I have tried importing many tables of different sizes from a database into Qlik Sense, in a qvd file. I found that one table with around 8 million rows (1.5 GB) took 20 mins to load. On the other hand, a table with 9 million rows and size 2 GB took just 5 mins. I'm not sure why there is such a disparity? Both the table are from the same database.
If you are loading from the database performance could be dependent on the database itself regarding usage and perhaps table locks etc on the data itself. Was data already cached for 1 table and needed to be loaded from I/O for another. Did a log backup kick off that impacted I/O speed etc.
Assuming database is the same and nobody using it except for you ... then I would ask do you do any Transformations for 1 table and not for the other.
If you comment out the STORE to QVD what is the time it takes? Are the two closer?
If it results from the literal storing to the QVD's it could be disk fragmentation when writing the first file. Or it could have been the result of Virus scanning software. When I was a Qlik Admin I had to constantly ensure my Network team excluded directories where I stored QVD files from their scanning to ensure they didn't slow me down.
Hope some of those things might help you pinpoint why.