Embedding Python into C++ application in a multi-threaded/concurrent environment?

I am able to use **Boost.Process** to successfully execute a Python script from within a C++ application. The C++ app can obtain the data returned by Python and this is very trivial.

The problem is performance. Creating a new process needed to execute the Python script is very expensive, e.g. 60ms on a mid-level desktop Windows PC. I have not tested on Linux yet but I'm sure that there will still be a performance hit for each process.

Things that have come to mind:

- fastcgi for python
- execute python scripts concurrently in different threads (obviously without creating a new process for each script as I do at this time)

Can **Boost.Python** help me with either of these ideas? I've seen older posts online and people seem to give up with the concurrency method. Please advise, thank you.