Digital Foundry Article Technical Discussion [2018]

[see colon]

When DF analysed performance differences between OG One vs X (linked above) they reached the opposite conclusion and RDR2 seems to have a fixed resolution on all platforms.

I was speaking in general terms about most games and specifically about S vs OG. Of course X is going to run with a different performance profile, that's sort of the point of the X after all.

 

[Recop]

What a strange turn this RDR2 discussion has turned into. We've got one of the best looking open world games to have ever come out, and in 4K too, with no virtually no pop-in , and everyone is focused on XBO One vs XBO One S?

I have to agree that i'm very impressed by the X version. With this recent example, the X performs far beyond my expections. It was also probably the case with the 60fps mode of FH4, but i didn't quite realize it. What a great piece of engineering. Moreover at a reasonable size + quiet. Before the DF analysis, i was certain that the game would not run at native 4K on X. I was 100% wrong.

But to be 100% certain, i would like to see the VGtech analysis because he seems to be more accurate than DF. But until proven otherwise, i consider it's native 4K on XBX.

 

[Jay]

I usually believe that devs choose the best option for their game given whatever constrains their working within.
But would a simple native resolution have been better for the 4pro?
It would have also been the easiest way, although people would've complained that they should have used cb.
If they patched it now, everyone would actually be happy

I agree the 1X does come away looking very good in all respects, I'm not sure anyone expected it to be native 4k without looking at sky (using a dynamic resolution)

 

[Silent_Buddha]

Yea I'm among them. But we aren't given stats on active consoles, only on sales.

I mean generally speaking, the life span of OG xbox cut short at 3 years, because shortly after it was just the S. They had major trade in programs for the S. The year following was the X, and they had major trade in programs for that.

I guess for me I don't see the point for them to expend the effort to do it. It's not a different version of the game, the only thing the OG can do is a frame rate test. That's about it. everything else should be the same.

Increasingly sales are less reflective of actual ownership. Is the console still working? Did the owner upgrade? Did the owner keep their machine or sell it? How many used machines are still working, etc.

User numbers are far more important, IMO. Regardless of whether a person has upgraded multiple times, had to replace their machine, or whatever, if they are playing is the most useful information.

NOTE - this isn't saying sales numbers are irrelevant, they'll still likely give you relative sizes of a console's useable install base (machines in use) just not anything resembling an accurate number.

Regards,
SB

 

[Deleted member 11852]

I was speaking in general terms about most games and specifically about S vs OG. Of course X is going to run with a different performance profile, that's sort of the point of the X after all.

That was a typo, I meant to write 'OG Xbox vs S'. The DF article is indeed about S - X wasn't out then. Subsequent DF articles that also compared OG/S performance highlighted more framerate differences, which is why I questioned why S was now the base.

 

[see colon]

That was a typo, I meant to write 'OG Xbox vs S'. The DF article is indeed about S - X wasn't out then. Subsequent DF articles that also compared OG/S performance highlighted more framerate differences, which is why I questioned why S was now the base.

I still contend that the difference between S and OG in framerate is probably less now than it was a year ago. It's more likely that there is a difference in resolution because most games ship with a dynamic scaler. Honestly, I think the biggest difference between the S and OG is that I feel the install times are noticeably less on S.

 

[BRiT]

Yes, the improved hard drive model in the S is a difference but not so much at the same capacity. Drive data density makes some improvements so theres more of a difference between Launch One 500GB and S 1TB, but not as large of a difference as Launch One 500GB and One X 1TB however there are improvements elsewhere that impact that comparison so its not genuine like for like.

 

[vipa899]

Impressed by the One X, runs at a full native 4k steady 30fps for the most. The pro with a less stable framerate at about half the resolution, DF suspects PS was the lead platform. X is a monster of a console.

 

[iroboto]

I usually believe that devs choose the best option for their game given whatever constrains their working within.
But would a simple native resolution have been better for the 4pro?
It would have also been the easiest way, although people would've complained that they should have used cb.
If they patched it now, everyone would actually be happy

I agree the 1X does come away looking very good in all respects, I'm not sure anyone expected it to be native 4k without looking at sky (using a dynamic resolution)

generally speaking effects, shadows, etc don't always or normally run at the same resolution at the full frame buffer so that's not always the best indicator to determine if a game is native 4K. If I understand correctly, most measurements are around polygons at specific angles looking for steps.

Anyway for most people it's very difficult to determine resolution visually like the specific number, but it's much easier to see that something can look sharper, your brain has a very acute ability to know when detail should drop off at a distance, and native 4K gets very close to that point.

