Hi Sherry, I have this running with OMP, with cuSparse solves (PETSc CPU factorizations) Building SuperLU_dist w/o _OPENMP was not easy for me. We need to get a better way to do this. (Satish or Barry?) SuperLU works with one thread and two subdomains. With two threads I see this (appended). So this seems to be working in that before it was hanging. I set the solver up so that it does not use threads the first time it is called so that solvers can get any lazy allocations done in serial. This is just to be safe in that we do not use a Krylov method here and I don't believe "preonly" allocates any work vectors, and SuperLU does the symbolic factorizations without threads. Let me know how you want to proceed. Thanks, Mark ijcusparse -dm_vec_type cuda' NC=2 |g energy 0) species-0: charge density= -1.6022862392985e+01 z-momentum= -3.4369550192576e-19 energy= 9.6063873494138e+04 0) species-1: charge density= 1.6029950760009e+01 z-momentum= -2.7844197929124e-18 energy= 9.6333444502318e+04 0) Total: charge density= 7.0883670236874e-03, momentum= -3.1281152948382e-18, energy= 1.9239731799646e+05 (m_i[0]/m_e = 1835.47, 92 cells) ex2: /gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/externalpackages/git.superlu_dist/SRC/dSchCompUdt-cuda.c:157: pdgstrf: Assertion `jjj-1<nub' failed. [h16n13:21073] *** Process received signal *** [h16n13:21073] Signal: Aborted (6) [h16n13:21073] Signal code: User function (kill, sigsend, abort, etc.) (0) [h16n13:21073] [ 0] [0x2000000504d8] [h16n13:21073] [ 1] /lib64/libc.so.6(abort+0x2b4)[0x200020ef2094] [h16n13:21073] [ 2] /lib64/libc.so.6(+0x356d4)[0x200020ee56d4] [h16n13:21073] [ 3] /lib64/libc.so.6(__assert_fail+0x64)[0x200020ee57c4] [h16n13:21073] [ 4] /gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libsuperlu_dist.so.6(pdgstrf+0x3848)[0x2000022fe5d8] [h16n13:21073] [ 5] /gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libsuperlu_dist.so.6(pdgssvx+0x1220)[0x2000022dc4a8] [h16n13:21073] [ 6] /gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(+0x9aff28)[0x200000a9ff28] [h16n13:21073] [ 7] /gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(MatLUFactorNumeric+0x144)[0x2000007d273c] [h16n13:21073] [ 8] /gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(+0xecffc4)[0x200000fbffc4] [h16n13:21073] [ 9] /gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(PCSetUp+0x134)[0x20000107dd38] [h16n13:21073] [10] /gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(KSPSetUp+0x9f8)[0x2000010b272c] [h16n13:21073] [11] /gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(+0xfc46f0)[0x2000010b46f0] [h16n13:21073] [12] /gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(KSPSolve+0x20)[0x2000010b6fb8] [h16n13:21073] [13] /gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(+0xf1e4f8)[0x20000100e4f8] [h16n13:21073] [14] /sw/summit/gcc/6.4.0/lib64/libgomp.so.1(+0x1a51c)[0x200020e2a51c] [h16n13:21073] [15] /lib64/libpthread.so.0(+0x8b94)[0x200020e78b94] [h16n13:21073] [16] /lib64/libc.so.6(clone+0xe4)[0x200020fd85f4] [h16n13:21073] *** End of error message *** ERROR: One or more process (first noticed rank 0) terminated with signal 6 (core dumped) make: [runasm] Error 134 (ignored) On Thu, Jan 21, 2021 at 11:57 PM Xiaoye S. Li <[email protected]> wrote:
All the OpenMP calls are surrounded by
#ifdef _OPENMP ... #endif
You can disable openmp during Cmake installation, with the following: -Denable_openmp=FALSE (the default is true)
(I think Satish knows how to do this with PETSc installation)
------- The reason to use mixed MPI & OpenMP is mainly less memory consumption, compared to pure MPI. Timewise probably it is just slightly faster. (I think that's the case with many codes.)
