genuine question: why not boot the machine with 'isolcpu' and place your tasks on said cpu's ? such tasks would be not subjected to vagaries of scheduler etc...
It all depends on how much control you have on the system you target. The strategy you refer to may well work for a dedicated deployment, but if you are developing a general-purpose memory allocator targeting a wide range of applications, you might not want to impose those constraints on your users.
You could still have N threads but only want M (M<N) CPUs. For example there are situations where the single threads are mostly I/O bound, but reducing the little CPU work they do allows you to pack more threads (and e.g. serve more clients) on the same number of CPUs.
Without restartable sequences, replacing per-thread data with per-CPU data had a trade-off: per-thread data can be accessed with less overhead, per-CPU data consumes less memory. Restartable sequences give you the best of both worlds.
