What is even there to call inaccurate? They said nothing about it, other than they haven't made statements about latency. Even though they talk about latency a few times in that very article, but just not in the context of specifically referring to ESRAM. They do, however, talk about it in documents to developers. I've regularly in talking about latency acknowledged, precisely as they did, that, yes, GPU's are latency tolerant, but that in no way suggests that you don't benefit from lower latency memory at all.
It's also mentioned in the leaked vgleaks info, which came from official documentation, unless people want to tell me that information isn't accurate?
http://www.vgleaks.com/durango-gpu-2/2/
See how this works? I'm sorry, but you people can't have it both ways. If the architects say something people don't want to hear, it's instantly bullshit or PR damage control. If they say something that people believe helps reinforce their already negative or pessimistic view of the system, then somehow you can take what they said to the bank, and ignore official documentation highlighting the ESRAM's low latency (although they don't specify just how low that is.) I'm not saying the architects are wrong like so many on here have said numerous times. I'm just taking all of what they say in the context of how they've presented it, in addition to how leaked official information has presented it, and what I've heard from a friend coding for the console.
I've already said on here many times that I have a friend in first party development for the system, and I've heard what I would consider very good things about the implication of the ESRAM's latency on actual, real world performance. If you want to believe I'm making that up, go right ahead, it doesn't bother me. And because there's that level of distrust, I've even posted direct quotes from an actual game developer, such as this one here
http://beyond3d.com/showpost.php?p=1738762&postcount=3325
where they explain some of the potential implications of low latency on chip memory. You might then be asking yourself "Well, how do you know the ESRAM is on chip memory and low latency." If nothing else, the architects have confirmed in these eurogamer articles that the ESRAM is built very close on chip with that GPU, and is, in fact, internal memory. It's also confirmed to be SRAM, so that means it's quite low latency by default. The architects don't need to say that the ESRAM is low latency, because it's already said in official Microsoft documentation that it is, and that it's one of the benefits or implications of the ESRAM being included in the system. The Xbox One ESRAM is a fast on chip memory scratchpad exactly as described in that post by the game developer.
Comments by architects that I think largely imply the ESRAM is much closer to the execution units of the GPU, thus making it rather low latency, are in plain site in the eurogamer articles that have been put out. They say explicitly multiple times that ESRAM is internal memory and DDR3 is external memory. They, I believe, say that very thing in 4 different times. I don't want to bother pointing it out to people, because it's in this article.
And I'll leave it at that, because I'm already explicitly aware of what kind of crowd I'm dealing with. I swear speaking with a positive outlook on this system is almost like insulting people's mothers or something.