But what is the difference between 'non-blocking' operations (using the CPU prefix 'LOCK' or LL / SC, etc.

You say “like the following (non-blocking only)”. But if you want your code to be very portable, it will not be the fastest in all cases. It is ugly and slow (doing 2 atoms instead of 1). So far (until openmp 3.1 is implemented), I used the built-in wrapper functions in C ++, where you can choose which version to use at compile time: template inline T my_fetch_add( T * ptr, T val) no checking how to run the C preprocessor. The following built-in functions are intended to be compatible with those described in the Intel Itanium Processor-specific Application Binary Interface, section 7.4. none needed checking whether we are using the Intel C compiler. Using the GNU Compiler Collection (GCC) 6.51 Legacy sync Built-in Functions for Atomic Memory Access. yes checking for gcc option to accept ISO C89. This function-like macro is available in C++20 by default, and is provided as an. He believes that the Intel C / C ++ compiler also supports this, but I have not tried it. o checking whether we are using the GNU C compiler. define hasbuiltin(x) 0 // Compatibility with non-clang compilers. There, good work if you use gcc (or g ++), see atomic built-in functions: Since we must bring the value from memory in order to gradually increase it, it only makes sense to access it, say, the CPU register and put it in the stream variable. As in openmp 3.1, there is support for collecting atomic updates, you can fix either the old value or the new value.

Intel c compiler __sync_fetch_and_add