More detailed threadpool test report [sumtec]

Because Dudu's test results are inconsistent with mine (the threadpool on his side is slower than single, and the result is exactly the opposite of mine), I suspect that my tests are inaccurate, too much. So I re-tested it again. The test results are as follows:

-- About 6 or 7 warm-up tests have been performed, and the results have been deleted.

-- During the test, MSN and QQ have been canceled and the network cable is connected. There is only one ie window, and VS is enabled.
-- The release version is used for testing and is not in the debugging status.

If there are 100 testing objects and the base of testing intensity is 10000:
The total threadpool test time is 141,703,760.00.
The total single test time is 142,204,480.00.
If there are 100 testing objects and the base of testing intensity is 10000:
The total threadpool test time is 140,902,608.00.
The total single test time is 141,904,048.00.
If there are 100 testing objects and the base of testing intensity is 10000:
The total threadpool test time is 140,502,032.00.
The total single test time is 141,904,048.00.
If there are 100 testing objects and the base of testing intensity is 10000:
The total threadpool test time is 140,301,744.00.
The total single test time is 142,104,336.00.

-- In this test, MSN, QQ, and vs quit and the network cable is unplugged, without the IE window.
-- The test uses the same version (same run)

If there are 100 testing objects and the base of testing intensity is 10000:
The total threadpool test time is 139,100,016.00.
The total single test time is 139,400,448.00.
If there are 100 testing objects and the base of testing intensity is 10000:
The total threadpool test time is 138,899,728.00.
The total single test time is 139,200,160.00.
If there are 100 testing objects and the base of testing intensity is 10000:
The total threadpool test time is 138,799,584.00.
The total single test time is 139,300,304.00.
If there are 100 testing objects and the base of testing intensity is 10000:
The total threadpool test time is 139,200,160.00.
The total single test time is 139,200,160.00.

-- I can even turn off taskman. You can close all of them.

If there are 100 testing objects and the base of testing intensity is 10000:
The total threadpool test time is 138,399,008.00.
The total single test time is 138,399,008.00.
If there are 100 testing objects and the base of testing intensity is 10000:
The total threadpool test time is 138,198,720.00.
The total single test time is 138,399,008.00.
If there are 100 testing objects and the base of testing intensity is 10000:
The total threadpool test time is 138,298,864.00.
The total single test time is 138,599,296.00.
If there are 100 testing objects and the base of testing intensity is 10000:
(Note: In this test, the disk operation occurs when it is about single. I don't know what to ask. It is estimated that the result will be affected)
The total threadpool test time is 137,998,432.00.
The total single test time is 138,399,008.00.
If there are 100 testing objects and the base of testing intensity is 10000:
(Note: In this test, the disk operation occurs when it is about threadpool. I don't know what to ask, so it is estimated that the result will be affected)
The total threadpool test time is 138,499,152.00.
The total single test time is 138,399,008.00.
If there are 100 testing objects and the base of testing intensity is 10000:
(Note: In this test, the disk operation occurs when it is about threadpool. I don't know what to ask, so it is estimated that the result will be affected)
The total threadpool test time is 138,298,864.00.
The total single test time is 138,499,152.00.
If there are 100 testing objects and the base of testing intensity is 10000:
(Note: In this test, the disk operation occurs when it is about threadpool. I don't know what to ask, so it is estimated that the result will be affected)
The total threadpool test time is 138,198,720.00.
The total single test time is 138,499,152.00.

-- Minimize and then perform the same test.

If there are 100 testing objects and the base of testing intensity is 10000:
(Note: In this test, the disk operation occurs when it is about single. I don't know what to ask. It is estimated that the result will be affected)
The total threadpool test time is 138,098,576.00.
The total single test time is 138,899,728.00.

-- If this test is not performed, use another parameter (enable a notepad and copy the result)
-- Starting from here, I/O operations on the disk are no longer recorded.

If there are 150 testing objects and the base of testing intensity is 3000:
The total threadpool test time is 92,733,344.00.
The total single test time is 92,733,344.00.
If there are 150 testing objects and the base of testing intensity is 3000:
The total threadpool test time is 92,633,200.00.
The total single test time is 92,833,488.00.
If there are 150 testing objects and the base of testing intensity is 3000:
The total threadpool test time is 92,432,912.00.
The total single test time is 92,733,344.00.
If there are 150 testing objects and the base of testing intensity is 3000:
The total threadpool test time is 92,232,624.00.
The total single test time is 92,633,200.00.

-- Change another set of parameters.

If there are 1000 testing objects and the base of testing intensity is 300:
The total threadpool test time is 408,887,952.00.
The total single test time is 408,787,808.00.

-- The above results show that pure mathematical computation with a relatively small amount of computing is not dominant in multithreading.
-- We plan to use a group of abnormal parameters to test it once. Let's take a look at the results of a large number of test objects with high strength.

If there are 500 testing objects and the base of testing intensity is 5000:
The total threadpool test time is 1,721,976,080.00.
The total single test time is 1,724,780,112.00.

-- Do not try to test: 1000 tasks with a 10000 intensity. This takes about four and a half hours to complete a test.
-- Unexpected, too tired.

At least I proved that my test was quite accurate. I don't know why Dudu has the opposite result.
Oh, just use internal information ......