I never quite understood why the largest cores are clocked highest. Is it because the more cores there are, the more chance that one of them will reach the highest frequency, or is just the dark silicon providing more surface area to dissipate heat?
If it's the more cores increasing the chance of one of them being higher frequency, does that also not apply to the low end - in that more cores increases the likelihood that one of them is a dog and can't clock very high at all?
Below is a silicon wafer, this happens to be one from Intel. The best chips are always at the centre of the wafer, these are normally "perfect" and tend to be the ones that can run at the highest speeds. The further out the edge you go, the more chance of a defect. Even if the yield is say 80%, i.e. 20% of the chips on the wafer are not good enough to make a product out of, the remaining 80% will be of varying "quality". So in this case, in the middle you have the Core i7's, then the further out you go you end up with i5's, i3's, Pentiums and Celerons. Ok, this is a bit over simplified, but it gives you a rough idea how it works. It's possible that it's a very good wafer, so there are no Pentium and Celerons, but a larger share of all the other parts. However, the CPU makers need to grade the CPUs in such a way that they can get resonable yields from each category, ideally as many of the highest category as possible, as they can charge the highest price for those cores.
Obviously this also means that by using this process of elimination, there's a bigger chance that an i7 is going to win the silicon lottery over an i5 and be that one chip in a thousand that can overclock like crazy. However, that doesn't mean Intel can sell that as a 6GHz chip, as they'd only get a dozen of those a month, so it doesn't make financial sense to stretch too far either. But as these chips are the very best of the wafer, they end up also being the ones that are clocked the highest. Obviously, if there's a flaw in one of the centre parts, say the cache doesn't fully work, then that ends up as an i5 for example and that might still be that one in a thousand chip that can hit 6GHz+ and the person who buys that got really lucky.
Does this make sense?
Oh come on. Lisa Su wouldn't say 'Hey guys I got a good one for you today - all the base/boost clocks of every Ryzen 3000 CPU and their prices on top' back in December 2018 and ask money for this info, because it would be made up. Certainly this info would not be privy to some guy sitting in front of his PC all day up in Scotland.
I think you forget something, the plan was to launch Ryzen 3000 much earlier in the year, maybe not January, but February/March. The delay is largely down to the motherboards with the X570 chipset not being ready and in fact, they're still being tuned and will be tuned until the last minute.
So yes, I do believe AMD had everything figured out at that point in time already, but had to push things back. There might be more to it than the motherboards, but it did have a huge contributing factor as to why the platform is only launching at Computex. I guess you don't really work in the industry, so you wouldn't know the first or the last things about this, but instead, you hang out on forums and spreading FUD about something you know very little about.
As to why this info was sent to Jim, I don't know. Maybe someone likes him and thought he should be in know? Kind of like when you get a free donut at dunkin' for being a regular customer.
