Implementation of HIP for sys and vec packages

This implements the vec functionality including tests.
In the sys level, we handle the initialization issues, but it is tested
at the vec level.

Works

Implementation of HIP for sys and vec packages

This implements the vec functionality including tests.
In the sys level, we handle the initialization issues, but it is tested
at the vec level.

Works with ROCm 3.9.

show more ...

Add -matlab_engine_host hostname runs the Matlab on the machine hostname

Also cleaned up error handling for PETSC_MATLAB_ENGINE_()

Commit-type: error-checking, feature
/spend 1.5h
Reported-by: Kaus

Add -matlab_engine_host hostname runs the Matlab on the machine hostname

Also cleaned up error handling for PETSC_MATLAB_ENGINE_()

Commit-type: error-checking, feature
/spend 1.5h
Reported-by: Kaustubh Khedkar <kkhedkar9879@sdsu.edu>

show more ...

Additional material in PetscAttachDebugger() manual page

Add link to https://github.com/Azrael3000/tmpi allowing debugging with OpenMPI

Commit-type: documentation
/spend 10m
Reported-by: Sajid Ali

Additional material in PetscAttachDebugger() manual page

Add link to https://github.com/Azrael3000/tmpi allowing debugging with OpenMPI

Commit-type: documentation
/spend 10m
Reported-by: Sajid Ali <sajidsyed2021@u.northwestern.edu>

show more ...

PetscAttachDebugger() should return error code of PETSC_ERR_SYS, not 1 when a system called failed

Commit-type: error-checking
/spend 3m

Add support for -debug_terminal Terminal to use Apple's Terminal App

Make Terminal the default device to display the debugger on Apple
One can still use xterm with -debug_terminal xterm

This all

Add support for -debug_terminal Terminal to use Apple's Terminal App

Make Terminal the default device to display the debugger on Apple
One can still use xterm with -debug_terminal xterm

This allows the use of cut-and-paste which is painful with Apple's xterm

Commit-type: feature, debugging
/spend 90m
Reported-by: Hapla Vaclav <vaclav.hapla@erdw.ethz.ch>

show more ...

Merge remote-tracking branch 'origin/release' into master

xl: rename petscwaitonerror to petscwaitonerrorflg to avoid name conflict with PetscWaitOnError() fortran interface

Reported-by: "Dobrev, Veselin A." <dobrev1@llnl.gov>

Merge remote-tracking branch 'origin/release' into master

Better message when aborting
Suggested-by: Mark McClure <mark@resfrac.com>

Report all GPU errors as PETSC_ERR_GPU so that the test harness can retry any GPU failure

We still get occassional errors:

ot ok ksp_ksp_tutorials-ex71_bddc_cusparse # Error code: 180

That occur a

Report all GPU errors as PETSC_ERR_GPU so that the test harness can retry any GPU failure

We still get occassional errors:

ot ok ksp_ksp_tutorials-ex71_bddc_cusparse # Error code: 180

That occur after the cuSPARSE library must have been loaded since the load command already succeeded

Commit-type: error-checking, CI, enhancement
/spend 15m

show more ...

Remove PetscMallocValidate(__LINE__,PETSC_FUNCTION_NAME,__FILE__); I added to the signal handler
because I forgot just a few days before I had already added it (with a nice message as well).

Commi

Remove PetscMallocValidate(__LINE__,PETSC_FUNCTION_NAME,__FILE__); I added to the signal handler
because I forgot just a few days before I had already added it (with a nice message as well).

Commit-type: fix of unneeded code
/spend 5m

show more ...

Call PetscMallocValidate() on signal and error conditions to check if the final memory state is correct

Commit-type: error-checking
/spend 10m

Merge branch 'barry/2020-09-26/error-mans' into 'master'

Improve manual pages for PetscError() and the various error handlers including...

See merge request petsc/petsc!3249

Improve manual pages for PetscError() and the various error handlers including providing cross links in the manual pages

Commit-type: documentation
/spend 45m

Add specific PETSc error flag for GPU not having resources available

Try three times to initialize CUDA/cuBLAS and cuSPARSE before erroring. This is for GPU tests that sometimes fail since the GPU
i

Add specific PETSc error flag for GPU not having resources available

Try three times to initialize CUDA/cuBLAS and cuSPARSE before erroring. This is for GPU tests that sometimes fail since the GPU
is busy with other tests when a new test is started.

Commit-type: error-checking, testing
/spend 1h

show more ...

Need initialize to prevent test systems from complaining about possibly uninitialized variables.

Commit-type: portability-fix
/spend 5m
Reported-by: GitLab CI

Call PetscMallocValidate() when SEGV or SIGBUS are received

Commit-type: debugging
/spend 10m

Deprecate -debugger_nodes and replace with -debugger_ranks

PETSCABORT calls abort if run under -start_in_debugger, rather than MPI_Abort, this allows debugging the crashed program

\spend 1h

Commit

Deprecate -debugger_nodes and replace with -debugger_ranks

PETSCABORT calls abort if run under -start_in_debugger, rather than MPI_Abort, this allows debugging the crashed program

\spend 1h

Commit-type: bug-fix, usability
/spend 5m

show more ...

Fix -debugger_nodes x,y,z so that ranks that are not in the debugger sleep before calling MPI_Abort so they do
not bring down the user's debug session.

Commit-type: bug-fix, feature
/spend 1.5h
Repo

Fix -debugger_nodes x,y,z so that ranks that are not in the debugger sleep before calling MPI_Abort so they do
not bring down the user's debug session.

Commit-type: bug-fix, feature
/spend 1.5h
Reported-by: Ed Bueler <elbueler@alaska.edu>

show more ...

doc: fix build errors

Turn on checkbadSource test to generate an error when found; fix all source code that causes errors

Commit-type: portability-fix, testing-fix, style-fix, feature, maintainability
/spend 1.5h

Add support of CUDA/HIP initialization at the same time