Tool/software:
Hello TI Experts,
According to the TI Arm Clang compiler documentation (2.5. Characteristics and Implementation of Arm C++ — TI Arm Clang Compiler Tools User's Guide), it is stated that C++ atomic operations are "not implemented or fully supported."
Could you please clarify what this means in practical terms?
In my experience, I’ve been able to declare std::atomic<int> variables and perform basic operations such as store, load, and fetch_add using std::memory_order_seq_cst. However, I would like to understand the limitations more clearly:
- Are there specific atomic operations or types that are not supported or may behave unexpectedly?
- Are all standard memory orderings (e.g., std::memory_order_seq_cst, relaxed, acq_rel) implemented and correctly honored?
- Are there known issues or caveats when using atomic operations in multi-threaded scenarios with this compiler?
Any additional technical insights or documentation references would be greatly appreciated.
Best regards
Luca