3 ms·
I believe GP is not talking about thread parallelism, but rather the compiler-controlled instruction reordering. Compilers typically reorder instructions in the
by kccqzy 8y ago
I believe GP is not talking about thread parallelism, but rather the compiler-controlled instruction reordering. Compilers typically reorder instructions in the hope that it will improve performance. For example if you have v+=2, the compiler can choose to load v into a register, and then continue with the rest of the statements, and ten instructions later perform the actual addition and store it back into memory.
- caf 8y agoYes, that's what I was referring to. When single-stepping such code you'll often see it skip backwards and forwards between the interleaved source statements.