> The first is that a cpu core is a massively distributed system, and there is no single point in time when an operation is executed.
this is critical: there might be hundreds of cycles between an instruction entering the pipeline and existing it, so you have to be precise what "ordering" means. Typically the only thing it matters is visible side effects, i.e access to the coherence layer.
this is critical: there might be hundreds of cycles between an instruction entering the pipeline and existing it, so you have to be precise what "ordering" means. Typically the only thing it matters is visible side effects, i.e access to the coherence layer.