> On Mar 17, 2015, at 11:38 AM, <[email protected]> > <[email protected]> wrote: > > Barry, > > Thanks - I have used the preallocation routines to allocate the non zero row > length and am also looking at other ideas as discussed in these web pages. We > are expecting a petsc structured matrix to solve more quickly than the > corresponding CSR matrix but it is still slower!!
In order to know why it is slower we need to understand what part is slower: assembling the matrix, building the preconditioner, solving the system etc. Also the time depends a great deal on problem size, for smaller problems one method may be much faster, while for a much larger problem a different method will be much better. > > Would it be correct to also use the PetscWrapper::SparseMatrix::SparseMatrix > method to speed things up further? I don't know what that is. If you can run the problem with -log_summary and send us the output it would give a starting point for understanding where the time is being spent. Barry > > Thanks, > > Carol > > > > Dr Carol Brickley > BSc,PhD,ARCS,DIC,MBCS > > Senior Software Engineer > Applied Computer Science > DS+T, > AWE > Aldermaston > Reading > Berkshire > RG7 4PR > > Direct: 0118 9855035 > > > -----Original Message----- > From: Barry Smith [mailto:[email protected]] > Sent: 16 March 2015 21:35 > To: Brickley Carol AWE > Cc: [email protected] > Subject: EXTERNAL: Re: [petsc-users] petsc 3.4.3 > > > We'd need more specific details on why it is slower than you think it should > be. Is the it built with optimization (i.e. ./configure --with-debugging=0) ? > Are the solvers converging with more iterations than expected > http://www.mcs.anl.gov/petsc/documentation/faq.html#newton > http://www.mcs.anl.gov/petsc/documentation/faq.html#kspdiverged ? Is the > machine perhaps not suitable to for the simulations > http://www.mcs.anl.gov/petsc/documentation/faq.html#computers? > > Barry > > >> On Mar 16, 2015, at 2:18 PM, <[email protected]> >> <[email protected]> wrote: >> >> All >> >> I am using petsc 3.4.3 but finding it not working as quickly as it should. >> Any ideas? >> >> Carol >> >> Dr Carol Brickley >> BSc,PhD,ARCS,DIC,MBCS >> >> Senior Software Engineer >> Applied Computer Science >> DS+T, >> AWE >> Aldermaston >> Reading >> Berkshire >> RG7 4PR >> >> Direct: 0118 9855035 >> >> ___________________________________________________ >> ____________________________ The information in this email and in any >> attachment(s) is commercial in confidence. If you are not the named >> addressee(s) or if you receive this email in error then any distribution, >> copying or use of this communication or the information in it is strictly >> prohibited. Please notify us immediately by email at >> admin.internet(at)awe.co.uk, and then delete this message from your >> computer. While attachments are virus checked, AWE plc does not accept any >> liability in respect of any virus which is not detected. AWE Plc Registered >> in England and Wales Registration No 02763902 AWE, Aldermaston, Reading, RG7 >> 4PR >> > > > ___________________________________________________ > ____________________________ > > The information in this email and in any attachment(s) is > commercial in confidence. If you are not the named addressee(s) > or > if you receive this email in error then any distribution, copying or > use of this communication or the information in it is strictly > prohibited. Please notify us immediately by email at > admin.internet(at)awe.co.uk, and then delete this message from > your computer. While attachments are virus checked, AWE plc > does not accept any liability in respect of any virus which is not > detected. > > AWE Plc > Registered in England and Wales > Registration No 02763902 > AWE, Aldermaston, Reading, RG7 4PR
