MATMPIAIJCUSPARSE: add support for sparse MatMat operations

MATAIJ: some fixes for device builds

MatSetValuesCOO: use cuda memory if possible

MatMPIAIJGetLocalMatMerge: support for merging two matrices

MatSetValuesCOO: perform addition of repeated entries when INSERT_VALUES is specified

use temporary data instead of persistent storage in cusparse implementation

MatSeqAIJCompactOutExtraColumns_SeqAIJ: no nested loops

MatTransposeMatMultSymbolic_MPIAIJ_MPIAIJ: minor

MATSEQAIJKOKKOS: add partial code for MatMat operations

MATSEQAIJCUSPARSE: add support for sparse MatMat operations

Minor

CUDA: add a couple of missing Lazy initializations

Convert MPI error type to PETSc error with string message for all MPI calls

Now PETSc examples will ONLY return PETSc error codes and never MPI error codes directly so we can understand and post-pro

Convert MPI error type to PETSc error with string message for all MPI calls

Now PETSc examples will ONLY return PETSc error codes and never MPI error codes directly so we can understand and post-process their
errors better.

The test harness will now automatically retry tests that fail with MPI, this may help with Intel MPI that produces
seemingly random failures.

Commit-type: error-checking
/spend 30m

show more ...

checkbadSource (master) fixes

Merge remote-tracking branch 'origin/release'

checkbadSource: apply rules to *.cu *.cpp sources, and expand CHKERRQ check to CHKERR(Q|MPI|CUDA|CUBLAS|CUSPARSE)

Merge remote-tracking branch 'origin/release'

call PetscCUDAInitializeCheck() in MatConvert_SeqAIJ_SeqAIJCUSPARSE since it uses cuSparse()

Missing this sometimes produces errors in the CU tests, but other times it passes. I don't understand why

call PetscCUDAInitializeCheck() in MatConvert_SeqAIJ_SeqAIJCUSPARSE since it uses cuSparse()

Missing this sometimes produces errors in the CU tests, but other times it passes. I don't understand why it is not needed.

Commit-type: error-checking, Bug-fix, portability-fix, testing-fix, CUDA
/spend 10m
Reported-by: GitLab CI

show more ...

Merge branch 'barry/2020-09-07/matstructure-unknown' into 'master'

Barry/2020 09 07/matstructure unknown

See merge request petsc/petsc!3182

Merge branch 'adams/cusparse-host-buffer' into 'master'

Adams/cusparse host buffer

See merge request petsc/petsc!3502

added cache of buffer to avoid managed memory mallocs

Introduce UNKNOW_NONZERO_PATTERN for MatAXPY() make it the default for TS when using both IJacobian and RHSJacobian

Provide TSSetMatStructure() to allows users to indicate the nonzero structure rela

Introduce UNKNOW_NONZERO_PATTERN for MatAXPY() make it the default for TS when using both IJacobian and RHSJacobian

Provide TSSetMatStructure() to allows users to indicate the nonzero structure relationship between the RHS and I Jacobian

When the RHS Jacobian matrix is not set and is obtained with a MatDuplicate() from the I Jacobian set the flag to SAME_NONZERO_PATTERN

Add MatStructures and update SLEPc to compile with no warning message after the new MatStructure MAT_UNKNOWN_PATTERN was introduced

Commit-type: optimization
/spend 25m
Thanks-to: Ed Buler

/spend 35m

Update SLEPc branch to support new enum type

show more ...

Merge branch 'adams/landau-gpu-assembly' into 'master'

Adams/landau gpu assembly

See merge request petsc/petsc!3423

Adding Cuda and Kokkos assembly. Added Device assembly to Landau operator. Added Kokkos test mat/ex5k.

Merge remote-tracking branch 'origin/release'

AIJMKL: cannot print a scalar in case of complex