Except when you look at jaguar level CPUs being used on the PC with a dGPU you get this. What that is showing is that across 6 games a 28.125% clock speed advantage averages out at a 10.6% FPS advantage when CPU limited. The caveats are that this is done with 4 core Jaguar CPUs (not 8 as in the consoles), it is being paired with a 7970 which is substantially more powerful than the GPUs found in the consoles making CPU limited scenarios a more common occurrence and the API overhead and optimisation level is exactly the same for both CPUs. Even if you ignore those facts though it means that the 9.375% clock speed advantage the Xbox One has would translate to an average 3.6% FPS advantage in similar scenarios. Nothing to really write home about in the end is it? The only real scenario where the Xbox One will receive a decent performance advantage over the PS4 is the scenario where the Xbox One had more optimisation done.

The only games where the Xbox One is smoother are AC:U, which is a total mess and DA:I where it runs at 900p vs 1080p. Most other games the performance is basically the same on both and in a few the PS4 had both a graphical advantage and an FPS advantage.

It would be really interesting to compare the above CPUs in games that are running mantle, that would give a better picture of what the CPU bottlenecking looks like on consoles as their APIs are closer to Mantle in terms of overhead than they are DX11.

For once Ron I cannot disagree with anything you have said above. Minor correction on the CPU clockspeed though as it is 1.75 Ghz vs 1.6 Ghz but as you said it is a minor bump. I could maybe argue that the Xbox One ROPS will only be the effective equivalent of 17 on the PS4 if it is pulling the data from ESRAM. If it needs to grab the data from the system ram then it is also bandwidth limited but again this is a very minor thing.

In all to think the Xbox One is the PS4s equal in terms of horsepower is a delusional. Yes it has a minor CPU clock speed advantage which in a like for like situation will only yield the tiniest of advantages in certain scenarios. The issue is though at 1080p the GPUs in both consoles are too weak to really put the burden on the CPU so in all but the most edge cases games will be GPU bound not CPU bound. Further in those edge cases that are CPU bound you very rarely end up with 100% clock speed scaling, unless you are starcraft 2, in which case the 9.375% bump in clock speed would not yield a 9.375% performance advantage.

This shows that the Athlon 5350 with a 7970 can only outperform the 5150 with a 7970 by around 10 % despite the fact it has a 28.125% clock speed advantage. Now the 5350 is a 4 core Jaguar APU running at 2.05Ghz, the 5150 is a 4 core Jaguar APU running at 1.6 Ghz and the 7970 is a much faster GPU than what is in either the PS4 or the Xbox One so what conclusions can we draw.

1) At best a 28.125% clock speed increase sees a 21% increase in average frame rate. This occured in Company of Heroes which is an RTS game and these are known for being very CPU intensive just like 4x strategy games. On average the advantage is around 10.8%

2) The 7970 is a much more powerful GPU than the PS4 GPU and it would have been interesting to see if these were just as CPU limited with an R7 265 which is much closer to the PS4 in terms of performance metrics.

3) The Jaguar CPUs are miles slower than even a low end i3. This means to maximise the consoles use of GPU Compute is an absolute must on both pieces of hardware.

4) This means, on average in part CPU, part GPU bound scenarios which are the most likely to occur a 1% clock speed advantage results in a 0.384% FPS increase.

Based on the above in you would expect the Xbox One to outperform the PS4 by around 3.6% in part CPU, part GPU limited scenarios. How that averages out over an entire game, or an entire level though I do not know and it will depend on CPU usage throughout but I doubt any game, or any level, is CPU limited all of the time. It may be in specific parts but in general it will be GPU limited with the occasion part being CPU limited. This is also assuming the API overhead and optimisation is the same for both platforms and that is also unlikely to be the case. This gap might be a bit bigger in specific areas where the Xbox One API has less overhead than the PS4 API or where the Xbox One received more optimisation than the PS4 and the gap might be smaller in the reverse scenario.

Ultimately a 9.375% clock speed advantage does not mean much and will not lead to any meaningful advantage on the Xbox One.

Wrong.

@Tighaman: They are both GCN 1.1. I do not need to look at a diagram to calculate bandwidth you just do math. Memory clock * bus width / 8 = bandwidth in bytes per second. 5.5 * 256 / 8 = 176 GB/s. Sure ESRAM has no contention but it is tiny so even though it is fast it cannot do everything so the gpu will be hitting the slow main ram a lot.

SHAPE was confirmed as being for kinect and voice recognition. There is no advantage in the sound systems as they are both equal for game purposes.

One gpu used as 2 or 2 used as one is the same damn thing and there ia no way that is how the gpu is designed. It is a single 12 cu design. So you can use shaders to do back end work? I think not as it is a different stage in the pipeline.

