Re: [deal.II] Running multiple programs at the same time
Hello Wolfgang, The memory bus bandwidth seems the most reliable answer. But two more questions arise: - Matrix-free methods would have this problems minimised? - Will this affect also to MPI parallelization scalability in this computer? Regards, Toni El dimarts, 11 febrer de 2020 22:37:31 UTC+1, Wolfgang Bangerth va escriure: > > On 2/11/20 2:27 PM, Toni Vidal wrote: > > > > That did not solve the problem with step 6. I got the same times. > > > > Indeed I have installed deal.II without threads (DEAL_II_WITH_THREADS = > > OFF) and I set in my .basrc OMP_NUM_THREADS=1. > > There are many other possible reasons for contention. For example, most > finite element programs are limited by the transfer of data from memory > to the processor. If you have just one program running, then only one > program is using the memory bus and is getting its full speed. But if > you have multiple programs running, then they are all competing for the > same bandwidth on the memory bus, and they will also be slowed down by > more than a single program would be. > > It could also be that you have, say, 4 cores on your processor but 3 > other programs currently running. Then running one instance of your code > will get a full core, but if you ran four, the total of 7 codes would > have to compete for 4 cores, and all would be slowed down. > > By the way, to see whether your program really is using only one thread, > you can run the program 'top' in a separate command line window. It will > show you which percentage of a processor each running job takes up. > > Best > W. > > > -- > > Wolfgang Bangerth email: bang...@colostate.edu > > www: http://www.math.colostate.edu/~bangerth/ > -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to dealii+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/0dab48ed-19c0-4d99-af01-717d008cfdf2%40googlegroups.com.
Re: [deal.II] Running multiple programs at the same time
Hi Ahmad, I commented the output results method but it is the times did not change significantly. Toni El dimarts, 11 febrer de 2020 22:40:42 UTC+1, Ahmad Shahba va escriure: > > I was just wondering how much I/O operations contribute to your timings. > What would happen if you minimize the I/O activities, maybe comment out > output_results method and see if anything changes? > > Regards > Ahmad > > On Tue, Feb 11, 2020 at 16:27 Toni Vidal > wrote: > >> Hi David, >> >> That did not solve the problem with step 6. I got the same times. >> >> Indeed I have installed deal.II without threads (DEAL_II_WITH_THREADS = >> OFF) and I set in my .basrc OMP_NUM_THREADS=1. >> >> Any other idea? >> >> >> El dimarts, 11 febrer de 2020 17:53:53 UTC+1, David Wells va escriure: >>> >>> Hi Toni, >>> >>> I think that this is due to each individual program creating the same >>> number of threads as you have physical processors. Try adding >>> >>> MultithreadInfo::set_thread_limit(1); >>> >>> at the top of your code to prevent this from happening. Let us know if >>> this works! >>> >>> Thanks, >>> David >>> >>> On Tue, Feb 11, 2020 at 11:45 AM Toni Vidal >>> wrote: >>> > >>> > >>> > Dear deal.ii users and developers, >>> > >>> > I am currently running a deal.II based code thousands of times with >>> different input parameters. Each simulation takes about 30 seconds in a >>> single processor. To do this, I have made a python script that runs 4 >>> simulations at a time (using the multiprocessing module). However, each >>> simulation takes about 60 seconds and my have 8 cores (Intel® Core™ >>> i7-9700K CPU @ 3.60GHz × 8). Is it not supposed to take approximately the >>> same amount of because the processors are independent? Am I >>> > >>> > In order to isolate the problem I have executed deal.II's step 6 (with >>> 12 cycles and 1e5 maximum solver steps) 1, 2 and 4 times at the same (using >>> different terminals). >>> > >>> > 1 running programs ~ 32 s user time >>> > 2 running programs ~ 44 s user time >>> > 4 running programs ~ 80 s user time >>> > >>> > Why the programs does not take the same time even though my computer >>> have 8 cores? >>> > Any idea? Am I missing something obvious? >>> > >>> > Ton Vidal >>> > >>> > -- >>> > The deal.II project is located at http://www.dealii.org/ >>> > For mailing list/forum options, see >>> https://groups.google.com/d/forum/dealii?hl=en >>> > --- >>> > You received this message because you are subscribed to the Google >>> Groups "deal.II User Group" group. >>> > To unsubscribe from this group and stop receiving emails from it, send >>> an email to dea...@googlegroups.com. >>> > To view this discussion on the web visit >>> https://groups.google.com/d/msgid/dealii/767ae142-a02d-4bc2-a454-d4c385d1b217%40googlegroups.com. >>> >>> >>> >> -- >> The deal.II project is located at http://www.dealii.org/ >> For mailing list/forum options, see >> https://groups.google.com/d/forum/dealii?hl=en >> --- >> You received this message because you are subscribed to the Google Groups >> "deal.II User Group" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to dea...@googlegroups.com . >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/dealii/36de50e0-355b-4337-a9ad-0512229a2f80%40googlegroups.com >> >> <https://groups.google.com/d/msgid/dealii/36de50e0-355b-4337-a9ad-0512229a2f80%40googlegroups.com?utm_medium=email_source=footer> >> . >> > -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to dealii+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/d54ed4f2-8b68-4b02-b641-dce0440e2d17%40googlegroups.com.
