Use Process to actually run the task in multiple processes to speed up the performance

[minhhieugma]

• Use Process to actually run the task in multiple processors to speed up the performance
• This PR is meant to solve "Always run on 1 core CPU only "

Before

• The rate is about 5,000 tokens/s

After

• The rate is about 70,000 tokens/s
• With the same computer, 8 logical processors

Issue Link

[minhhieugma]

Due to the fact that I don't have enough time to convert all tasks to using the Process. I converted the random_brute. That's why I need this flag

[vlnahp]

It will be better if the value is declared in the if __name__ == "__main__": and then passed to the class.

[minhhieugma]

I want use_process to be used for this particular option. The rest should be used the old approach. Then, slowly we can migrate to the new approach.

If we move use_process to the outside of the class, how can we switch and meet this expectation?

[vlnahp]

I mean to share the values between speed() and the rest.
The problem is that we need to measure the speed and display it. Thus, the current number should be shared to the speed() function.

[minhhieugma]

I believe that the metrics are shared across the threads. It means that t1 = threading.Thread(target=obj.speed) is able to display it correctly.

In details, these lines of code help to to share metrics between speed()

self.cur_n = self.manager.Value('i', 0)
self.start_t = self.manager.Value('i', 0)

Could you specify which information is not shown so that I can look deeper?

[minhhieugma]

Please ignore the CodeFactor. It is an existing issue. I will try to improve it in the next coming PRs

[vlnahp]

I'll check if it really works and merge soon.

[CooperWaNg-py]

its been 2 years