Relationship between application software threads and CPU cores

As a general rule , It is believed that the empirical value of thread pool size should be set as follows ：（ among N by CPU The number of ）

• If it is CPU Intensive application , Then the thread pool size is set to N+1
• If it is IO Intensive application , Then the thread pool size is set to 2N+1（ because io When reading data or caching , Thread waiting , At this time, if you open more threads , Can effectively improve cpu utilization ）

If only one application is deployed on a server and there is only one thread pool , So this estimate may be reasonable , It needs to be tested and verified by itself .

however ,IO Optimization , Such an estimation formula may be more suitable for ：

The best number of threads = （（ Thread wait time + Threads CPU Time ）/ Threads CPU Time ）* CPU number

Because obviously , The higher the percentage of thread waiting time , More threads needed . Threads CPU The higher the proportion of time , The fewer threads needed .

Here's an example ：

For example, average per thread CPU Run at 0.5s, And thread waiting time （ Not CPU The elapsed time , such as IO） by 1.5s,CPU The core number is 8, So according to the above formula, it is estimated that ：((0.5+1.5)/0.5)*8=32. This formula is further transformed into ：

The best number of threads = （ Thread waiting time and thread CPU The ratio of time + 1）* CPU number

The empirical value of thread pool size just mentioned , It's actually an estimate of this formula .