Re: [deal.II] Running multiple programs at the same time
Hi David, That did not solve the problem with step 6. I got the same times. Indeed I have installed deal.II without threads (DEAL_II_WITH_THREADS = OFF) and I set in my .basrc OMP_NUM_THREADS=1. Any other idea? El dimarts, 11 febrer de 2020 17:53:53 UTC+1, David Wells va escriure: > > Hi Toni, > > I think that this is due to each individual program creating the same > number of threads as you have physical processors. Try adding > > MultithreadInfo::set_thread_limit(1); > > at the top of your code to prevent this from happening. Let us know if > this works! > > Thanks, > David > > On Tue, Feb 11, 2020 at 11:45 AM Toni Vidal > wrote: > > > > > > Dear deal.ii users and developers, > > > > I am currently running a deal.II based code thousands of times with > different input parameters. Each simulation takes about 30 seconds in a > single processor. To do this, I have made a python script that runs 4 > simulations at a time (using the multiprocessing module). However, each > simulation takes about 60 seconds and my have 8 cores (Intel® Core™ > i7-9700K CPU @ 3.60GHz × 8). Is it not supposed to take approximately the > same amount of because the processors are independent? Am I > > > > In order to isolate the problem I have executed deal.II's step 6 (with > 12 cycles and 1e5 maximum solver steps) 1, 2 and 4 times at the same (using > different terminals). > > > > 1 running programs ~ 32 s user time > > 2 running programs ~ 44 s user time > > 4 running programs ~ 80 s user time > > > > Why the programs does not take the same time even though my computer > have 8 cores? > > Any idea? Am I missing something obvious? > > > > Ton Vidal > > > > -- > > The deal.II project is located at http://www.dealii.org/ > > For mailing list/forum options, see > https://groups.google.com/d/forum/dealii?hl=en > > --- > > You received this message because you are subscribed to the Google > Groups "deal.II User Group" group. > > To unsubscribe from this group and stop receiving emails from it, send > an email to dea...@googlegroups.com . > > To view this discussion on the web visit > https://groups.google.com/d/msgid/dealii/767ae142-a02d-4bc2-a454-d4c385d1b217%40googlegroups.com. > > > -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to dealii+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/36de50e0-355b-4337-a9ad-0512229a2f80%40googlegroups.com.
[deal.II] Running multiple programs at the same time
Dear deal.ii users and developers, I am currently running a deal.II based code thousands of times with different input parameters. Each simulation takes about 30 seconds in a single processor. To do this, I have made a python script that runs 4 simulations at a time (using the multiprocessing module). However, each simulation takes about 60 seconds and my have 8 cores (Intel® Core™ i7-9700K CPU @ 3.60GHz × 8). Is it not supposed to take approximately the same amount of because the processors are independent? Am I In order to isolate the problem I have executed deal.II's step 6 (with 12 cycles and 1e5 maximum solver steps) 1, 2 and 4 times at the same (using different terminals). 1 running programs ~ 32 s user time 2 running programs ~ 44 s user time 4 running programs ~ 80 s user time Why the programs does not take the same time even though my computer have 8 cores? Any idea? Am I missing something obvious? Ton Vidal -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to dealii+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/767ae142-a02d-4bc2-a454-d4c385d1b217%40googlegroups.com.
