Thanks, Jed On Thu, Jan 20, 2022 at 4:34 PM Jed Brown <j...@jedbrown.org> wrote:
> You can't create CUDA or Kokkos Vecs if you're running on a node without a > GPU. I am running the code on compute nodes that do have GPUs. With PETSc-3.16.1, I got good speedup by running GAMG on GPUs. That might be a bug of PETSc-main. Thanks, Fande KSPSetUp 13 1.0 6.4400e-01 1.0 2.02e+09 1.0 0.0e+00 0.0e+00 0.0e+00 0 5 0 0 0 0 5 0 0 0 3140 64630 15 1.05e+02 5 3.49e+01 100 KSPSolve 1 1.0 1.0109e+00 1.0 3.49e+10 1.0 0.0e+00 0.0e+00 0.0e+00 0 87 0 0 0 0 87 0 0 0 34522 69556 4 4.35e-03 1 2.38e-03 100 KSPGMRESOrthog 142 1.0 1.2674e-01 1.0 1.06e+10 1.0 0.0e+00 0.0e+00 0.0e+00 0 27 0 0 0 0 27 0 0 0 83755 87801 0 0.00e+00 0 0.00e+00 100 SNESSolve 1 1.0 4.4402e+01 1.0 4.00e+10 1.0 0.0e+00 0.0e+00 0.0e+00 21100 0 0 0 21100 0 0 0 901 51365 57 1.10e+03 52 8.78e+02 100 SNESSetUp 1 1.0 3.9101e-05 1.0 0.00e+00 0.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 0 0 0 0.00e+00 0 0.00e+00 0 SNESFunctionEval 2 1.0 1.7097e+01 1.0 1.60e+07 1.0 0.0e+00 0.0e+00 0.0e+00 8 0 0 0 0 8 0 0 0 0 1 0 0 0.00e+00 6 1.92e+02 0 SNESJacobianEval 1 1.0 1.6213e+01 1.0 2.80e+07 1.0 0.0e+00 0.0e+00 0.0e+00 8 0 0 0 0 8 0 0 0 0 2 0 0 0.00e+00 1 3.20e+01 0 SNESLineSearch 1 1.0 8.5582e+00 1.0 1.24e+08 1.0 0.0e+00 0.0e+00 0.0e+00 4 0 0 0 0 4 0 0 0 0 14 64153 1 3.20e+01 3 9.61e+01 94 PCGAMGGraph_AGG 5 1.0 3.0509e+00 1.0 8.19e+07 1.0 0.0e+00 0.0e+00 0.0e+00 1 0 0 0 0 1 0 0 0 0 27 0 5 3.49e+01 9 7.43e+01 0 PCGAMGCoarse_AGG 5 1.0 3.8711e+00 1.0 0.00e+00 0.0 0.0e+00 0.0e+00 0.0e+00 2 0 0 0 0 2 0 0 0 0 0 0 0 0.00e+00 0 0.00e+00 0 PCGAMGProl_AGG 5 1.0 7.0748e-01 1.0 0.00e+00 0.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 0 0 0 0.00e+00 0 0.00e+00 0 PCGAMGPOpt_AGG 5 1.0 1.2904e+00 1.0 2.14e+09 1.0 0.0e+00 0.0e+00 0.0e+00 1 5 0 0 0 1 5 0 0 0 1661 29807 26 7.15e+02 20 2.90e+02 99 GAMG: createProl 5 1.0 8.9489e+00 1.0 2.22e+09 1.0 0.0e+00 0.0e+00 0.0e+00 4 6 0 0 0 4 6 0 0 0 249 29666 31 7.50e+02 29 3.64e+02 96 Graph 10 1.0 3.0478e+00 1.0 8.19e+07 1.0 0.0e+00 0.0e+00 0.0e+00 1 0 0 0 0 1 0 0 0 0 27 0 5 3.49e+01 9 7.43e+01 0 MIS/Agg 5 1.0 4.1290e-01 1.0 0.00e+00 0.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 0 0 0 0.00e+00 0 0.00e+00 0 SA: col data 5 1.0 1.9127e-02 1.0 0.00e+00 0.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 0 0 0 0.00e+00 0 0.00e+00 0 SA: frmProl0 5 1.0 6.2662e-01 1.0 0.00e+00 0.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 0 0 0 0.00e+00 0 0.00e+00 0 SA: smooth 5 1.0 4.9595e-01 1.0 1.21e+08 1.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 244 2709 15 1.97e+02 15 2.55e+02 90 GAMG: partLevel 5 1.0 4.7330e-01 1.0 6.98e+08 1.0 0.0e+00 0.0e+00 0.0e+00 0 2 0 0 0 0 2 0 0 0 1475 4120 5 1.78e+02 10 2.55e+02 100 PCGAMG Squ l00 1 1.0 2.6027e+00 1.0 0.00e+00 0.0 0.0e+00 0.0e+00 0.0e+00 1 0 0 0 0 1 0 0 0 0 0 0 0 0.00e+00 0 0.00e+00 0 PCGAMG Gal l00 1 1.0 3.8406e-01 1.0 5.48e+08 1.0 0.0e+00 0.0e+00 0.0e+00 0 1 0 0 0 0 1 0 0 0 1426 4270 1 1.48e+02 2 2.11e+02 100 PCGAMG Opt l00 1 1.0 2.4932e-01 1.0 7.20e+07 1.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 289 2653 1 6.41e+01 1 1.13e+02 100 PCGAMG Gal l01 1 1.0 6.6279e-02 1.0 1.09e+08 1.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 1645 3851 1 2.40e+01 2 3.64e+01 100 PCGAMG Opt l01 1 1.0 2.9544e-02 1.0 7.15e+06 1.