No need to bring the PS4 into this specific debate we're having. I know the PS4 memory architecture is better and more efficient. I'm strictly talking about the Xbone's memory here, and how it compares to the previous generation. I want to know how something that is considered an evolution, and more capable than its previous iteration, make it automatically more difficult, and where people are all of a sudden getting the idea that the 360's memory architecture was difficult to develop for? I feel like some people here are just grasping for straws, just to make the Xbone's memory architecture seem THATT much more inferior to the PS4's method.
No one is saying that the 360 was difficult they were saying it was "more difficult than not having edram" that doesn't add much complexity.
and of course the PS4 is relevant in this, the difficulty of development is entirely based on the difficulty of development for the opposition console or PC, That's what gives us a baseline of what is easy and what is hard. It's like playing a sport, it's only as hard as whoever your playing against.
Let's look again at that quote you posted:
now texture out of ESRAM if you want to
Ok so a new ability to texture out of esram, you know what this means? new API for it. But 32 mb is small? can you put full textures in there or are you forced to use PRT? I'd guess PRT, new API for it. So you would have to use 2 new APIs to do this, how good are these new APIs? we don't know, from the developer quote in the OP of this thread we could say "horrible"
"Gosh, it would sure be nice if an entire render target didn't have to live in eDRAM," and so we fixed that on Xbox One where we have the ability to overflow from ESRAM into DDR3 so the ESRAM is fully integrated into our page tables and so you can kind of mix and match the ESRAM and the DDR memory as you go.
Do you know what this means? It means the memory is now addressable. that means the developers can choose where to put things, This allows for more possibilites, HOWEVER, this means developers need to put things in the right place, now that we know they do this, let's look again at the esram and the "1024 bit bus"
It's not a 1024 bit bus, it's 4x 256 bit bus's linked on 4 pipelines each of 8mb esram, the developers need to manage these as they are now addressable to get access to the 109gb/s read speed, it seems you need to have your data in all four "8mb chunks"
Esram is not edram, the Article you quoted is MS PR and shows falsehoods even in the small snippet you quoted.