Just been testing out Ryzen 5700 in x300 over the last day and I am surprised at the instantaneous temp spikes.. There's a good chance it is just a poor installation by myself and i'll do it again to be sure. Spikes from 30c to 70c in 1 second are something I have never seen before. Even though AMD has said "running at 95C is no problem" (which may be true), there is something funky going on with the thermal resistance.
It seems like there is unusually bad thermal resistance somewhere between the cores and the heatsink. The heat isn't even getting the chance to saturate the heatsink tested on both the wraith stealth and noctua nh-l9a. I am not expecting perfection, but in an older rig from a few years back, I measured a result of 9C delta between the internal core temperature reading and an external temperature sensor placed on the exterior heatsink fins.
But on the x300, the delta between the heatsink and the core reading seems absolutely massive. Just a guess its at least 25C. Even after benchmarking to reach equilibrium, the heatsink does not feel hot like it should. It's just... warm.
I saw some posts about Ryzen heat spreaders not being perfectly flat and possibly concave, causing an issue when combined with a very flat heatsink. Or also, if they are flat but paired with a convex heatsink, the edges might have a bit too much clearance and not enough pressure (not a problem with previous chip designs with die in the center). So since I used the minimal amount of TIM, it would produce a poor result in both of these scenarios & I would be better off over applying it. One user reported he re-seated his cooler 10 times before noticing the uneven IHS edges, then sanded it flat.
The other thing I read is that the chiplet layout used by AMD puts these dense cores towards the side/edges of the chip where they are less likely to be directly beneath a row of centered heatpipes. This could explain the terrible resistance, as the heat cannot be instantly absorbed as it would with the properly located heatpipe, and the heat has to first creep into the center region of the chip before it can be soaked. I'm curious whether the same spikes happen on a water block, which can extend past the edge of the chip and provide low thermal resistance on the edge regions just as well where the air coolers cannot.
Anyway, going to re-do the cooling to see
