As the single computing power of modern CPUs is gradually increasing (the physical core and logical computing power are increasing synchronously), how to better determine the required number of threads when configuring the number of threads in multi-threaded tasks?
Considering a typical traditional server scenario that responds to requests in a multi-threaded mode, does it mean that when the throughput of the logic core doubles, I can complete the same task with half of the threads while keeping the response latency of the service unchanged?