Valentina database uses cache to reduce disk input/output operations.
When Valentina engine starts it allocates fixed sized piece of RAM to keep here the hottest disk pages and the ones last used. The more size of cache, the more file pages it can keep.
We have upper limit of cache size also. It should not exceed the half of available RAM on start of Valentina engine. Let your computer has 1GB of RAM. Let your OS and other system utilities eat 500Mb. So on start of Valentina it gets 500MB of RAM, and maximal size of cache in such conditions should not be bigger than 250Mb. The second part of available RAM will be used to keep runtime things, such as joins, bitsets, RAM tmp tables, …
Usually the size of cache in 20-100MB is quite enough.
Keep in mind that modern MacOS X, Linux and Windows can swap out any RAM information to the disk. In this case we get disk IO operation to WRITE and later to READ. The bigger is your cache, the more chances there are to get this, especially if other active applications exist on the computer, which may require dynamically big allocations.