Merge branch 'adams/aijkokkos-gpu-logging' into 'main'

Cleanup diagnostics and remove unneeded calls to WaitForKokkos

See merge request petsc/petsc!4748

Sys: widen log_view header

Merge branch 'jacobf/2021-11-02/attribute-format-sys' into 'main'

Feature: PETSC_ATTRIBUTE_FORMAT - sys

See merge request petsc/petsc!4520

apply PETSC_ATTRIBUTE_FORMAT and related fixes to sys sources

Merge branch 'jacobf/2021-09-04/petscdevice-cupminitialization' into 'main'

Replace current CUPM initialization with unified PetscDevice version

See merge request petsc/petsc!4310

- Replace the current CUPM initialization infrastructure with PetscDevice.
- Remove -petsc_use_default_null_stream and -[cuda|hip]_synchronize options
- Remove PetscCUDASynchronize and PetscHIPSynchr

- Replace the current CUPM initialization infrastructure with PetscDevice.
- Remove -petsc_use_default_null_stream and -[cuda|hip]_synchronize options
- Remove PetscCUDASynchronize and PetscHIPSynchronize
- Remove PetscCUDAInitialize(), PetscCUDAInitializeCheck(), PetscHIPInitialize(), and
PetscHIPInitializeCheck()
- Remove PetscCUBLASInitializeHandle(), PetscCUSOLVERDnInitializeHandle(),
PetscHIPBLASInitializeHandle(), and PetscHIPSOLVERInitializeHandle().
- Remove petsc_gputimer_begin and petsc_gputimer_begin
- Add -device_enable, -device_select and -device_view options
- Replace -[cuda|hip]_device with split options -device_enable_[cuda|hip] and -device_select_[cuda|hip]
- Replace -[cuda|hip]_view with -device_view_[cuda|hip]
- Add PetscDeviceInitType
- Add PetscDeviceInitialize() and PetscDeviceInitialized()
- Add PetscDeviceView()

show more ...

Merge branch 'jczhang/reuse-c-fortran-petscinitialize' into 'main'

Reuse PetscInitialize code between C and Fortran

See merge request petsc/petsc!4364

Kokkos: fix a bug in Kokkos::initialize. We need to set args.num_threads

Merge branch 'psanan-release-patch-82973' into 'main'

Developer docs, style: remove rule about using "if (!rank)" and replace all "!rank" usage with "rank == 0"

See merge request petsc/petsc!4146

Style: replace "!rank" with "rank == 0"

In SF tutorial ex1 (src/vec/is/sf/tutorials/ex1.c), add parentheses around "rank == 0". This is relevant because ! binds more tightly than +, but == binds les

Style: replace "!rank" with "rank == 0"

In SF tutorial ex1 (src/vec/is/sf/tutorials/ex1.c), add parentheses around "rank == 0". This is relevant because ! binds more tightly than +, but == binds less
tightly, so the result of the computation would otherwise change.

show more ...

Merge remote-tracking branch 'origin/jose/fix-manpages'

Fix manpages to avoid errors reported by doctext

See merge request petsc/petsc!4309

Merge branch 'balay/fix-misleading-indentation' into 'main'

g++11: fix -Werror=misleading-indentation

See merge request petsc/petsc!4298

Fix manpages: Input/Output Parameter --> Parameters

g++11: fix -Werror=misleading-indentation

Merge branch 'jacobf/2021-05-18/petscdevice-context' into 'main'

Feature: PetscDeviceContext

See merge request petsc/petsc!3994

Merge branch 'jed/cuda-timing-skip' into 'main'

PetscLogGpuTime: skip event recording when not logging

See merge request petsc/petsc!4260

PetscLogGpuTime: skip event recording when not logging

The present profiling approach waits eagerly on the event to record its
time, rather than deferring the cudaEventSynchronize until later (or
us

PetscLogGpuTime: skip event recording when not logging

The present profiling approach waits eagerly on the event to record its
time, rather than deferring the cudaEventSynchronize until later (or
using cudaEventQuery) when it isn't an expensive new synchronization. A
proper fix for this will be somewhat involved, but when we aren't
logging (e.g., no -log_view) then there is no reason to hold up
performance (by making every elementary vector operation blocking).

show more ...

Added PetscDevice to manage interaction and configuration of available devices.

Added PetscDeviceContext class to manage stream interactions for CUDA and HIP device
backends. The new class is design

Added PetscDevice to manage interaction and configuration of available devices.

Added PetscDeviceContext class to manage stream interactions for CUDA and HIP device
backends. The new class is designed to facilitate a fork-join parallelization model for
asynchronous compute.

show more ...

Merge branch 'barry/2021-07-09/add-dmdavecgetarraydofwrite' into 'main'

Add DMDAVecGetArrayDOFWrite() and DMDAVecRestoreArrayDOFWrite()

See merge request petsc/petsc!4161

Log movement of scalars between CPU and GPU seperate from vectors and large arrays

Commit-type: profiling
\spend 15m

Merge branch 'myRelease' into 'main'

Add CSV file format to petsc_tas_analysis.py

See merge request petsc/petsc!4134

Add CSV file format to petsc_tas_analysis.py

Merge branch 'barry/2021-05-16/fix-double-lines' into 'main'

Fix typos in source

See merge request petsc/petsc!3984

Remove all double blank lines from source

Commit-type: petsc-style
/2h

Merge branch 'hongzh/add-cuda-event-timer' into 'main'

Add CUDA and HIP event timers

See merge request petsc/petsc!3852