Remove --with-kokkos-cuda-arch=VOLTA70 from your configure Barry: It seems we should either disable this option, or rename it to --with-cuda-arch and get it accepted by both kokkos.py and cuda.py --Junchao Zhang On Sun, May 16, 2021 at 6:15 AM Mark Adams <[email protected]> wrote:
progress:
04:08 cgpu11 adams/cusparse-cpu-solve *= ~/petsc$ make PETSC_DIR=/global/homes/m/madams/petsc PETSC_ARCH=arch-cori-gpu-opt-kokkos-gcc check Running check examples to verify correct installation Using PETSC_DIR=/global/homes/m/madams/petsc and PETSC_ARCH=arch-cori-gpu-opt-kokkos-gcc C/C++ example src/snes/tutorials/ex19 run successfully with 1 MPI process Possible error running C/C++ src/snes/tutorials/ex19 with 2 MPI processes See http://www.mcs.anl.gov/petsc/documentation/faq.html srun: error: Unable to create step for job 1921749: More processors requested than permitted 2,15c2,32 < 0 SNES Function norm 2.391552133017e-01 < 0 KSP Residual norm 2.325621076120e-01 < 1 KSP Residual norm 1.654206318674e-02 < 2 KSP Residual norm 7.202836119880e-04 < 3 KSP Residual norm 1.796861424199e-05 < 4 KSP Residual norm 2.461332992052e-07 < 1 SNES Function norm 6.826585648929e-05 < 0 KSP Residual norm 2.347339172985e-05 < 1 KSP Residual norm 8.356798075993e-07 < 2 KSP Residual norm 1.844045309619e-08 < 3 KSP Residual norm 5.336386977405e-10 < 4 KSP Residual norm 2.662608472862e-11 < 2 SNES Function norm 6.549682264799e-11 < Number of SNES iterations = 2 ---
Kokkos::Cuda::initialize ERROR: likely mismatch of architecture [cgpu11:39927] *** Process received signal *** [cgpu11:39927] Signal: Aborted (6) [cgpu11:39927] Signal code: (-6) [cgpu11:39927] [ 0] /lib64/libpthread.so.0(+0x12360)[0x2aab12731360] [cgpu11:39927] [ 1] /lib64/libc.so.6(gsignal+0x110)[0x2aab12973160] [cgpu11:39927] [ 2] /lib64/libc.so.6(abort+0x151)[0x2aab12974741] [cgpu11:39927] [ 3] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libkokkoscore.so.3.4(+0x408e5)[0x2aaabbfb88e5] [cgpu11:39927] [ 4] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libkokkoscore.so.3.4(_ZN6Kokkos4Impl12CudaInternal10initializeEiP11CUstream_st+0xb5c)[0x2aaabbfc762c] [cgpu11:39927] [ 5] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libkokkoscore.so.3.4(_ZN6Kokkos4Impl16ExecSpaceManager17initialize_spacesERKNS_13InitArgumentsE+0x24)[0x2aaabbfa1224] [cgpu11:39927] [ 6] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libkokkoscore.so.3.4(_ZN6Kokkos4Impl70_GLOBAL__N__46_tmpxft_00011209_00000000_7_Kokkos_Core_cpp1_ii_889c95a619initialize_internalERKNS_13InitArgumentsE+0x11)[0x2aaabbfa49d1] [cgpu11:39927] [ 7] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libpetsc.so.3.015(PetscKokkosInitializeCheck+0x1f9)[0x2aaaaafad809] [cgpu11:39927] [ 8] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libpetsc.so.3.015(+0x68541b)[0x2aaaab35841b] [cgpu11:39927] [ 9] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libpetsc.so.3.015(+0x36a224)[0x2aaaab03d224] [cgpu11:39927] [10] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libpetsc.so.3.015(+0x35f596)[0x2aaaab032596] [cgpu11:39927] [11] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libpetsc.so.3.015(+0x2f9d94)[0x2aaaaafccd94] [cgpu11:39927] [12] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libpetsc.so.3.015(PetscSFBcastWithMemTypeBegin+0x11f)[0x2aaaab04bc3f] [cgpu11:39927] [13] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libpetsc.so.3.015(+0x37f53a)[0x2aaaab05253a] [cgpu11:39927] [14] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libpetsc.so.3.015(VecScatterBegin+0x70)[0x2aaaab057b30] [cgpu11:39927] [15] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libpetsc.so.3.015(DMGlobalToLocalBegin_DA+0x1f)[0x2aaaabc27a5f] [cgpu11:39927] [16] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libpetsc.so.3.015(DMGlobalToLocalBegin+0x1dd)[0x2aaaabe61b3d] [cgpu11:39927] [17] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libpetsc.so.3.015(+0x144567a)[0x2aaaac11867a] [cgpu11:39927] [18] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libpetsc.so.3.015(SNESComputeFunction+0xf5)[0x2aaaac138ef5] [cgpu11:39927] [19] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libpetsc.so.3.015(+0x14ad0de)[0x2aaaac1800de] [cgpu11:39927] [20] /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/lib/libpetsc.so.3.015(SNESSolve+0x821)[0x2aaaac146ed1] [cgpu11:39927] [21] /global/u2/m/madams/petsc/src/snes/tutorials/./ex19[0x401a69] [cgpu11:39927] [22] /lib64/libc.so.6(__libc_start_main+0xea)[0x2aab1295df8a] [cgpu11:39927] [23] /global/u2/m/madams/petsc/src/snes/tutorials/./ex19[0x4021ba] [cgpu11:39927] *** End of error message *** srun: error: cgpu11: task 0: Aborted srun: Terminating job step 1921749.6 /global/homes/m/madams/petsc/src/snes/tutorials Possible problem with ex19 running with cuda, diffs above ========================================= gmake[3]: [makefile:102: runex3k_kokkos] Error 1 (ignored)
On Sun, May 16, 2021 at 12:31 AM Barry Smith <[email protected]> wrote:
Looks like nvcc_wrapper gets confused when both -arch and -gencode are used. But handles either of them separately correctly.
Mark,
PETSc configure now handles the -arch and --with-kokkos-cuda-arch= business automatically so you do not, and generally shouldn't pass -arch sm_70 in --CUDAFLAGS anymore or use the --with-kokkos-cuda-arch option anymore.
If you need to set the -arch to something earlier than the system supports (rare) you can now use -with-cuda-gencodearch to set that instead of using CUDAFLAGS.
Barry
$ NVCC_WRAPPER_DEFAULT_COMPILER=mpicxx /home/bsmith/petsc/arch-main/bin/nvcc_wrapper --show --expt-extended-lambda -c -Xcompiler -fPIC -g -gencode arch=compute_70,code=sm_70 -I/home/bsmith/soft/gnu-mpich/include -Wno-deprecated-gpu-targets -I/home/bsmith/petsc/include -I/home/bsmith/petsc/arch-main/include -I/nfs/apps/spacks/2021-02-09/opt/spack/linux-centos7-x86_64/gcc-7.3.0/cuda-11.2.0-ikruu5mo6dtt3avvmwsejouqhxu4btdm/include /home/bsmith/petsc/src/ts/utils/dmplexlandau/kokkos/landau.kokkos.cxx -o arch-main/obj/ts/utils/dmplexlandau/kokkos/landau.o nvcc --expt-extended-lambda -gencode arch=compute_70,code=sm_70 -Wno-deprecated-gpu-targets -ccbin mpicxx -g -I/home/bsmith/soft/gnu-mpich/include -I/home/bsmith/petsc/include -I/home/bsmith/petsc/arch-main/include -I/nfs/apps/spacks/2021-02-09/opt/spack/linux-centos7-x86_64/gcc-7.3.0/cuda-11.2.0-ikruu5mo6dtt3avvmwsejouqhxu4btdm/include -Xcompiler -fPIC -x cu /home/bsmith/petsc/src/ts/utils/dmplexlandau/kokkos/landau.kokkos.cxx -c -o arch-main/obj/ts/utils/dmplexlandau/kokkos/landau.o ~/petsc* (main=)* arch-main
$ NVCC_WRAPPER_DEFAULT_COMPILER=mpicxx /home/bsmith/petsc/arch-main/bin/nvcc_wrapper --show --expt-extended-lambda -c -arch=sm_70 -Xcompiler -rdynamic -DLANDAU_DIM=2 -DLANDAU_MAX_SPECIES=10 -DPETSC_HAVE_CUDA_ATOMIC -DLANDAU_MAX_Q=4 -Xcompiler -fPIC -O3 -gencode arch=compute_80,code=sm_80 -I/usr/common/software/sles15_cgpu/openmpi/4.0.3/gcc/include -Wno-deprecated-gpu-targets nvcc --expt-extended-lambda -arch=sm_70 -Wno-deprecated-gpu-targets -ccbin mpicxx -DLANDAU_DIM=2 -DLANDAU_MAX_SPECIES=10 -DPETSC_HAVE_CUDA_ATOMIC -DLANDAU_MAX_Q=4 -O3 -I/usr/common/software/sles15_cgpu/openmpi/4.0.3/gcc/include -Xcompiler -rdynamic,-fPIC,arch=compute_80,code=sm_80 -c ~/petsc* (main=)* arch-main