[deal.II] Re: Configuration of PETSc
Hello Konrad, I thing PETSc PCILU does not work in parallel. You could use the preconditioner PETScWrappers::PreconditionBlockJacobi that uses an ILU preconditioner for each process block. https://www.dealii.org/current/doxygen/deal.II/classPETScWrappers_1_1PreconditionBlockJacobi.html Also youcould use HYPRE parallell ILU from petsc if you have installed it. https://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/PC/PCHYPRE.html#PCHYPRE Or any other preconditioner that works in parallel. El divendres, 20 setembre de 2019 10:16:05 UTC+2, Konrad Simon va escriure: > > Dear deal.ii community, > > I am using deali.ii with PETSc and Trilinos. However, when I am using the > PETSc PreconditionILU I get an error that suggests that a solver package is > missing (with Trilinos it works). Petsc's PreconditionAMG works fine > (although not very efficiently for my problem). > Do I need to do any special configuration steps for PETSC? I followed the > instructions that are documented on the deal.ii pages on "how to configure > Petsc". https://www.dealii.org/current/external-libs/petsc.html > > Best, > Konrad > > This is the error: > > Running using PETSc. > Number of active cells: 262144 > Total number of cells: 24865 (on 7 levels) > Number of degrees of freedom: 1609920 (811200+798720) > [0]PETSC ERROR: - Error Message > -- > [0]PETSC ERROR: See > http://www.mcs.anl.gov/petsc/documentation/linearsolvertable.html for > possible LU and Cholesky solvers > [0]PETSC ERROR: Could not locate a solver package. Perhaps you must > ./configure with --download- > [0]PETSC ERROR: See http://www.mcs.anl.gov/petsc/documentation/faq.html > for trouble shooting. > [0]PETSC ERROR: Petsc Release Version 3.9.4, Sep, 11, 2018 > [0]PETSC ERROR: main on a x86_64 named thunder5 by u290231 Fri Sep 20 > 10:00:23 2019 > [0]PETSC ERROR: Configure options --with-shared-libraries=1 --with-x=0 > --with-mpi=yes --download-hypre=yes --with-64-bit-indices > --with-debugging=yes --with-hypre=yes > [0]PETSC ERROR: #1 MatGetFactor() line 4328 in > /scratch/cen/numgeo/lib/petsc-3.9.4/src/mat/interface/matrix.c > [0]PETSC ERROR: #2 PCSetUp_ILU() line 142 in > /scratch/cen/numgeo/lib/petsc-3.9.4/src/ksp/pc/impls/factor/ilu/ilu.c > [0]PETSC ERROR: #3 PCSetUp() line 923 in > /scratch/cen/numgeo/lib/petsc-3.9.4/src/ksp/pc/interface/precon.c > - > TimerOutput objects finalize timed values printed to the > screen by communicating over MPI in their destructors. > Since an exception is currently uncaught, this > synchronization (and subsequent output) will be skipped to > avoid a possible deadlock. > - > WARNING! There are options you set that were not used! > WARNING! could be spelling mistake, etc! > Option left: name:-p value: ../MsFEComplex/parameter.in > ERROR: Uncaught exception in MPI_InitFinalize on proc 0. Skipping > MPI_Finalize() to avoid a deadlock. > > > > Exception on processing: > > > An error occurred in line <421> of file > > > in function >void dealii::PETScWrappers::PreconditionILU::initialize(const > dealii::PETScWrappers::MatrixBase&, const > dealii::PETScWrappers::PreconditionILU::AdditionalData&) > The violated condition was: >ierr == 0 > Additional information: > deal.II encountered an error while calling a PETSc function. > The description of the error provided by PETSc is "See > http://www.mcs.anl.gov/petsc/documentation/linearsolvertable.html for > possible LU and Cholesky solvers". > The numerical value of the original error code is 92. > > > Aborting! > > -- > Primary job terminated normally, but 1 process returned > a non-zero exit code. Per user-direction, the job has been aborted. > -- > -- > mpirun detected that one or more processes exited with non-zero status, > thus causing > the job to be terminated. The first process to do so was: > > Process name: [[37149,1],0] > Exit code:1 > -- > -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to