Sherry
On Thu, Jan 21, 2021 at 7:20 PM Mark Adams <[email protected]> wrote:
On Thu, Jan 21, 2021 at 10:16 PM Barry Smith <[email protected]> wrote:
On Jan 21, 2021, at 9:11 PM, Mark Adams <[email protected]> wrote:
I have tried it and it hangs, but that is expected. This is not something she has prepared for.
I am working with Sherry on it.
And she is fine with just one thread and suggests it if she is in a thread.
Now that I think about it, I don't understand why she needs OpenMP if she can live with OMP_NUM_THREADS=1.
It is very possible it was just a coding decision by one of her students and with a few ifdef in her code should would not need the OpenMP but I don't have the time or energy to check her code and design decision.
Oh yea there OMP calls like omp_num_threads() that need something. There is probably a omp1.h file somewhere in the world like our serial MPI.
Barry
Mark
On Thu, Jan 21, 2021 at 9:30 PM Barry Smith <[email protected]> wrote:
On Jan 21, 2021, at 5:37 PM, Mark Adams <[email protected]> wrote:
This did not work. I verified that MPI_Init_thread is being called correctly and that MPI returns that it supports this highest level of thread safety.
I am going to ask ORNL.
And if I use:
-fieldsplit_i1_ksp_norm_type none -fieldsplit_i1_ksp_max_it 300
for all 9 "i" variables, I can run normal iterations on the 10th variable, in a 10 species problem, and it works perfectly with 10 threads.
So it is definitely that VecNorm is not thread safe.
And, I want to call SuperLU_dist, which uses threads, but I don't want SuperLU to start using threads. Is there a way to tell superLU that there are no threads but have PETSc use them?
My interpretation and Satish's for many years is that SuperLU_DIST has to be built with and use OpenMP in order to work with CUDA.
def formCMakeConfigureArgs(self): args = config.package.CMakePackage.formCMakeConfigureArgs(self) if self.openmp.found: self.usesopenmp = 'yes' else: args.append('-DCMAKE_DISABLE_FIND_PACKAGE_OpenMP=TRUE') if self.cuda.found: if not self.openmp.found: raise RuntimeError('SuperLU_DIST GPU code currently requires OpenMP. Use --with-openmp=1')
But this could be ok. You use OpenMP and then it uses OpenMP internally, each doing their own business (what could go wrong :-)).
Have you tried it?
Barry
Thanks, Mark
On Thu, Jan 21, 2021 at 5:19 PM Mark Adams <[email protected]> wrote:
OK, the problem is probably:
PetscMPIInt PETSC_MPI_THREAD_REQUIRED = MPI_THREAD_FUNNELED;
There is an example that sets:
PETSC_MPI_THREAD_REQUIRED = MPI_THREAD_MULTIPLE;
This is what I need.
On Thu, Jan 21, 2021 at 2:26 PM Mark Adams <[email protected]> wrote:
On Thu, Jan 21, 2021 at 2:11 PM Matthew Knepley <[email protected]> wrote:
On Thu, Jan 21, 2021 at 2:02 PM Mark Adams <[email protected]> wrote:
On Thu, Jan 21, 2021 at 1:44 PM Matthew Knepley <[email protected]> wrote:
On Thu, Jan 21, 2021 at 11:16 AM Mark Adams <[email protected]> wrote:
Yes, the problem is that each KSP solver is running in an OMP thread (So at this point it only works for SELF and its Landau so it is all I need). It looks like MPI reductions called with a comm_self are not thread safe (eg, the could say, this is one proc, thus, just copy send --> recv, but they don't)
Instead of using SELF, how about Comm_dup() for each thread?
OK, raw MPI_Comm_dup. I tried PetscCommDup. Let me this. Thanks,
You would have to dup them all outside the OMP section, since it is not threadsafe. Then each thread uses one I think.
Yea sure. I do it in SetUp.
Well that worked to get *different Comms*, finally, I still get the same problem. The number of iterations differ wildly. This two species and two threads (13 SNES its that is not deterministic). Way below is one thread (8 its) and fairly uniform iteration counts.
