proxy generator - performance question

i got a shoot with multiple camera's in my resolve project, and i decided to give the proxy generator a try to generate 1080p prores files for all of the camera's. (sony fx6, gopro, vhs, dji drone). Everything looks like it has been replaced like it should by the proxies, indicated by the purple icons on every clip. the option "proxy handling" is set to "prefer proxies". So practically, all the camera's are the same files, and should perform the same, right?

but working from the source tape, i still notice a difference in performance when i scrub through the sony camera and for example the dji. scrolling through the sony is still a tiny little bit choppy, while scrolling through the drone files is buttery smooth. I'm talking about quick scrolling with the jogwheel, or playing back 8/16x speed.

the confusing part is that when i place the clips on a timeline (1080p), i don't see any difference, it's all super smooth. It's all coming from the same file location (qnap server, +-900Mb read/write)

There is a difference between the amount of audio channels, but i turned all the channels off for a test, and that did not make a difference. I can't seem to find an explanation for the difference in performance, and that's bugging me what could be it?

Could it be that the system has to try harder to correlate? Somehow it keeps the connection if you want to switch back, while the clips alone in a timeline will behave like originals.

not sure.. i now tested it with another project, same-ish results, what originally 4k sony mxf files were, are scrubbing choppier than the sony mp4 files, while they are all coverted to 1080p prores files.