You know next to nothing about this and you just spout the shit mr x media dribbles out. If you want to learn look at gcn deepdives done at anandtech as that is a good start. If you just want to believe fantasy land bollocks then carry on.

The Xbox One has 2 ACE units which has can store 2 commands each giving a total of 4 stored commands for compute.

Based on the die shots it only has 1 command processor for issuing rendering commands but if it does have 2 that is a waste of die space that could have been used for something else.

The PS4 has 176 GB/s of memory bandwidth, this drops to 156 GB/s of bandwidth if the CPU is using its full 20GB/s allocation. This is about the same as the 7850 and the 7870 with the shader and ROP performance being closer to that of the 7850 than the 7870.

The Xbox One DDR3 has 68 GB/s of bandwidth and the ESRAM has upto 190 GB/s of bandwidth but in most scenarios it tops out around 120 GB/s. The CPU can use upto 30 GB/s of the DDR3 bandwidth.

SHAPE is for kinect audio decoding and has nothing to do with game audio, both consoles are likely using TruAudio that is built into GCN 1.1 for their in game audio..

If it is so difficult to target multiple command processor then it would stop SLI or Xfire from working. The fact these technologies work just fine (most of the time) suggests that this is not actually a problem and if the Xbox Is based around a 2x 6CU 8 ROP design it just seems rather daft as that will not perform any better than a 1x 12CU 16 ROP design and would probably perform worse as you never see 100% scaling when going from single GPU to dual GPU. Further it would mean that you need to store each asset twice so each 'card' could use it which would be a huge memory hog. There is no logical reason, or evidence, to suggest this is the case and it would be detrimental to performance if it were.

You might read this stuff but it does not make it true, or likely, or logical. I am not telling you anything from a blog, or forums, I am telling you stuff based on the die shots of the APUs and my knowledge of GCN architecture.

You suck at maths.

60 is 100% more than 30.

30 is 50% of 60.

1.84 TF is 40.5% more than 1.31 TF

1.31 TF is 71.2% of 1.84 TF

25.6 Gpix/s is 88.2% more than 13.6 GPix/s

13.6 GPix/s is 53.1% of 25.6 GPix/s

Interesting thing about Pixel fillrate is that it is bandwidth limited in most cases. On Xbox One it can maximise ROP performance only if it is getting the data from ESRAM. If it is getting it from the DDR3 the actual real world difference is a lot more in favour of the PS4.

So you are saying that multi GPU systems running SLI or Xfire will not work until DX12 because of multiple command processors? Also you do realise that the compute command processor is in-fact the ACE unit, of which the PS4 has 8 compared to the Xbox Ones 2. Why waste die space with multiple contexts when you could use it for extra shaders. It is just a waste of resources and I do not buy it because the die shot of the Xbox One APU just shows a standard GCN config based on 14 CUs and 4 render backends. If you want to believe in fairy tails though then fine, go ahead.

I would prefer the later. 1080p improves the quality in every frame of the game. The extra grass, while a nice bonus to those who have a PS4 is not worth dropping the resolution for, especially when in other areas the games look almost identical.

With how it is the Xbox One looks anywhere from the same, to somewhat worse depending on where you are in the map. If they had chosen 900p the Xbox One version would look a bit worse all of the time, all relative to the PS4 of course.

Lets say you have a box that can store a certain amount of stuff in it. If you throw the stuff in the box haphazardly and you end up filling it up before you have put all of your stuff in can you blame the box because you did not bother to maximise the space?

This is pretty much the issue, the CPU can handle the workload but Ubisoft have not optimised the code to enable it to so it sits idle or it is doing stuff it does not need to because it has not been optimised, this will affect all versions of the game and hopefully once they do optimise it they can hit a much steadier 30 FPS on both the Xbox One and the PS4.

As has been previously stated many times the 9.4% CPU advantage the Xbox One has is not enough, on its own, to explain why it is getting a 20% increase in FPS.

I had this but I just logged into the profile attached to my US account and the date reset back to the US launch date and when I went back to my main account it was fine.

There is more going on than just the CPU speed difference here. The FPS is around 20% higher on the Xbox One but the CPU is only 9% faster on the Xbox One meaning that the CPU alone does not account for the difference.

The most likely explanation based on PC and console benchmarks is that they developed the game with Nvidia Gameworks and then ported it over to the Xbox One, fixed some issues (but not all) and then did a quick port to the PS4 but did not do any performance tuning.

In general the game is just a mess and it needed another 3 to 6 months of bug fixing and optimisation before it was released but Ubi wanted a next gen AC game out for the holidays so they released it in the currently unfinished state.