disable tests for no device as it breaks ci intermittently
C99 fixups
remove the idle cache mechanism from PetscDeviceContext, as it gave false positives when using multiple processes in CI
PetscDevice: Add missing Log initialization for -device_view
PetscDevice: fix scrambled output
rename PetscAssert() -> PetscCheck() and PetscAssertDebug() -> PetscAssert()
make PetscDevice now rely on PETSC_HAVE_CXX rather than PETSC_HAVE_CXX_DIALECT_CXX11
deprecate PETSC_STATIC_INLINE
apply PetscAssertFalse() -> PetscAssert() transformation to headers
add PetscAssert() and PetscAssertFalse()
make PetscInfo() variadic
SETERRQ[1-9]+ begone
require C99 and C++11 in PETSc source files:- Removed checks for inline (C/C++), and restrict (C). They are standard keywords since C99.- Removed PETSC_NOEXCEPT, PETSC_NOEXCEPT_ARG(), and PETSC_
require C99 and C++11 in PETSc source files:- Removed checks for inline (C/C++), and restrict (C). They are standard keywords since C99.- Removed PETSC_NOEXCEPT, PETSC_NOEXCEPT_ARG(), and PETSC_CONSTPEXR. They are standard keywords since C++11.- Removed checks for PETSC_HAVE_CXX_DIALECT_CXX11 in device headers.
show more ...
ci: disable cuda test (nsize=10) that's failing regularlylinux-cuda11-double:not ok sys_objects_device_tests-ex3_cuda+nsize-10local_device_context_stream_type-default_blocking # Error code: 96
re-namespaced device code
split blas interface for cupm into cupmblasinterface
add petscmacros.h and move many macros over into it
cleanup leading underscores from objpool.hpp
Sys: add sycl device
Merge remote-tracking branch 'origin/release'
Sys: require PETSC_HAVE_CUPM for device/impls/cupm
Merge branch 'jacobf/2021-11-15/better-cupmdevice-no-device' into 'main'Handle no CUDA/HIP devices betterSee merge request petsc/petsc!4556
handle no devices better
minor PetscDevice and PetscDeviceContext doc fixes and other miscellaneous tweaks
apply PETSC_ATTRIBUTE_FORMAT fixes to sys tests
1234567891011