Maybe this MPI is just not thread safe at all. Let me look into it. Thanks anyway,
0 SNES Function norm 4.974994975313e-03 In PCFieldSplitSetFields_FieldSplit with -------------- link: 0x80017c60. Comms pc=0x67ad27c0 ksp=*0x7ffe1600* newcomm=0x8014b6e0 In PCFieldSplitSetFields_FieldSplit with -------------- link: 0x7ffdabc0. Comms pc=0x67ad27c0 ksp=*0x7fff70d0* newcomm=0x7ffe9980 Linear fieldsplit_e_ solve converged due to CONVERGED_RTOL iterations 282 1 SNES Function norm 1.836376279964e-05 Linear fieldsplit_e_ solve converged due to CONVERGED_ATOL iterations 19 2 SNES Function norm 3.059930074740e-07 Linear fieldsplit_e_ solve converged due to CONVERGED_ATOL iterations 15 3 SNES Function norm 4.744275398121e-08 Linear fieldsplit_e_ solve converged due to CONVERGED_ATOL iterations 4 4 SNES Function norm 4.014828563316e-08 Linear fieldsplit_e_ solve converged due to CONVERGED_RTOL iterations 456 5 SNES Function norm 5.670836337808e-09 Linear fieldsplit_e_ solve converged due to CONVERGED_ATOL iterations 2 6 SNES Function norm 2.410421401323e-09 Linear fieldsplit_e_ solve converged due to CONVERGED_ATOL iterations 18 7 SNES Function norm 6.533948191791e-10 Linear fieldsplit_e_ solve converged due to CONVERGED_RTOL iterations 458 8 SNES Function norm 1.008133815842e-10 Linear fieldsplit_e_ solve converged due to CONVERGED_ATOL iterations 9 9 SNES Function norm 1.690450876038e-11 Linear fieldsplit_e_ solve converged due to CONVERGED_ATOL iterations 4 10 SNES Function norm 1.336383986009e-11 Linear fieldsplit_e_ solve converged due to CONVERGED_RTOL iterations 463 11 SNES Function norm 1.873022410774e-12 Linear fieldsplit_e_ solve converged due to CONVERGED_RTOL iterations 113 12 SNES Function norm 1.801834606518e-13 Linear fieldsplit_e_ solve converged due to CONVERGED_ATOL iterations 1 13 SNES Function norm 1.004397317339e-13 Nonlinear solve converged due to CONVERGED_SNORM_RELATIVE iterations 13
0 SNES Function norm 4.974994975313e-03 In PCFieldSplitSetFields_FieldSplit with -------------- link: 0x6e265010. Comms pc=0x56450340 ksp=0x6e2168d0 newcomm=0x6e265090 In PCFieldSplitSetFields_FieldSplit with -------------- link: 0x6e25bc40. Comms pc=0x56450340 ksp=0x6e22c1d0 newcomm=0x6e21e8f0 Linear fieldsplit_e_ solve converged due to CONVERGED_RTOL iterations 282 1 SNES Function norm 1.836376279963e-05 Linear fieldsplit_e_ solve converged due to CONVERGED_RTOL iterations 380 2 SNES Function norm 3.018499983019e-07 Linear fieldsplit_e_ solve converged due to CONVERGED_RTOL iterations 387 3 SNES Function norm 1.826353175637e-08 Linear fieldsplit_e_ solve converged due to CONVERGED_RTOL iterations 391 4 SNES Function norm 1.378600599548e-09 Linear fieldsplit_e_ solve converged due to CONVERGED_RTOL iterations 392 5 SNES Function norm 1.077289085611e-10 Linear fieldsplit_e_ solve converged due to CONVERGED_RTOL iterations 394 6 SNES Function norm 8.571891727748e-12 Linear fieldsplit_e_ solve converged due to CONVERGED_RTOL iterations 395 7 SNES Function norm 6.897647643450e-13 Linear fieldsplit_e_ solve converged due to CONVERGED_RTOL iterations 395 8 SNES Function norm 5.606434614114e-14 Nonlinear solve converged due to CONVERGED_SNORM_RELATIVE iterations 8
Matt
Matt
On Thu, Jan 21, 2021 at 10:46 AM Matthew Knepley < [email protected]> wrote:
On Thu, Jan 21, 2021 at 10:34 AM Mark Adams <[email protected]> wrote:
It looks like PETSc is just too clever for me. I am trying to get a different MPI_Comm into each block, but PETSc is thwarting me:
It looks like you are using SELF. Is that what you want? Do you want a bunch of comms with the same group, but independent somehow? I am confused.
