Re: [petsc-users] Strange CUDA failure with a second petscfinalize with PETSc 3.16

19 Jan 2022

      Thanks Jacob for looking into this –

You can see the updated source code of ex11fc in the attachment – although there is not much that I modified (except for the jabbers I outputted). I also attached the full output (ex11fc.log) along with the configure.log file. It’s an old dual Xeon workstation (one of my “production” machines) with Linux kernel 5.4.0 and gcc 9.3.

I simply ran the code with

mpiexec -np 2 ex11fc -usecuda

for GPU test. And as stated before, calling without the “-usecuda” option shows no errors.

Please let me know if you find anything wrong with the configure/code.

Cheers,
Hao

From: Jacob Faibussowitsch<mailto:[email protected]>
Sent: Wednesday, January 19, 2022 3:38 AM
To: Hao DONG<mailto:[email protected]>
Cc: Junchao Zhang<mailto:[email protected]>; petsc-users<mailto:[email protected]>
Subject: Re: [petsc-users] Strange CUDA failure with a second petscfinalize with PETSc 3.16

Apologies, forgot to mention in my previous email but can you also include a copy of the full printout of the error message that you get? It will include all the command-line flags that you ran with (if any) so I can exactly mirror your environment.

Best regards,

Jacob Faibussowitsch
(Jacob Fai - booss - oh - vitch)

On Jan 18, 2022, at 14:06, Jacob Faibussowitsch <[email protected]<mailto:[email protected]>> wrote:

Can you send your updated source file as well as your configure.log (should be $PETSC_DIR/configure.log). I will see if I can reproduce the error on my end.
Best regards,

Jacob Faibussowitsch
(Jacob Fai - booss - oh - vitch)

On Jan 17, 2022, at 23:06, Hao DONG <[email protected]<mailto:[email protected]>> wrote:

Dear Junchao and Jacob,

Thanks a lot for the response – I also don’t understand why this is related to the device, especially on why the procedure can be successfully finished for *once* – As instructed, I tried to add a CHKERRA() macro after (almost) every petsc line – such as the initialization, mat assemble, ksp create, solve, mat destroy, etc. However, all other petsc commands returns with error code 0. It only gives me a similar (still not very informative) error after I call the petscfinalize (again for the second time), with error code 97:

[0]PETSC ERROR: --------------------- Error Message --------------------------------------------------------------
[0]PETSC ERROR: GPU error
[0]PETSC ERROR: cuda error 709 (cudaErrorContextIsDestroyed) : context is destroyed
[0]PETSC ERROR: See https://petsc.org/release/faq/ for trouble shooting.
[0]PETSC ERROR: Petsc Release Version 3.16.3, unknown
[0]PETSC ERROR: ./ex11f on a  named stratosphere by donghao Tue Jan 18 11:39:43 2022
[0]PETSC ERROR: Configure options --prefix=/opt/petsc/complex-double-with-cuda --with-cc=mpicc --with-cxx=mpicxx --with-fc=mpif90 COPTFLAGS="-O3 -mavx2" CXXOPTFLAGS="-O3 -mavx2" FOPTFLAGS="-O3 -ffree-line-length-none -mavx2" CUDAOPTFLAGS=-O3 --with-cxx-dialect=cxx14 --with-cuda-dialect=cxx14 --with-scalar-type=complex --with-precision=double --with-cuda-dir=/usr/local/cuda --with-debugging=1
[0]PETSC ERROR: #1 PetscFinalize() at /home/donghao/packages/petsc-current/src/sys/objects/pinit.c:1638
[0]PETSC ERROR: #2 User provided function() at User file:0

I can also confirm that rolling back to petsc 3.15 will *not* see the problem, even with the new nvidia driver. And petsc 3.16.3 with an old nvidia driver (470.42)  also get this same error. So it’s probably not connected to the nvidia driver.

Any idea on where I should look at next?
Thanks a lot in advance, and all the best,
Hao

From: Jacob Faibussowitsch<mailto:[email protected]>
Sent: Sunday, January 16, 2022 12:12 AM
To: Junchao Zhang<mailto:[email protected]>
Cc: petsc-users<mailto:[email protected]>; Hao DONG<mailto:[email protected]>
Subject: Re: [petsc-users] Strange CUDA failure with a second petscfinalize with PETSc 3.16

I don’t quite understand how it is getting to the CUDA error to be honest. None of the code in the stack trace is anywhere near the device code. Reading the error message carefully, it first chokes on PetscLogGetStageLog() from a call to PetscClassIdRegister():

PetscErrorCode PetscLogGetStageLog(PetscStageLog *stageLog)
{
  PetscFunctionBegin;
  PetscValidPointer(stageLog,1);
  if (!petsc_stageLog) {
    fprintf(stderr, "PETSC ERROR: Logging has not been enabled.\nYou might have forgotten to call PetscInitialize().\n");
    PETSCABORT(MPI_COMM_WORLD, PETSC_ERR_SUP); // Here
  }
  ...

But then jumps to PetscFinalize(). You can also see the "You might have forgotten to call PetscInitialize().” message in the error message, just under the topmost level of the stack trace.

