Will DirectX 12 Help the Xbox 1?

Yep all he does is that and when he gets own like when he quoted like 4 developers claiming the difference would not be much,yet games had a huge difference on launch resolution wise or frame wise he backtracked and claimed that a 12 7850 will not be a 18 CU 7850,when he also damn well knows that GCN on xbox one is not Pitcairn is Bonaire.

Now he quotes rebellion and when games still perform better on PS4 a year from now he would move to another crap stated by a developers who wanted to be polite or been apologist for MS.

Yes that same developer say the game could be 1080p without MSAAx4 which no xbox one game does by the way,The order on xbox one with MSAA 4X would be 720p if not less,as MSAA 4X would devastate the ESRAM.

Sure the 7950 with 32 ROP deliver better results than the PS4 it was several more CU and faster speed to.

I'm aware of The Order's 1920x1080p without MSAA.

TitanFall Xbox One has 1408x782 with 2X MSAA. Respawn has experimented with 1920x1080p+non-AA and 1600x900p +FXAA.

On the PC, the end user can select all of the above settings i.e. the ability customise the experience based on user's wants e.g. trade resolution levels vs AA levels.

You haven't debunk anything since AMD Temash APU is just a lower TDP profile for quad core AMD Jaguar CPU + 2 CU GCN combo. AMD Kabini APU is another AMD Jaguar + 2 CU GCN combo with higher TDP profile. AMD Temash/Kabini APU includes 4 ACE units and it was released several months earlier than PS4 i.e.

Jan 2013 http://en.wikipedia.org/wiki/AMD_Accelerated_Processing_Unit#Jaguar_architecture_.28Kabini_and_Temash.29

Didn't you know AMD Z-01 or C-50 APU is just a down-clocked E-350 APU? From http://hwbot.org/hardware/processor/fusion_c_50/ the fastest C-50 APU is clocked at 2Ghz which is faster than the official E-350 APU. Model names and code names are just for marketing suckers.

-------------

Like your Pitcairn vs Bonaire, your mistake is making the codenames to be a larger issue than the actual processing elements. Your not looking at the debate deep enough.

Bonaire XT (7790) doesn't have the option for faster ESRAM memory, hence it forever gimped at 96GB/s memory bandwidth.

Since PC's Bonaire XT SKU doesn't have 150 GB/s memory bandwidth option, the closest 12 CU GCN with the nearest 1.32 TFLOPS and 150 GB/s memory bandwidth is the prototype-7850 with 12 CU at 860Mhz and 150 GB/s memory bandwidth.

I have already stated Xbox One would need "tiling tricks" to replicate prototype-7850 with 12 CU at 860Mhz and 150 GB/s memory bandwidth setup. I have Rebellion's statements to back my POV.

-------------

I have already debunked Cow's 16 ROPS vs 32 ROPS mistake. 7950's 32 ROPS says hi.

I have already debunked your rookie PS4 bandwidth math mistake. Let me post it again.

Yeah and the Order is one of the best looking games ever made Titanfall look like average crap,so yeah with those visuals you can have MSAA 2X but yeah thanks for validating my argument not game as MSAA 4X on xbox one while running at 800p is a little higher than 720p with average graphics that they can achieve MSAA 2X which isn't quite the same either,hell i don't even know how FXAA which basically has no hit to performance will send the game into 900p,the xbox one has problems with ESRAM is to small and regardless of the trick developers may pull those tricks are always a downgrade,Crysis 3 runs on PS3 and xbox 360,yeah with a allot of tricks and cuts.

You are a joke quoting wikipedia which even you can edit..lol

Is not temash period and i proved that Temash is 1.4ghz with turbo mode max period,Kabini is up to 2ghz.

And your were wrong twice because been Temash or Kabini based wasn't the reason why the PS4 has 8 ACES which was your argument yo begin with since the xbox one has the same damn CPU and doesn't have 8 ACES so yeah you were WRONG admit it and move on..

