E
In version A24 the Wifi disappeared after restarting the system due to freezing. My configuration was done with negative curves individually on the cores. Generally used between 11 and 29 negatives, boost on 200+, temps 95c. The instability was random, with freezes. I want to see how far the rabbit hole goes.
95c is pretty high and definitely thermal throttling you.
I just got my separate system I'm building up to A25 and set the configs back to what I had them before on that board (It's going to be the Wife's machine but her current one is working fine, so I get to experiment a bit...) and it seems to be running stable. Drive speeds did improve
some but not back up to the speeds that
@zonachristiany12280298 was seeing - but read and write broke the 1.2m IOPS barrier on multiple attempts. I'm seeing really unusually high TBW for those drives, especially since one is completely empty, hasn't been used yet (just the 4 performance tests) and the TBW is already over 1TBW. Though I don't think that's the board's issue but something with Samsung.
Otherwise, it does seem fairly stable on A25 so far. My RAM came in slightly slower with the same settings as before (a couple of thousand MB/S transfer speed slower) and with a bit higher latency (.2ns, but that could be statistically insignificant). I'm running -20 curve on all cores with +200 Mhz boost clock at the moment, and while all cores will boost to the max 5425mhz a few of the cores averages are still showing around 5350's or so, so not able to maintain that boost for as long as some of the others that are able to maintain 5390+
average.
Temps don't break 73 on mine and set my max to 85 in the BIOS. But really not approaching that at all. So I probably have some headroom with my cooler there, but probably not going to push that too much because the wife just wants it to
work and not fiddle with it once I'm done.

Anyway, didn't see that slightly off boost performance in A23 or A24. But I'd say that's still fairly acceptable if it's stable.
Hopefully they can find a fixable issue with the NVMe ports for the speeds to get back to normal. A drive going through the chipset definitely shouldn't show better performance than one attached directly to the CPU.