[gpderetta]

Yes, in practice you have to dedicate the whole machine for a specific application, but the one thread per isolated core is a proven one for high performance/low latency applications.