If 6C can clock higher than 8C with the same silicon this means that the 8C limitation is likely due to temperature/power and that the suggested clocks are for an all cores overclocking, otherwise there would be no reason that a 6C enabled die would clock higher than a 8C when only one core is overclocked.
Die is said to be 82mm2 (or is it 88mm2 ?), generaly full CPUs are set at around 0.6-0.7W/mm2 TDP at most, in this case this could be more since part of the CPU is on an external I/O and that the power hungry part (the cores) dissipate much more than this, typicaly 2x at least looking at usual SoCs..
Or it could be a combination of "grey silicon" on the 6c chiplets and binning.
Not to go too OT, but they are all over the house (heat distribution) Under the dining room table, under the kitchen table beside my bed, etc.....
That's actually pretty smart. Unless you have special wiring arrangements, you don't want to pull too many amps per 20a circuit anyway. I had to spread out my mining rigs a bit back when I was doing that sort of thing.
Slight correction, The Crosshair VI is an 8+4 VRM. I wouldn't buy it new for a 3900X, but I would be willing to use it (or my Taichi) for the new chips. I'm not ready to upgrade my 1800X yet, my budget has been going to Threadripper.
Woops! I get confused sometimes. I remember there being a 6+1 X370 board that had no doublers, giving it good frequency response or . . . something. I dunno, brain getting rotten.
I like the Gigabyte and Asrock X570 lineup. Waiting for actual reviews to see if Gigabyte finally has a decent UEFI.
Only thing that cheeses me off about Gigabyte is that $599 board with 70a VRMs doesn't have an integrated monoblock like the Crosshair VIII Formula. Guess Asus wants the watercoolers while Gigabyte doesn't. Instead Gigabyte has that big-arsed aluminum plate/heatsink instead of a fan.
Excellent write-up! Just wanted to point out that the X370 Crosshair VI Hero has an
8+4 VRM design. Not as robust as the X370 Taichi but still very solid and I don't think it would have any problem running a heavily oced 3800X as it does very well with the 1800X and 2700X.
Thanks, and yeah I screwed up that one board.
3800X seems weak and the 3900X is better than expected. I've been eyeing the 3900X so I hope that part is true but we'll see.
3900x is a day-one buy for me.
I don't see the point in having a massive overkill on the vrm design. It just increases cost and decreases efficiency, unless you are running at 50%+ of it's capacity all the time. At this point it seems more like marketing bling than usefull.
I can understand having a super highend model offering it, but for 90% of overclockers it won't help any.
X370 Taichi was pretty cheap and didn't have any major efficieny problems, even when not pushing 300W through the socket. I enjoy watching Asus, Gigabyte, and MSI getting on-board with quality VRM configs. I enjoy even more watching them make that a part of their marketing push. They're paying attention.
Yeah 8+4 is really going to be "enough" for a lot of OCers who are not going to go past 150W on their chip. I was pushing 200W+ on my R7 1800x which made me happy that I got that X370 Taichi. That is going to be my minimum config for my 3900x. I would prefer an extra 2 phases on the primary just to be safe, so my choices are a bit limited. Still wondering whether I'll regret going with only 2 phases in the secondary. Probably not.
What’s the estimated support of a B350 board?
Look at what the board can do with Summit Ridge and extrapolate from there. If your B350 was only good for a 1600/1600x, then try a 3600. If your B350 was okay-ish with a 1700 then maybe you can try a 3700x. Beware of 4+3 VRM configs! Those were mostly designed for Bristol Ridge (and to a much lesser extent, Raven Ridge).
AMD has done a poor job in terms of Motherboard support, i don't like it.
Eh? X470 was a major improvement over X370, in terms of UEFI support and availability. AMD really cleaned up the design requirements too.