@werner this is meant as the issue for the buffer sizes etc. which we talked about yesterday.
We need to improve the performance of Kleopatra when working with large files. On my Windows System when encrypting for example I have 15% CPU usage of GnuPG, 15% CPU usage of Kleopatra and it is also not bottlenecked by Disk IO.
A starting point for me would be:
GPGME on Windows uses a buffer size of 512 byte. This is ok when data is only exchanged internally in GPGME but with QGpgME callbacks are used for IO and everything is slowed down through a lot of overhead when for example a qprocess is passed new data. Every read / write callback call causes a lot of code to be exectuted and to do this every 512 bytes seems very wasteful.