$ NVCC_WRAPPER_DEFAULT_COMPILER=mpicxx /home/bsmith/petsc/arch-main/bin/nvcc_wrapper --show --expt-extended-lambda -c -Xcompiler -fPIC -g -arch=sm_70 -gencode arch=compute_70,code=sm_70 -I/home/bsmith/soft/gnu-mpich/include -Wno-deprecated-gpu-targets -I/home/bsmith/petsc/include -I/home/bsmith/petsc/arch-main/include -I/nfs/apps/spacks/2021-02-09/opt/spack/linux-centos7-x86_64/gcc-7.3.0/cuda-11.2.0-ikruu5mo6dtt3avvmwsejouqhxu4btdm/include /home/bsmith/petsc/src/ts/utils/dmplexlandau/kokkos/landau.kokkos.cxx -o arch-main/obj/ts/utils/dmplexlandau/kokkos/landau.o nvcc --expt-extended-lambda -arch=sm_70 -Wno-deprecated-gpu-targets -ccbin mpicxx -g -I/home/bsmith/soft/gnu-mpich/include -I/home/bsmith/petsc/include -I/home/bsmith/petsc/arch-main/include -I/nfs/apps/spacks/2021-02-09/opt/spack/linux-centos7-x86_64/gcc-7.3.0/cuda-11.2.0-ikruu5mo6dtt3avvmwsejouqhxu4btdm/include -Xcompiler -fPIC,arch=compute_70,code=sm_70 -x cu /home/bsmith/petsc/src/ts/utils/dmplexlandau/kokkos/landau.kokkos.cxx -c -o arch-main/obj/ts/utils/dmplexlandau/kokkos/landau.o
On May 15, 2021, at 8:44 PM, Junchao Zhang <[email protected]> wrote:
Add --show to the failing nvcc_wrapper command and see how nvcc_wrapper parses the options
PATH=/global/common/sw/cray/sles15/x86_64/zlib/1.2.11/gcc/8.2.0/pep2pal/bin:/global/common/sw/cray/cnl7/haswell/cmake/3.18.2/bin:/opt/cray/rca/2.2.20-7.0.1.1_4.62__g8e3fb5b.ari/bin:/opt/cray/alps/6.6.58-7.0.1.1_6.20__g437d88db.ari/sbin:/opt/cray/alps/default/bin:/opt/cray/job/2.2.4-7.0.1.1_3.48__g36b56f4.ari/bin:/opt/cray/pe/craype/2.6.2/bin:/usr/common/software/sles15_cgpu/openmpi/4.0.3/gcc/bin:/usr/common/software/sles15_cgpu/hpcsdk/20.11/Linux_x86_64/20.11/compilers/bin:/usr/common/software/sles15_cgpu/hpcsdk/20.11/Linux_x86_64/20.11/cuda/bin:/usr/common/software/sles15_cgpu/cuda/11.1.1/bin:/usr/common/software/sles15_cgpu/gcc/8.3.0/bin:/opt/esslurm/bin:/opt/cray/pe/modules/ 3.2.11.4/bin:/usr/common/software/bin:/usr/common/nsg/bin:/opt/ovis/bin:/opt/ovis/sbin:/usr/local/bin:/usr/bin:/bin:/usr/lib/mit/bin:/usr/lib/mit/sbin:/opt/cray/pe/bin:`dirname <http://3.2.11.4/bin:/usr/common/software/bin:/usr/common/nsg/bin:/opt/ovis/bin:/opt/ovis/sbin:/usr/local/bin:/usr/bin:/bin:/usr/lib/mit/bin:/usr/lib/mit/sbin:/opt/cray/pe/bin:dirname> /usr/common/software/sles15_cgpu/cuda/11.1.1/bin/nvcc` NVCC_WRAPPER_DEFAULT_COMPILER=/usr/common/software/sles15_cgpu/openmpi/4.0.3/gcc/bin/mpicxx /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/bin/nvcc_wrapper --expt-extended-lambda -c -arch=sm_70 -Xcompiler -rdynamic -DLANDAU_DIM=2 -DLANDAU_MAX_SPECIES=10 -DPETSC_HAVE_CUDA_ATOMIC -DLANDAU_MAX_Q=4 -Xcompiler -fPIC -O3 -gencode arch=compute_80,code=sm_80 -I/usr/common/software/sles15_cgpu/openmpi/4.0.3/gcc/include -Wno-deprecated-gpu-targets -I/global/homes/m/madams/petsc/include -I/global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/include -I/usr/common/software/sles15_cgpu/openmpi/4.0.3/gcc/include -I/usr/common/software/sles15_cgpu/cuda/11.1.1/include /global/u2/m/madams/petsc/src/sys/objects/kokkos/kinit.kokkos.cxx -o arch-cori-gpu-opt-kokkos-gcc/obj/sys/objects/kokkos/kinit.o --show
--Junchao Zhang
On Sat, May 15, 2021 at 7:14 PM Mark Adams <[email protected]> wrote:
Thanks, Now I get an error in make all. THis was working a few weeks ago. Make.log was empty but here is the output. .... CC arch-cori-gpu-opt-kokkos-gcc/obj/sys/error/ftn-custom/zerrf.o CC arch-cori-gpu-opt-kokkos-gcc/obj/sys/f90-src/f90_cwrap.o CC arch-cori-gpu-opt-kokkos-gcc/obj/vec/is/utils/hdf5io.o
