As this issue seems to have been resolved with new drivers I'll mark this thread as solved now.
I've just answered this post on CTE.
For now, CTE is mainly focused on netcode issues, but there's a better interaction with developpers on the CTE forum since the community is smaller than on retail game forums (and more involved/less whining), so 2 developpers directly answer to the community (unlike retail game forums, that are only read by community managers).
It's a good thing a developper now directly looks into this issue but it doesn't mean he will have time to solve it...especially because there are many memory leaks causes (AMD HD5/6XXX, Realtek sound chips, Windows NDU, ...)
In my case, this solution reduced leak occurences to about once a week in x64, instead of every 2 rounds: http://forums.hardwarezone.com.sg/windows-8-338/awareness-thread-windows-8-memory-leak-ndu-sys-4309046.html
This in nothing to do with the NDU "fix"... as stated by many people after many other people have also suggested it. Thanks for having a go at helping though.
I know this thread is really really long now and it's unlikely people are going to read all the way through it, but the FIRST post in the thread already summarises the things that have been tried for the myriad fix suggestions posted.
To anyone else coming to this thread, please at least take the time to read through the first post. Carbonic has actually put a lot of detail about the issue and it's history there.
My game seem to have begun be affected by this immediately after the first map is loaded and not only after mapchange. Anybody else can confirm this?
