I think the issue here is the shutting down of all possible conclusions.  A rumor comes up that the Xbox One has heterogeneous coherent memory, that the processor speed has been increased, or that the amount of memory available to the system has been increased and there is an immediate squashing of any speculation.  Not with facts, but unsubstantiated opinion. 

You're right.  Ethomaz didn't exactly say the CPU couldn't be increased in processor speed, and as he pointed out, he stated that the conclusion on how to determine the CPU speed was incorrect.  But your conclusion on that thread is incomplete and thusly wrong.  He had to be dragged, kicking and screaming to offer a conclusion that a CPU speed was possible.  It wasn't his initial conclusion.  More to the point, he never could offer why the original author's conclusion as to a CPU increase was inaccurate, only that it was.  And his argument for why it was so, was because he could make a similarly inaccurate claim.

The truth of the matter is, neither were correct on the hows and whys because there is absolutely no way to determine the frequency of a processor without putting it on a scope.  No component, other than the CPU itself, determines the processor frequency.  You can analyze a motherboard six ways to Sunday and there isn't one component on that board that'll help you determine the CPU frequency.  None. 

The fact is that the author heard a rumor, likely was given information from a reliable source, and in order to justify the reporting of the rumor as reliable, concocted a smoke and mirrors method of determining the CPU speed.  And I've said this before. 

Coherency and heterogeneous memory is actually a big deal.  In fact, it was a big enough deal that an AMD marketing person felt it was important enough to imply that the PS4 had it and it would contribute to the PS4's performance and that the Xbox One didn't.  So, to you it may not be important.  However to AMD's Marketing team, it was.  It was also a big deal to the original poster of that thread and those who staunchly suggested that in no possible way could the Xbox One feature coherent and heterogeneous memory. 

You argue that it is disingenuous for me to infer RAM and memory when NAND memory isn't true RAM.  Fine.  Now acknowledge the disingenuousness of the various people who have posted to counter any argument that the Xbox One could in fact have any of the features mentioned.  The argument against coherency and heterogeneous memory was extremely disingenuous, because any sane, rational, and logical person could see and infer that it was possible.

Now finally, sure.  NAND isn't RAM.  However, NAND can be used as RAM.  That's the beauty of NAND memory.  It can be used both as a storage format as a replacement for magnetic media, but it can also be used as a system memory replacement for DRAM. 

Likewise, whenever there is a negative rumor about the Xbox One, with the same group of people they ALWAYS seem to be true.  The source, despite being wrong frequently, is always accurate and trustworthy.

So, no.  I'm not being disingenuous.  People seem to have this inability to take a problem, look at it objectively, and see the possible conclusions.  That, in my opinion, is disingenuous.  Look at MY arguments in those threads, and you'll see I'm not blind, but not every rumor is as cut and dry as you'd like to believe.  Rumors are obfuscations of truth, often with details missing.  So rather than shoot them down from the hip, they are something that requires an outside-of-the-box analysis.  Either there are people incapable of doing that, or they are smart enough to do it but unwilling to do so because they want to crush any positive discussion and foster any negative discussion of the Xbox One. 

Your argument precludes that the cost of a second GPU hasn't already been factored into the costs of the Xbox One.

You assume the cost of a second GPU must be added into the existing costs.  You assume that the $600 retail cost of a GPU is the same as the component cost, it isn't.

If Microsoft were to add a second GPU the only thing it would be adding is the second GPU.  Not the memory.  There is 175mm of space for something in that 393mm SoC.  That's a lot of room for nothing.

Not in a million years. Do you know that the 8G NAND ram has a bandwidth of roughly 100MByte/s. Good luck with that as a replacement for 68GB/s main dram. (It will also degrade while being written to, so you would avoid anything that hs to to with frewuent reads/writes).

It's called cooling, air flow lol. Every system has it.

When did the SoC increase to 393mm^2? Where did you get the 175mm^2 from? You are obviously still living in the land of delusion so here are a few minor points.  This mistertroll fabulates about a dgpu with around 2000-2300units (of whatever) made in 20nm with a w2w connection to the SoC.

a) The only company that gets anything out (as complex as a cpu) near 20nm is Intel with its new 22nm processor lines. Yields are not great so there are limited numbers of these things. Process technology wise, Intel is ahead of TSMC by about 1-1.5years and ahead by about 2-2.5 years of AMD. Tell us, who do you think makes these miracle dgpus (it certainly isn't intel)?

b) 20nm chips use a different technology than 28nm chips. Tell us which company is capable of makeing a chip that incorporates two different process technologies into one chip?

c) Tell us, where is the memory for this magical dgpu? You certainly need fast memory for this super dgpu.

d) Why would there be a primary gpu at all? This magical dgpu is faster than anything on the PC market, so why waste money on a "measly gpu" at all?

e) How would you feel as a developer if you were developing on developer units for years, only to be told 1 month before release "It was all a joke, we have a completely different gpu in our box"?

How would you cool the two layers with air? You know, the SoC (bottom layer) that is radiating approx. 100Watts into the layer with the magical dgpu..

So you're saying they've dedicated 175mm, more than a third of the SoC to cooling.  Right. 

First, the way to cool an IC isn't to keep the heat inside the component, it's to disipate it through radiate the heat out of it.  Hence why they use heat sinks.

Second, the Xbox One SoC is Wafer-to-Wafer, meaning it's stacked.  Adding silicon between wafers wouldn't cool it, it would actually increase the heat held into the IC.  If we applied your logic to a real-world-example it'd be like placing a cool/hot pack between two heating pads and thinking it'll keep things cooler.  No, eventually the cool gel absorbs the heat and then begins radiating it's own heat out into the layer above it. 

So laugh at me too then I guess. Shouldn't be too much trouble for you.

I have no idea how powerful the X1 will be in the end, but I think that the dGPU rumours are legit. Call me stupid if you will

I can't believe this thread is still going and some people actually believe this.

Numerous people have debunked the claims made in the OP with legitimate questions whose answers have no basis in reality.

Wow, funny how time flies.  You're talking about the top performance of a card, not an embedded IC.  And DDR3 DRAM in the Xbox One is 68Gb/s not GB/s.  You're mixing up your bits and bytes.  100MB/s is pretty close to 1Gb/s.  That said, the performance available through NAND memory is capable of more than that.  NAND-based SSDs are capable of over 200MB/s, or roughly 2Gb/s.  Embeded NAND memory would be capable of potentially faster speeds.  Is it the same performance as DDR3 memory?  No.  It doesn't need to be.  It needs to be faster than the HDD, and it is.

With up to 100 million cycles on modern NAND memory, I doubt NAND degredation would be an issue considering NAND memory with 100K cycles was considered to have a 10 year lifespan.  If I'm not mistaken, 100M is 1000x greater than 100K, which would make that a 10,000 year lifespan.  I'm not worried.

Son, you really should know better than to bring a knife to a gun fight.  You're bound to get hurt.