PATH=/global/common/sw/cray/sles15/x86_64/zlib/1.2.11/gcc/8.2.0/pep2pal/bin:/global/common/sw/cray/cnl7/haswell/cmake/3.18.2/bin:/opt/cray/rca/2.2.20-7.0.1.1_4.62__g8e3fb5b.ari/bin:/opt/cray/alps/6.6.58-7.0.1.1_6.20__g437d88db.ari/sbin:/opt/cray/alps/default/bin:/opt/cray/job/2.2.4-7.0.1.1_3.48__g36b56f4.ari/bin:/opt/cray/pe/craype/2.6.2/bin:/usr/common/software/sles15_cgpu/openmpi/4.0.3/gcc/bin:/usr/common/software/sles15_cgpu/hpcsdk/20.11/Linux_x86_64/20.11/compilers/bin:/usr/common/software/sles15_cgpu/hpcsdk/20.11/Linux_x86_64/20.11/cuda/bin:/usr/common/software/sles15_cgpu/cuda/11.1.1/bin:/usr/common/software/sles15_cgpu/gcc/8.3.0/bin:/opt/esslurm/bin:/opt/cray/pe/modules/ 3.2.11.4/bin:/usr/common/software/bin:/usr/common/nsg/bin:/opt/ovis/bin:/opt/ovis/sbin:/usr/local/bin:/usr/bin:/bin:/usr/lib/mit/bin:/usr/lib/mit/sbin:/opt/cray/pe/bin:`dirname <http://3.2.11.4/bin:/usr/common/software/bin:/usr/common/nsg/bin:/opt/ovis/bin:/opt/ovis/sbin:/usr/local/bin:/usr/bin:/bin:/usr/lib/mit/bin:/usr/lib/mit/sbin:/opt/cray/pe/bin:dirname> /usr/common/software/sles15_cgpu/cuda/11.1.1/bin/nvcc` NVCC_WRAPPER_DEFAULT_COMPILER=/usr/common/software/sles15_cgpu/openmpi/4.0.3/gcc/bin/mpicxx /global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/bin/nvcc_wrapper --expt-extended-lambda -c -arch=sm_70 -Xcompiler -rdynamic -DLANDAU_DIM=2 -DLANDAU_MAX_SPECIES=10 -DPETSC_HAVE_CUDA_ATOMIC -DLANDAU_MAX_Q=4 -Xcompiler -fPIC -O3 -gencode arch=compute_80,code=sm_80 -I/usr/common/software/sles15_cgpu/openmpi/4.0.3/gcc/include -Wno-deprecated-gpu-targets -I/global/homes/m/madams/petsc/include -I/global/homes/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/include -I/usr/common/software/sles15_cgpu/openmpi/4.0.3/gcc/include -I/usr/common/software/sles15_cgpu/cuda/11.1.1/include /global/u2/m/madams/petsc/src/sys/objects/kokkos/kinit.kokkos.cxx -o arch-cori-gpu-opt-kokkos-gcc/obj/sys/objects/kokkos/kinit.o g++: error: arch=compute_80: No such file or directory g++: error: code=sm_80: No such file or directory gmake[3]: *** [gmakefile:188: arch-cori-gpu-opt-kokkos-gcc/obj/sys/objects/kokkos/kinit.o] Error 1 gmake[3]: *** Waiting for unfinished jobs.... CC arch-cori-gpu-opt-kokkos-gcc/obj/vec/is/utils/isio.o CC arch-cori-gpu-opt-kokkos-gcc/obj/vec/is/utils/ftn-auto/pmapf.o CC arch-cori-gpu-opt-kokkos-gcc/obj/vec/is/utils/pmap.o CC arch-cori-gpu-opt-kokkos-gcc/obj/vec/is/utils/psort.o CC arch-cori-gpu-opt-kokkos-gcc/obj/vec/is/utils/isltog.o FC arch-cori-gpu-opt-kokkos-gcc/obj/vec/f90-mod/petscvecmod.o CUDAC arch-cori-gpu-opt-kokkos-gcc/obj/sys/memory/cuda/mcudahost.o CUDAC.dep arch-cori-gpu-opt-kokkos-gcc/obj/sys/memory/cuda/mcudahost.o gmake[2]: *** [/global/homes/m/madams/petsc/lib/petsc/conf/rules:50: libs] Error 2 **************************ERROR************************************* Error during compile, check arch-cori-gpu-opt-kokkos-gcc/lib/petsc/conf/make.log Send it and arch-cori-gpu-opt-kokkos-gcc/lib/petsc/conf/configure.log to [email protected] ******************************************************************** gmake[1]: *** [makefile:40: all] Error 1 make: *** [GNUmakefile:9: all] Error 2
On Sat, May 15, 2021 at 5:30 PM Barry Smith <[email protected]> wrote:
Bitbucket is off-line. Do you need fortran stubs at this moment? If not use --with-fortran-interface=0
On May 15, 2021, at 2:51 PM, Mark Adams <[email protected]> wrote:
I can build a non-kokkos PETSc here (Cori/GPU) but I get this error with Kokkos. Any suggestions? Thanks, Mark
=============================================================================================
=============================================================================================
Trying to download git://https://bitbucket.org/petsc/pkg-sowing.git for SOWING