Bonaire is smaller than Pictcairn and has 16 ROP,not 32 which you insist con saying that MS some how remove is not the same architectures is Bonaire and sites like DF which you give credit when it serve you best.

Microsoft already shown that their 16 ROPS at 853Mhz = 17 ROPS at 800Mhz can saturate their 150 GB/s memory bandwidth. Eurogamer already stated X1's 16 ROPS is not a major issue for 1920x1080p.

So Eurogamer is right when they claim the X1 16 ROP are no major issue for 1080p,but they are wrong when they claim the GPU is Bonaire,.?

The so call tricks are downgrades just like the developer on Beyond3D say nothing more nothing less Tilling will not make the xbox one magically stay close to the PS4,tilling has always been a problem and requires more work to.

So yeah the difference is here to stay and since launch the difference has been there and even beyond what i use to quote from Anandatech many times which you claimed it didn't represent the xbox one because of this or that well all fall down the difference is even bigger now.

LOL, you still play codename marketing games. AMD/NVIDIA loves fools like you. Radeon R9 270X's codename is "Curacao XT", when it's just Pitcairn XT rename with slightly faster memory.

AMD Temash and Kabini are the same design and the only major difference between the two of them is the TDP profile. Lower TDP profile dictates lower clock speed. AMD Temash and Kabini already sports 4 ACE units and hence the IP option exist for PS4. Microsoft didn't exercise that option i.e. X1 retained two ACE units per 12 CU ratio which is close to R9-290X's two ACE per 11 CU ratio.

From http://www.tomshardware.com/news/amd-kabini-temash-apu,20200.html

"AMD Expected to Launch Kabini and Temash in Q1 2013". Go ahead and edit tomshardware.com. LOL.

-------------

Unlike you, I'm not interested in codenames i.e. I'm interested in 1.32 TFLOPS GCN with ~150 GB/s memory bandwidth since this will set the upper limit for X1.

You didn't read Beyond3D post i.e. you missed the word "OR" in that post e.g. you either reduce the resolution or tile it. I don't see any conflicts with Rebellion's statements and your Beyond3D's re-post.

You can continue to post anandtech's link since PC's 7790 SKU will stay at ~96 GB/s memory bandwidth i.e. it's TFLOPS and memory bandwidth profile doesn't come close to X1.

Note that, we both agree that X1 is less than retail 7850 i.e. both the prototype 7850 and 7790 is less than the retail 7850.

Both the X1 and the PS4 have custom AMD chips built up from their catalogue of component parts. Both are using 2 Jaguar clusters but I would hazard a guess that the X1 is using GCN 1.0 and the PS4 is using GCN 1.1. I think this is the case because,

Why X1 GCN 1.0?

1) The X1 only has 2 ACE units which is the same configuration as GCN 1.0 parts

2) The X1 does not have TrueAudio afaik. It does have an audio co processor but it is a custom design

Why PS4 GCN 1.1?

1) The PS4 has 8 ACE units like Hawaii. granted the other GCN 1.1 chip only has 2 ACE units but it is a low end part.

2) The PS4 also has TrueAudio which is part of GCN 1.1

There is no real match in the PC space for either chip which makes sense as GCN 1.1 does not really give a performance boost over 1.0. Pitcairn and Cape Verde are perfectly good in their respective segments with Bonaire filling the gap between them nicely.

Xbox One doesn't have the option for 32 ROPS, hence it is forever gimped at 16 ROPS.

Xbox One doesn't have the option for sustained 150GB/s throughput for multiple high bandwidth consumers, hence it is forever gimped at 68GB/s + one high bandwidth consumer at 204GB/s assuming the data for that consumer can fit into 32MB of ESRAM.

Since Xbox One does not have 32 ROPS or the ability to sustain 150GB/s of throughput to multiple high bandwith consumers the closest GCN card with 1.32TF and 68GB/s of bandwidth is the 7770.