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 242 1671 1 4.84e+00 1 1.23e+01 100 PCGAMG Gal l02 1 1.0 1.8874e-02 1.0 3.72e+07 1.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 1974 3636 1 5.04e+00 2 6.58e+00 100 PCGAMG Opt l02 1 1.0 7.4353e-03 1.0 2.40e+06 1.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 323 1457 1 7.71e-01 1 2.30e+00 100 PCGAMG Gal l03 1 1.0 2.8479e-03 1.0 4.10e+06 1.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 1440 2266 1 4.44e-01 2 5.51e-01 100 PCGAMG Opt l03 1 1.0 8.2684e-04 1.0 2.80e+05 1.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 339 1667 1 6.72e-02 1 2.03e-01 100 PCGAMG Gal l04 1 1.0 1.2238e-03 1.0 2.09e+05 1.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 170 244 1 2.05e-02 2 2.53e-02 100 PCGAMG Opt l04 1 1.0 4.1008e-04 1.0 1.77e+04 1.0 0.0e+00 0.0e+00 0.0e+00 0 0 0 0 0 0 0 0 0 0 43 165 1 4.49e-03 1 1.19e-02 100 PCSetUp 2 1.0 9.9632e+00 1.0 4.95e+09 1.0 0.0e+00 0.0e+00 0.0e+00 5 12 0 0 0 5 12 0 0 0 496 17826 55 1.03e+03 45 6.54e+02 98 PCSetUpOnBlocks 44 1.0 9.9087e-04 1.0 2.88e+03 1.0 > The point of lazy initialization is to make it possible to run a solve > that doesn't use a GPU in PETSC_ARCH that supports GPUs, regardless of > whether a GPU is actually present. > > Fande Kong <fdkong...@gmail.com> writes: > > > I spoke too soon. It seems that we have trouble creating cuda/kokkos vecs > > now. Got Segmentation fault. > > > > Thanks, > > > > Fande > > > > Program received signal SIGSEGV, Segmentation fault. > > 0x00002aaab5558b11 in > > > Petsc::CUPMDevice<(Petsc::CUPMDeviceType)0>::CUPMDeviceInternal::initialize > > (this=0x1) at > > > /home/kongf/workhome/sawtooth/moosegpu/petsc/src/sys/objects/device/impls/cupm/cupmdevice.cxx:54 > > 54 PetscErrorCode CUPMDevice<T>::CUPMDeviceInternal::initialize() > noexcept > > Missing separate debuginfos, use: debuginfo-install > > bzip2-libs-1.0.6-13.el7.x86_64 elfutils-libelf-0.176-5.el7.x86_64 > > elfutils-libs-0.176-5.el7.x86_64 glibc-2.17-325.el7_9.x86_64 > > libX11-1.6.7-4.el7_9.x86_64 libXau-1.0.8-2.1.el7.x86_64 > > libattr-2.4.46-13.el7.x86_64 libcap-2.22-11.el7.x86_64 > > libibmad-5.4.0.MLNX20190423.1d917ae-0.1.49224.x86_64 > > libibumad-43.1.1.MLNX20200211.078947f-0.1.49224.x86_64 > > libibverbs-41mlnx1-OFED.4.9.0.0.7.49224.x86_64 > > libmlx4-41mlnx1-OFED.4.7.3.0.3.49224.x86_64 > > libmlx5-41mlnx1-OFED.4.9.0.1.2.49224.x86_64 libnl3-3.2.28-4.el7.x86_64 > > librdmacm-41mlnx1-OFED.4.7.3.0.6.49224.x86_64 > > librxe-41mlnx1-OFED.4.4.2.4.6.49224.x86_64 libxcb-1.13-1.el7.x86_64 > > libxml2-2.9.1-6.el7_9.6.x86_64 numactl-libs-2.0.12-5.el7.x86_64 > > systemd-libs-219-78.el7_9.3.x86_64 xz-libs-5.2.2-1.el7.x86_64 > > zlib-1.2.7-19.el7_9.x86_64 > > (gdb) bt > > #0 0x00002aaab5558b11 in > > > Petsc::CUPMDevice<(Petsc::CUPMDeviceType)0>::CUPMDeviceInternal::initialize > > (this=0x1) at > > > /home/kongf/workhome/sawtooth/moosegpu/petsc/src/sys/objects/device/impls/cupm/cupmdevice.cxx:54 > > #1 0x00002aaab5558db7 in > > Petsc::CUPMDevice<(Petsc::CUPMDeviceType)0>::getDevice > > (this=this@entry=0x2aaab7f37b70 > > <CUDADevice>, device=0x115da00, id=-35, id@entry=-1) at > > > /home/kongf/workhome/sawtooth/moosegpu/petsc/src/sys/objects/device/impls/cupm/cupmdevice.cxx:344 > > #2 0x00002aaab55577de in PetscDeviceCreate (type=type@entry > =PETSC_DEVICE_CUDA, > > devid=devid@entry=-1, device=device@entry=0x2aaab7f37b48 > > <defaultDevices+8>) at > > > /home/kongf/workhome/sawtooth/moosegpu/petsc/src/sys/objects/device/interface/device.cxx:107 > > #3 0x00002aaab5557b3a in PetscDeviceInitializeDefaultDevice_Internal > > (type=type@entry=PETSC_DEVICE_CUDA, > defaultDeviceId=defaultDeviceId@entry=-1) > > at > > > /home/kongf/workhome/sawtooth/moosegpu/petsc/src/sys/objects/device/interface/device.cxx:273 > > #4 0x00002aaab5557bf6 in PetscDeviceInitialize > > (type=type@entry=PETSC_DEVICE_CUDA) > > at > > > /home/kongf/workhome/sawtooth/moosegpu/petsc/src/sys/objects/device/interface/device.cxx:234 > > #5 0x00002aaab5661fcd in VecCreate_SeqCUDA (V=0x115d150) at > > > /home/kongf/workhome/sawtooth/moosegpu/petsc/src/vec/vec/impls/seq/seqcuda/veccuda.c:244 > > #6 0x00002aaab5649b40 in VecSetType (vec=vec@entry=0x115d150, > > method=method@entry=0x2aaab70b45b8 "seqcuda") at > > > /home/kongf/workhome/sawtooth/moosegpu/petsc/src/vec/vec/interface/vecreg.c:93 > > #7 0x00002aaab579c33f in VecCreate_CUDA (v=0x115d150) at > > > /home/kongf/workhome/sawtooth/moosegpu/petsc/src/vec/vec/impls/mpi/mpicuda/ > > mpicuda.cu:214 > > #8 0x00002aaab5649b40 in VecSetType (vec=vec@entry=0x115d150, > > method=method@entry=0x7fffffff9260 "cuda") at > > > /home/kongf/workhome/sawtooth/moosegpu/petsc/src/vec/vec/interface/vecreg.c:93 > > #9 0x00002aaab5648bf1 in VecSetTypeFromOptions_Private (vec=0x115d150, > > PetscOptionsObject=0x7fffffff9210) at > > > /home/kongf/workhome/sawtooth/moosegpu/petsc/src/vec/vec/interface/vector.c:1263 > > #10 VecSetFromOptions (vec=0x115d150) at > > > /home/kongf/workhome/sawtooth/moosegpu/petsc/src/vec/vec/interface/vector.c:1297 > > #11 0x00002aaab02ef227 in libMesh::PetscVector<double>::init > > (this=0x11cd1a0, n=441, n_local=441, fast=false, ptype=libMesh::PARALLEL) > > at > > > /home/kongf/workhome/sawtooth/moosegpu/scripts/../libmesh/installed/include/libmesh/petsc_vector.h:693 > > > > On Thu, Jan 20, 2022 at 1:09 PM Fande Kong <fdkong...@gmail.com> wrote: > > > >> Thanks, Jed, > >> > >> This worked! > >> > >> Fande > >> > >> On Wed, Jan 19, 2022 at 11:03 PM Jed Brown <j...@jedbrown.org> wrote: > >> > >>> Fande Kong <fdkong...@gmail.com> writes: > >>> > >>> > On Wed, Jan 19, 2022 at 11:39 AM Jacob Faibussowitsch < > >>> jacob....@gmail.com> > >>> > wrote: > >>> > > >>> >> Are you running on login nodes or compute nodes (I can’t seem to > tell > >>> from > >>> >> the configure.log)? > >>> >> > >>> > > >>> > I was compiling codes on login nodes, and running codes on compute > >>> nodes. > >>> > Login nodes do not have GPUs, but compute nodes do have GPUs. > >>> > > >>> > Just to be clear, the same thing (code, machine) with PETSc-3.16.1 > >>> worked > >>> > perfectly. I have this trouble with PETSc-main. > >>> > >>> I assume you can > >>> > >>> export PETSC_OPTIONS='-device_enable lazy' > >>> > >>> and it'll work. > >>> > >>> I think this should be the default. The main complaint is that timing > the > >>> first GPU-using event isn't accurate if it includes initialization, > but I > >>> think this is mostly hypothetical because you can't trust any timing > that > >>> doesn't preload in some form and the first GPU-using event will almost > >>> always be something uninteresting so I think it will rarely lead to > >>> confusion. Meanwhile, eager initialization is viscerally disruptive for > >>> lots of people. > >>> > >> >
