I have 100 xml files, I want assign 100 files into 100 threads and these threads parallel execution.

You have to define the scope of the problem you're trying to solve. In other words, where is the bottleneck holding up your processing of those files.

The higher the number of threads you have does NOT mean your processing occurs faster.

The benefit you get depends on the scope of the problem. How big are these XML files? Is there a lot of processing that needs to take place on the data? How are you doing the inserts into the database? Bulk insert or individual record inserts?

In other words, is the scope of the problem I/O bound, meaning is the time being wasted waiting for disk I/O to complete or is it waiting for a network access or an SQL query?

Is the problem compute bound? Is there a lot of processing of data that needs to be done before storing the results in your database?

If the problem is compute bound, the number of cores you have in all of your CPU's on the machine running your code determines the maximum number of threads you can expect to execute at the same time. This is NO WAY MENAS THAT YOU WILL HAVE x number of threads running at the exact same time. Remember, Windows already has a few hundred threads running in various processes besides yours. You share the CPU with every other process running on the machine.

You can have more threads created, waiting to run, than there are cores on the CPU, but swamping the CPU with work will only make your threads wait to run more than they are actually running.

As for I/O bound problems, it really depends on where the bottleneck is. Remember, a single disk can only ever read one bit at a time, for one thread at a time. So if your problem is disk I/O bound, throwing a couple hundred threads at the problem will do you no good at all.

If the I/O problem is waiting for network or SQL processing to finish, again, throwing all kinds of threads at the problem will only result in hundreds of threads waiting around for something to complete.