Hello,
To run on Windows, the first quick step would be to install on the WSL (Debian, Ubuntu, another: your choice) as if you were on Linux, including the graphical part. You can then install an X server to run the graphical interface.
The solution I mentioned to use Native Windows Python/Qt should be ready at the end of the year or beginning of 2020, but I will not have the time to work on it before mid-December. If you really want to test it and work on it yourself, I could explain briefly how we did a short test/proof of concept and which files you need to modify, but in any case it is easier to start from a full install on the WSL.
If you install MPI on the WSL, I see no reason why you would lose the parallelism as long a you are on the same Windows box. To run in parallel across multiple machines or on a cluster, you need to configure MPI across all those boxes. I assume this might be feasible if all are running the WSL, or a mix of Windows with the WSL and the same Linux version natively on other machines. But in this case, you also need to configure the batch system.
Also, the code_saturne scripts assume that in parallel, a shared filesystem is available. On most high-performance clusters, this is a true parallel filesystem such as GPFS or LUSTRE, but on smaller clusters, a shared NFS filesystem will do, as long as you avoid massive I/O.
In any case, running on a heterogeneous cluster is more difficult (you need to install the code on similar paths on each machine, avoid mixing big-endian/little-endian, and configure MPI accordingly), and load balancing can be a pain.
And when running on multiple nodes, performance will be highly dependent on network performance, so if you have Infiniband or some other fast network, things are fine, but if you only have Ethernet, things may be slow.
Running smaller cases and setting-up small to medium (or even quite large) cases on Windows and running cases which are too large for good performance on a single node on Linux clusters seems a practical/realistic approach.
Regards,
Yvan