSc source or APIs. Barry >

1 0

array ownership of MatCreateMPIAIJWithSplitArrays and MatCreateSeqAIJWithArrays
by Stefano Zampini 23 Jul '16

23 Jul '16

I’m interfacing a FEM library with PETSc. They already use Hypre and so they have a valid distributed CSR format (except for the diagonal entry of the diagonal block :-)) For efficiency reasons, I was thinking of using MatCreateMPIAIJWithSplitArrays and MatCreateSeqAIJWithArrays; However, If using these functions as they are now I will need to keep track of the allocated arrays, even with referencing objects, and this could be annoying and cumbersome. A trivial fix would be to add an extra PetscCopyMode parameter to these routines, so that PETSc can take ownership of the arrays. Thoughts? Any objection?

1 0

Re: [petsc-dev] Watching now
by Sean Farley 22 Jul '16

22 Jul '16

Barry Smith <bsmith(a)mcs.anl.gov> writes: >> On Jul 22, 2016, at 9:55 PM, Sean Farley <sean(a)farley.io> wrote: >> >> Barry Smith <bsmith(a)mcs.anl.gov> writes: >> >>> Man, why are comments disabled? >> >> Because julia indexes arrays by starting with 1. > > Damn Fortran programmers Not the first time I've uttered those words.

1 0

Re: [petsc-dev] Watching now
by Barry Smith 22 Jul '16

22 Jul '16

> On Jul 22, 2016, at 9:55 PM, Sean Farley <sean(a)farley.io> wrote: > > Barry Smith <bsmith(a)mcs.anl.gov> writes: > >> Man, why are comments disabled? > > Because julia indexes arrays by starting with 1. Damn Fortran programmers

1 0

Re: [petsc-dev] Watching now
by Sean Farley 22 Jul '16

22 Jul '16

Barry Smith <bsmith(a)mcs.anl.gov> writes: > Man, why are comments disabled? Because julia indexes arrays by starting with 1.

1 0

Re: [petsc-dev] Watching now
by Barry Smith 22 Jul '16

22 Jul '16

Man, why are comments disabled? > On Jul 22, 2016, at 8:28 PM, Matthew Knepley <knepley(a)gmail.com> wrote: > > https://www.youtube.com/watch?v=Sxb3U_PjmPA&list=PLP8iPy9hna6SQPwZUDtAM59-w… > > 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

Watching now
by Matthew Knepley 22 Jul '16

22 Jul '16

https://www.youtube.com/watch?v=Sxb3U_PjmPA&list=PLP8iPy9hna6SQPwZUDtAM59-w… 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

Call Center Services for Development Titan LLC
by Emily Kirkland 20 Jul '16

20 Jul '16

How many times have you had to apologize to a customer for a delay in answering an inquiry? How many times does this need to happen before you realize that you’re losing out? Gocallcenter.click is a service that compares quotes for Call Centers that can provide your business with Telemarketing, Customer Support, Appointment Setting and more. Please fill in the form or reply with your contact information to receive a Quote. Best regards, Emily Kirkland Call Center Specialist Emily(a)gocallcenter.click gocallcenter.click (http://www.gocallcenter.click/default.aspx?c=&p=(407) 388-5337&f=&l=&z=32792&e=petsc-dev(a)mcs.anl.gov&utm_source=is&utm_medium=email&utm_campaign=ZO3) Click here to unsubscribe (http://www.gocallcenter.click/[email protected])

1 0

Re: [petsc-dev] petsc4py and DMDA
by Lisandro Dalcin 20 Jul '16

20 Jul '16

On 20 July 2016 at 08:14, gouarin <loic.gouarin(a)math.u-psud.fr> wrote: > Le 20/07/2016 à 04:59, Matthew Knepley a écrit : >> >> On Tue, Jul 19, 2016 at 11:12 PM, gouarin <loic.gouarin(a)math.u-psud.fr >> <mailto:[email protected]>> wrote: >> >> Hi, >> >> I would like to use DMGetLocalVector and DMGetGlobalVector with >> petsc4py but I didn't find the appropriate methods. If they are not >> implemented, I can do a PR with these functions and their Restore >> part. >> >> >> You are right, these are missing. >> > OK I will add these functions and do a PR. We would also need the restore() methods. -- Lisandro Dalcin ============ Research Scientist Computer, Electrical and Mathematical Sciences & Engineering (CEMSE) Extreme Computing Research Center (ECRC) King Abdullah University of Science and Technology (KAUST) http://ecrc.kaust.edu.sa/ 4700 King Abdullah University of Science and Technology al-Khawarizmi Bldg (Bldg 1), Office # 0109 Thuwal 23955-6900, Kingdom of Saudi Arabia http://www.kaust.edu.sa Office Phone: +966 12 808-0459

1 0