| Features deprecated in versions 5.2, 5.1 and 5.0 were removed | N/A | Backward compatibility |
| Full support for C23 was added | P | |
| Full support for C++23 was added | P | |
_ALL suffix to the device-scope environment variables | P | Host device number wrongly accepted |
num_threads now accepts a list | N | |
Supporting increments with abstract names in OMP_PLACES | N | |
Extension of OMP_DEFAULT_DEVICE and new
OMP_AVAILABLE_DEVICES environment vars | N | |
New OMP_THREADS_RESERVE environment variable | N | |
The decl attribute was added to the C++ attribute syntax | Y | |
| The OpenMP directive syntax was extended to include C 23 attribute
specifiers | Y | |
| All inarguable clauses take now an optional Boolean argument | N | |
| For Fortran, locator list can be also function reference with
data pointer result | N | |
| Concept of assumed-size arrays in C and C++ | N | |
| directive-name-modifier accepted in all clauses | N | |
| For Fortran, atomic with BLOCK construct and, for C/C++, with
unlimited curly braces supported | N | |
| For Fortran, atomic compare with storing the comparison result | N | |
New looprange clause | N | |
Ref-count change for use_device_ptr/use_device_addr | N | |
| Support for inductions | N | |
| Implicit reduction identifiers of C++ classes | N | |
| Change of the map-type property from ultimate to
default | N | |
self modifier to map and self as
defaultmap argument | N | |
| Mapping of assumed-size arrays in C, C++ and Fortran | N | |
groupprivate directive | N | |
local clause to declare target directive | N | |
part_size allocator trait | N | |
pin_device, preferred_device and target_access
allocator traits | N | |
access allocator trait changes | N | |
Extension of interop operation of append_args, allowing all
modifiers of the init clause | N | |
interop clause to dispatch | N | |
message and severity clauses to parallel directive | N | |
self clause to requires directive | N | |
no_openmp_constructs assumptions clause | N | |
reverse loop-transformation construct | N | |
interchange loop-transformation construct | N | |
fuse loop-transformation construct | N | |
apply code to loop-transforming constructs | N | |
omp_curr_progress_width identifier | N | |
safesync clause to the parallel construct | N | |
omp_get_max_progress_width runtime routine | N | |
strict modifier keyword to num_threads | N | |
atomic permitted in a construct with order(concurrent) | N | |
workdistribute directive for Fortran | N | Renamed just after TR12; added in TR12 as coexecute |
Fortran DO CONCURRENT as associated loop in a loop construct | N | |
threadset clause in task-generating constructs | N | |
nowait clause with reverse-offload target directives | N | |
Boolean argument to nowait and nogroup may be non constant | N | |
memscope clause to atomic and flush | N | |
omp_is_free_agent and omp_ancestor_is_free_agent routines | N | |
omp_target_memset and omp_target_memset_rect_async routines | N | |
| Routines for obtaining memory spaces/allocators for shared/device memory | N | |
omp_get_memspace_num_resources routine | N | |
omp_get_submemspace routine | N | |
ompt_target_data_transfer and ompt_target_data_transfer_async
values in ompt_target_data_op_t enum | N | |
ompt_get_buffer_limits OMPT routine | N | |