Matt
if (jac->use_openmp) { ierr = KSPCreate(MPI_COMM_SELF,&ilink->ksp);CHKERRQ(ierr); PetscPrintf(PETSC_COMM_SELF,"In PCFieldSplitSetFields_FieldSplit with -------------- link: %p. Comms %p %p\n",ilink,PetscObjectComm((PetscObject)pc),PetscObjectComm((PetscObject)ilink->ksp)); } else { ierr = KSPCreate(PetscObjectComm((PetscObject)pc),&ilink->ksp);CHKERRQ(ierr); }
produces:
In PCFieldSplitSetFields_FieldSplit with -------------- link: 0x7e9cb4f0. Comms 0x660c6ad0 0x660c6ad0 In PCFieldSplitSetFields_FieldSplit with -------------- link: 0x7e88f7d0. Comms 0x660c6ad0 0x660c6ad0
How can I work around this?
On Thu, Jan 21, 2021 at 7:41 AM Mark Adams <[email protected]> wrote:
On Wed, Jan 20, 2021 at 6:21 PM Barry Smith <[email protected]> wrote:
On Jan 20, 2021, at 3:09 PM, Mark Adams <[email protected]> wrote:
So I put in a temporary hack to get the first Fieldsplit apply to NOT use OMP and it sort of works.
Preonly/lu is fine. GMRES calls vector creates/dups in every solve so that is a big problem.
It should definitely not be creating vectors "in every" solve. But it does do lazy allocation of needed restarted vectors which may make it look like it is creating "every" vectors in every solve. You can use -ksp_gmres_preallocate to force it to create all the restart vectors up front at KSPSetUp().
Well, I run the first solve w/o OMP and I see Vec dups in cuSparse Vecs in the 2nd solve.
Why is creating vectors "at every solve" a problem? It is not thread safe I guess?
It dies when it looks at the options database, in a Free in the get-options method to be exact (see stacks).
======= Backtrace: ========= /lib64/libc.so.6(cfree+0x4a0)[0x200021839be0]
/gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(PetscFreeAlign+0x4c)[0x2000002a368c]
/gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(PetscOptionsEnd_Private+0xf4)[0x2000002e53f0]
/gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(+0x7c6c28)[0x2000008b6c28]
/gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(VecCreate_SeqCUDA+0x11c)[0x20000052c510]
/gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(VecSetType+0x670)[0x200000549664]
/gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(VecCreateSeqCUDA+0x150)[0x20000052c0b0]
/gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(+0x43c198)[0x20000052c198]
/gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(VecDuplicate+0x44)[0x200000542168]
/gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(VecDuplicateVecs_Default+0x148)[0x200000543820]
/gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(VecDuplicateVecs+0x54)[0x2000005425f4]
/gpfs/alpine/csc314/scratch/adams/petsc/arch-summit-opt-gnu-cuda-omp/lib/libpetsc.so.3.014(KSPCreateVecs+0x4b4)[0x2000016f0aec]
Richardson works except the convergence test gets confused, presumably because MPI reductions with PETSC_COMM_SELF is not threadsafe.
One fix for the norms might be to create each subdomain solver with a different communicator.
Yes you could do that. It might actually be the correct thing to do also, if you have multiple threads call MPI reductions on the same communicator that would be a problem. Each KSP should get a new MPI_Comm.
OK. I will only do this.
-- 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
https://www.cse.buffalo.edu/~knepley/ <http://www.cse.buffalo.edu/~knepley/>
-- 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
https://www.cse.buffalo.edu/~knepley/ <http://www.cse.buffalo.edu/~knepley/>
-- 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
https://www.cse.buffalo.edu/~knepley/ <http://www.cse.buffalo.edu/~knepley/>