=============================================================================================
=============================================================================================
Trying to download git://https://bitbucket.org/petsc/pkg-sowing.git for SOWING
=============================================================================================
=============================================================================================
Trying to download https://bitbucket.org/petsc/pkg-sowing/get/v1.1.26-p1.tar.gz for SOWING
=============================================================================================
******************************************************************************* UNABLE to CONFIGURE with GIVEN OPTIONS (see configure.log for details):
------------------------------------------------------------------------------- Error during download/extract/detection of SOWING: Unable to clone sowing Could not execute "['git clone https://bitbucket.org/petsc/pkg-sowing.git /global/u2/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/externalpackages/git.sowing']": Cloning into '/global/u2/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/externalpackages/git.sowing'... fatal: unable to access 'https://bitbucket.org/petsc/pkg-sowing.git/': The requested URL returned error: 503 Unable to download package SOWING from: git:// https://bitbucket.org/petsc/pkg-sowing.git * If URL specified manually - perhaps there is a typo? * If your network is disconnected - please reconnect and rerun ./configure * Or perhaps you have a firewall blocking the download * You can run with --with-packages-download-dir=/adirectory and ./configure will instruct you what packages to download manually * or you can download the above URL manually, to /yourselectedlocation and use the configure option: --download-sowing=/yourselectedlocation Unable to clone sowing Could not execute "['git clone https://bitbucket.org/petsc/pkg-sowing.git /global/u2/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/externalpackages/git.sowing']": Cloning into '/global/u2/m/madams/petsc/arch-cori-gpu-opt-kokkos-gcc/externalpackages/git.sowing'... fatal: unable to access 'https://bitbucket.org/petsc/pkg-sowing.git/': The requested URL returned error: 503 Unable to download package SOWING from: git:// https://bitbucket.org/petsc/pkg-sowing.git * If URL specified manually - perhaps there is a typo? * If your network is disconnected - please reconnect and rerun ./configure * Or perhaps you have a firewall blocking the download * You can run with --with-packages-download-dir=/adirectory and ./configure will instruct you what packages to download manually * or you can download the above URL manually, to /yourselectedlocation and use the configure option: --download-sowing=/yourselectedlocation file could not be opened successfully Downloaded package SOWING from: https://bitbucket.org/petsc/pkg-sowing/get/v1.1.26-p1.tar.gz is not a tarball. [or installed python cannot process compressed files] * If you are behind a firewall - please fix your proxy and rerun ./configure For example at LANL you may need to set the environmental variable http_proxy (or HTTP_PROXY?) to http://proxyout.lanl.gov * You can run with --with-packages-download-dir=/adirectory and ./configure will instruct you what packages to download manually * or you can download the above URL manually, to /yourselectedlocation/v1.1.26-p1.tar.gz and use the configure option: --download-sowing=/yourselectedlocation/v1.1.26-p1.tar.gz
******************************************************************************* <configure.log>