Can you check the value of ierr of each function call (use the CHKERRA() macro to do so)? I suspect the problem here that errors occurring previously in the program are being ignored, leading to the garbled stack trace.

Best regards,

Jacob Faibussowitsch
(Jacob Fai - booss - oh - vitch)

On Jan 14, 2022, at 20:58, Junchao Zhang <[email protected]<mailto:[email protected]>> wrote:

Jacob,
   Could you have a look as it seems the "invalid device context" is in your newly added module?
   Thanks
--Junchao Zhang

On Fri, Jan 14, 2022 at 12:49 AM Hao DONG <[email protected]<mailto:[email protected]>> wrote:
Dear All,

I have encountered a peculiar problem when fiddling with a code with PETSC 3.16.3 (which worked fine with PETSc 3.15). It is a very straight forward PDE-based optimization code which repeatedly solves a linearized PDE problem with KSP in a subroutine (the rest of the code does not contain any PETSc related content). The main program provides the subroutine with an MPI comm. Then I set the comm as PETSC_COMM_WORLD to tell PETSC to attach to it (and detach with it when the solving is finished each time).

Strangely, I observe a CUDA failure whenever the petscfinalize is called for a *second* time.  In other words, the first and second PDE calculations with GPU are fine (with correct solutions). The petsc code just fails after the SECOND petscfinalize command is called. You can also see the PETSC config in the error message:

[1]PETSC ERROR: --------------------- Error Message --------------------------------------------------------------
[1]PETSC ERROR: GPU error
[1]PETSC ERROR: cuda error 201 (cudaErrorDeviceUninitialized) : invalid device context
[1]PETSC ERROR: See https://petsc.org/release/faq/ for trouble shooting.
[1]PETSC ERROR: Petsc Release Version 3.16.3, unknown
[1]PETSC ERROR: maxwell.gpu on a  named stratosphere by hao Fri Jan 14 10:21:05 2022
[1]PETSC ERROR: Configure options --prefix=/opt/petsc/complex-double-with-cuda --with-cc=mpicc --with-cxx=mpicxx --with-fc=mpif90 COPTFLAGS="-O3 -mavx2" CXXOPTFLAGS="-O3 -mavx2" FOPTFLAGS="-O3 -ffree-line-length-none -mavx2" CUDAOPTFLAGS=-O3 --with-cxx-dialect=cxx14 --with-cuda-dialect=cxx14 --with-scalar-type=complex --with-precision=double --with-cuda-dir=/usr/local/cuda --with-debugging=1
[1]PETSC ERROR: #1 PetscFinalize() at /home/hao/packages/petsc-current/src/sys/objects/pinit.c:1638
You might have forgotten to call PetscInitialize().
The EXACT line numbers in the error traceback are not available.
Instead the line number of the start of the function is given.
[1] #1 PetscAbortFindSourceFile_Private() at /home/hao/packages/petsc-current/src/sys/error/err.c:35
[1] #2 PetscLogGetStageLog() at /home/hao/packages/petsc-current/src/sys/logging/utils/stagelog.c:29
[1] #3 PetscClassIdRegister() at /home/hao/packages/petsc-current/src/sys/logging/plog.c:2376
[1] #4 MatMFFDInitializePackage() at /home/hao/packages/petsc-current/src/mat/impls/mffd/mffd.c:45
[1] #5 MatInitializePackage() at /home/hao/packages/petsc-current/src/mat/interface/dlregismat.c:163
[1] #6 MatCreate() at /home/hao/packages/petsc-current/src/mat/utils/gcreate.c:77

However, it doesn’t seem to affect the other part of my code, so the code can continue running until it gets to the petsc part again (the *third* time). Unfortunately, it doesn’t give me any further information even if I set the debugging to yes in the configure file. It also worth noting that PETSC without CUDA (i.e. with simple MATMPIAIJ) works perfectly fine.

I am able to re-produce the problem with a toy code modified from ex11f. Please see the attached file (ex11fc.F90) for details. Essentially  the code does the same thing as ex11f, but three times with a do loop. To do that I added an extra MPI_INIT/MPI_FINALIZE to ensure that the MPI communicator is not destroyed when PETSC_FINALIZE is called.  I used the PetscOptionsHasName utility to check if you have “-usecuda” in the options. So running the code with and without that option can give you a comparison w/o CUDA. I can see that the code also fails after the second loop of the KSP operation. Could you kindly shed some lights on this problem?

I should say that I am not even sure if the problem is from PETSc, as I also accidentally updated the NVIDIA driver (for now it is 510.06 with cuda 11.6). And it is well known that NVIDIA can give you some surprise in the updates (yes, I know I shouldn’t have touched that if it’s not broken). But my CUDA code without PETSC (which basically does the same PDE thing, but with cusparse/cublas directly) seems to work just fine after the update. It is also possible that my petsc code related to CUDA was not quite “legitimate” – I just use:
          MatSetType(A, MATMPIAIJCUSPARSE, ierr)
and
          MatCreateVecs(A, u, PETSC_NULL_VEC, ierr)
to make the data onto GPU. I would very much appreciate it if you could show me the “right” way to do that.

Thanks a lot in advance, and all the best,
Hao