Sure, you are correct they stated a 50% reduction in CPU from DriectX 12.  Mostly in this chart showing how it uses multiple cores more efficiently and effectively:

For the GPU they stated a 20% redution in processing time per frame.  It's not really shown exactly in a slide, but in numbers like the ones the slide like these. They show the DX11 and DX12, but hard to read I appoligize I didn' take it.

Well, assuming Xbox One will  at least some of the benifit of a 50% optimization, but probably already is ahead of PC's there.   But even if it doesn't, they stated a 20% reduction in GPU render time per frame, or something to that effect.

Mantle is supposed to give a 20% improvement at well.  It's a bit like mantle.

So I went with a conservative estimate of 20%.  The baseline of the GPU.  However, the scaling across the 8 core CPU might result in greater improvements.

Yet you offer nothing to support your claim other than a rude and snide remark.

Sure I believe you.

Valve killing MS or Windows does sound a bit far fetched ...

Gaming is only one part of windows but there are many more reasons to use windows such as multimedia and productivity purposes. I don't think steam OS was meant to replicate all of those things and even Valve MUST KNOW that PC  gamers can't avoid windows because directx is a PC gaming industry standard. 

At best I think they only meant to aim the clueless and vocal part of the PC gaming community but even then their intentions don't come off as being sinister. 

The improvements are relevant to configurations (as shown in the slides) where you are hamstrung by slow high-level API draw calls that depend on CPU to a high degree.

XB1 architect already confirmed that XB1 has a custom variant of DX that already has low-level API for efficient draw calls.

Translation : this will speed up PCs, but will do very little for XB1 (if anything). Now in PR/Marketing-speak, they would LOVE to be able to say 'this will massively improve XB1 graphics'. However, they can't, for two reasons :

(1)- It's not true. If it were, they'd be screaming it from the rooftops.

and

(2)- If it were true, it would mean that the XB1 dev team were complete idiots (obviously untrue), and massive liars (also obviously untrue).

XB1 graphics WILL improve, just like always, as people get more comfortable with the SDKs and learn how to maximize how things look without hitting areas that cause large performance hits.

At the end of the day, the choices to go with DDR3 and a considerably weaker GPU are just going to be something we all have to live with. There is no secret magic to make things substantially better, but at the same time there is no reason we can't have games that look and play great on XB1.

For that, I propose we try to minimize talk about these things, as none of this plays to XB1's strengths. We get it, it's a rung or two down on power. If people have fun with the games they play on it, who cares?

Games are supposed to be fun, and wherever someone has a big grin on their face, whether it be at 1080p on the screen, 2600p with some PC nut, or 792p on the XB1, more power to em.

That's why I didn't go off any CPU inprovement, eventhough I would assume it will also become more effective and efficient.  I was just looking at the GPU claim.  Any ideas how much the 360 improved?  If you compair crackdown to Halo 4, there is quite a bit of improvement.

Microsoft didn't want to go with DDR3, if you look at their old notes and plans it's always DDR4 memory.  But unfortunately, it's not ready yet.  They could have delayed the Xbox One for a year, but I think they went when they felt they really had too.  You can always wait an tech will get better.

I can't read anything on the slide you attached about a GPU side improvement.

What Mantle does it it improves framerates when you have a powerful GPU and weak CPU, because the CPU becomes the bottleneck.  I've never heard of any actual GPU side speed up per-se.

Also XB1 already has some features for multithreaded rendering, and pre-baking command lists so it already has a lot - if not all - of the gain DX12 will give.

It's also hard to say if CPU is the bottleneck in the XB1, since PS4 has basically the same CPU.

True, at some point you just have to get the thing out the door.

The 360 improved quite a lot, for different reasons. At the time, the tech they used was pretty exotic (shared memory, multi-core CPU, oddball architecture that needed all new APIs devved).

This time around things are very well documented and understood from the start, which is why we probably shouldn't expect too much magic down the line. Look at Forza 5. On a PC, a DDR3-based 7790 variant (which in reality would be about 7750 levels at best) wouldn't be able to play anywhere near that well.  It's a pretty fair example of how well optimized the API already was for the XB1 compared to PC DX11.x, as a PC would have needed something in the range of a 7870 or 7950 to hit those visuals for a similar game.

Well, a couple things, Microsoft has said the CPU was the bottle neck for consoles.  Might be spin but that's what they said.  I think they do believe that.  It has some different co processors and stuff that the PS4.  They are really good at squeezing power out of phones, writing popular operating system, business software, and games.  So they designed the Xbox One to do what they wanted it to do within a price.

And in the talk they said 20% GPU improvement on same hardware (or to that effect) - things would render and have 20% of the time left over – during the talk.  Wish I had the slides that were clear.   I get the other parts, like CPU enhancement not apply, but why would you think the GPU improvement would not apply to the Xbox One?

For the primary reason that the low-hanging fruit in DX11 (PC version) is the CPU bottleneck with regards to draw calls, which inhibits the GPU from working more efficiently. IOW, when you transition more of that to low-level API operations, the CPU becomes less of a bottleneck, and the GPU can work more efficiently at it's full capability.

The PS4 and XB1 have fairly weak CPUs. No two ways about that. Still, they pulled off Forza 5 (albeit with some downgrades, but still damned good work really) with a weak CPU and midrange GPU. The performance is a good bit better than a PC with similar specs would be capable of with DX 11, showing that what the XB1 architects said about the console pre-launch was true : that they already had optimized the XB1 SDK to allow for low-level hardware access.

Additional tweaks can be made, but it's not going to be like Mantle with a transition from bottlenecked API to low-level API. It's going to be going from a low-level API to another low-level API, with hopes of gaining some efficiency somewhere. At the same time, the hard limits are there with throughput, shaders, processing bandwidth, and so on. So perhaps they can go from 90% efficiency to 96% efficiency (obviously guesstimation, but you get the idea).

On the other hand, figuring out tricks, helping devs really get on board with optimizing engines to work with XB1, all that stuff can be great. They can figure ways where they're not technically moving more data, but because of how it's laid out and prioritized, it looks substantially better.