Now it is obvious that the above statement does not take into account the whole picture but what a 7770 to R7-265 comparison will give you are the scenarios where the ESRAM is not being effectively used. In this case the performance delta is 72% in favour of the PS4 surrogate which is inline with the differences between COD:G, MGSV:GZ and Tomb Raider:DE. Now if we move to a higher performing part like the 7790 the performance delta is 41% which is inline with ACIV:BF and BF4.

You might have Rebellion's statements to back up your POV but I have games and empirical evidence to back mine up. Until those statements become a reality all they are is an untested hypothesis. I think performance will vary between the above performance deltas of 72% and 41%. As the tools improve, and devs get better more and more games will be closer to the 41% delta than the 72% delta.

AMD has stated "semi-custom" for consoles.

Bonaire falls into Sea Islands GCN not in Southern Islands GCN i.e. there are minor instruction level difference that sets it apart from Southern Islands GCN. The changes are minor but real enough for system programmers. You are looking at the marketing level difference e.g. you didn't notice the new FLAT instructions nor the coherent system address space handling difference.

Xbox One's SoC also includes Tensilica Hi-Fi DSP IP i.e. the basis for AMD's TrueAudio.. http://www.design-reuse.com/news/33902/microsoft-xbox-one-cadence-tensilica-processors.html

-------------------

There's more to a GPU beyond the color ROPS i.e. VRAM bandwidth has to be shared with compute, TMUs, MSAA, Z-ROPS. You haven't factored in read and write ROPS workloads.

AMD ROP unit has multiple read and write ports.

Rebellion's statement has a specific criteria for it's 1920x1080p claim i.e. You didn't read Rebellion's statement.

From http://gamingbolt.com/sniper-elite-3-closer-to-60fps-on-ps4-may-be-a-bit-slower-on-xbox-one

“Well as close as possible to 60 frames per second on both the systems, it will probably be closer to 60 on the PS4 but it drops to around 40 and 50, but you won’t really feel or notice it. It may be a little bit slower on Xbox One"

PS4 = 60 to around 40-50 fps at 1920 x 1080p.

X1 = 60 to slightly less than 40 fps at 1920 x 1080p.

-------------------

@ronvalencia said:

There's more to a GPU beyond the color ROPS i.e. VRAM bandwidth has to be shared with compute, TMUs, MSAA, Z-ROPS. You haven't factored in read and write ROPS workloads.

AMD ROP unit has multiple read and write ports.

Yea, Vram bandwidth has to be shared, on the Xbox One you have a small pool of fast RAM which can be used for a certain task and the rest of the GPU has to share the 68GB/s DDR3 ram bandwidth, this will lead to a lot of bottlenecking. You can try and tile tasks with the ESRAM but you only have so much room to play with so ultimately your average sustained throughput for all tasks is going to be a fraction of what the PS4 is capable of.

Why do I need to factor in anything? I compared the 7770 to the R7-265 and that difference is similar to the COD:G, MGSV:GZ, TR:DE differences between the consoles, I also compared the 7790 to the R7-265 and that difference is similar to the ACIV:BF, BF4 differences between the consoles. That is empirical evidence that the PS4 is between 72% and 41% faster than the Xbox One.

Here Kojima stated that the PS4 and Xbox One power difference was minimal. As it turns out MGSV:GZ runs at 1080p 60 on PS4 with atmospheric modelling and 720p 60 on Xbox One with a standard skybox.

Where in that statement does it mention the Xbox One performance other than it may be slower on the Xbox One? Where has your X1 = 60 to slightly less than 40 fps at 1080p come from?

I will believe what rebellion say when the game is released and it can be compared, until then I will take the devs words with a pinch of salt as other devs have stated how similar the consoles are but when the games come out there are rather large differences. I bet the game will perform between 41% and 72% better on the PS4 than the Xbox One like the majority of other titles have so far.