we pretty much know that the Series X is not exactly 12.00TF, the options are:
60CU - 1.6GHZ - 12.28TF
56CU - 1.7GHZ - 12.18TF ⬅ most likely one
52CU - 1.8GHZ - 11.98TF
or they go with a more uneven clock speed to exactly hit 12TF but why would they do that? there's no real reason to it other than aesthetics
I don't know, seeing how MS desperately aimed for a round 6TF figure for X1X, I can see them doing everything possible to reach that round 12.0x TF again, so they can use it as a marketing tool instead of something like 11,8 for example. And the "twice as powerful as X1X!" as well.
So something like 1563Mhz (60CU), 1674MHz (56CU) or 1803Mhz (52CU), just to be that 0,01TF above the magic 12.
And if we do some fancy math, looking at 5700XT that has 40CU and a die size of 251mm, it takes an estimated 6,275mm^2 die size per a single CU. In comparison the 60Cu Radon VII that's also a 7nm chip uses estimated 5,5mm per a single CU. That's a dummy math I know, but the numbers of ROPs, TMUs, cache etc. are more or less associated with the number of CUs anyway, so let's use that 6,275mm number for some nerd science here, assuming RDNA2 won't be any bigger/more complex.
So a 60CU, given the same 7nm process and RDNA1 architecture, would translate into a 376mm GPU, which given the estimated die size of XBX APU of 405-410mm wouldn't leave any space left for either an 8C CPU, the RT cores, all of the the controllers - SSD, RAM, Ethernet, WiFi, BT, BD and what's not. Bare in mind a single 7nm Zen 2 CCX (4C+cache chiplet) is roughly 31mm^2, so for 8C we need about 62mm alone, and what about the rest of the mentioned components?
56Cu = 351mm, still rather too big for what the consoles will provide.
52CU = 326mm - now we're getting somewhere.
But the thing is, Navi GPUs are able to achieve a sustainable 2-2.1GHz clock speeds. Yes, I know it needs high voltage and a as a result a beefy cooling, but here the catch -the Hovis Method, which is basically nothing else than the combination of undervolting and overclocking, so a GPU/CPU uses just as little voltage as really necessary to reach the desired frequencies. So let's say the GPU in XBX will reach something between 1850/1900/1950Mhz, now we're at 50-48CU, 313, 307 and 301mm respectively, plus the mentioned 62mm for the CPU, leaving some actual space left for the other stuff.
So I don't know, do with those calculations whatever you want guys lol.
Dare you say how is this anyhow thread related? NV said long time ago they're not interested in consoles with their business model, a.k.a. tiny profit margins, and both next-gen consoles were as well long time ago confirmed to be yet again using complete AMD solution.
And I don't know how to even comment on that one, nice spam I guess?