 

[Tkumpathenurple]

Yup, very impressive on the part of Rockstar's technical ability and the X1X.

It makes the position of RT in the next generation consoles more interesting, because if 6TF, a relatively shitty CPU (whatever customisations MS applied to it, you can't polish a turd,) and 12GB of GDDR5 get you this game at native 4K, then maybe vastly more of those resources would leave a lot squandered?

 

[function]

Ditto. I recall the early DF performance testing where some games showed no framerate differences and others it was more pronounced, 1-2fps not uncommon, 5fps in one 60Hz (F1) up to 7fps difference. That's not marginal.

I suspect that games that show a noteworthy fps increase on the S would be those with uncharacteristically simple framebuffer arrangements and that also use little to no async compute. These aren't representative of modern AAA games, to the best of my understanding.

I would bet a large sum of money - maybe as much as £5 - that newer, more advanced games that mostly have to work from DDR3 for the phat buffers and async compute datasets show almost no difference on the S.

Game engines have moved in a direction that really punishes a system like the X1.

Interestingly, despite the resolution that RDR2 on X1 is running at - 66% of PS4 - it can still drop below it in performance. Meanwhile, 1X is running at above 6x the X1 resolution and at a more stable frame rate to boot.

I'd hazard a guess that X1 and X1S are within a gnat's nut of each other in terms of performance and that the X1 is under performing relative to it's GPU resources due to chronic BW deficiencies where it actually needs it.

I don't know where folks are getting the idea that there aren't many OG Xbox One's out there, it was the only option for almost three years before S was released. What happened to 34 months worth of OG Xbox Ones, did they evaporate?

Oh I think there are loads of OG X1s out there. I think it's still very important to MS and publishers too. I also think it would be nice to have both OG X1 and X1S in comparisons.

At the same time though, I get that resources are limited and that they are best put where DF can get the most meaningful content, and also that the OG is long discontinued and that S results are more valuable for anyone looking to buy a system.

I also think that in any modern, demanding AAA game the differences will be negligible.

 

[iroboto]

Yup, very impressive on the part of Rockstar's technical ability and the X1X.

It makes the position of RT in the next generation consoles more interesting, because if 6TF, a relatively shitty CPU (whatever customisations MS applied to it, you can't polish a turd,) and 12GB of GDDR5 get you this game at native 4K, then maybe vastly more of those resources would leave a lot squandered?

It takes a while for developers to optimize for another platform. Now that we're at a year in, we should be expecting bigger and better results from X1X. We started off early with some quick ports and sub native resolutions. But I think as time goes on we're going to see the X1X get pushed and optimized well. I'm expecting more of these types of results after seeing Horizon 4 and RDR2 knock it out.

And yes, this is also my line of reasoning for being able to sacrifice the silicon for supporting RT and AI. They will undoubtedly profile the X Enhanced titles, and ideally profile these RT titles and they should know what it takes to hit those performance numbers. For MS this shouldn't be a guessing game, at least not since the X1X came out. Over the past couple of years since Satya has taken over, MS across the company has been taking bigger risks and pushing innovation. Anyone following //Build could see they were completely deviating from their past. They continue to earn solid mind share in the tech side of things since Balmer has left.

I honestly don't know how many node shrinks are left before we hit a major wall, but I'd like to see more than 1 generation of RT. As much as I want to believe there will be innovation on the RT side of things, I think most of the innovation will be on the AI/Direct ML side of things. This is where all sorts of new algorithms will need to be made for real time what we can do with them, etc etc.

the hardware doesn't need to be in the same light as Nvidia, but tensor cores aren't exactly difficult complex processing units.

 

[JLinneman]

Now had a chance to review both videos, really interesting stuff. There is some hyperbolic commentary in places like there the claim the Xbox One S framerate "tanks" while the onscreen counter shows 28fps which is just silly.

DF keep referring to the Xbox One S as the "base Xbox" but isn't S clocked higher than the original Xbox One? Maybe there is nothing in it but why aren't DF testing the original Xbox One, have they covered this before?

Ha, well, this is a side-effect of the conversation style video. I had some data running live but not all so was taking from my experience. Normally I’d have scripted this and probably changed the wording a bit since it does sound a bit too harsh given the results.

Also, I simply wasn’t thinking about S vs launch One. I don’t have an original system in an easily accessible place and we agreed to move to the S as it seemed more relevant these days. I’ll keep this in mind and refer to it properly in the future.

 

[iroboto]

I don’t have an original system in an easily accessible place and we agreed to move to the S as it seemed more relevant these days. I’ll keep this in mind and refer to it properly in the future.

for all the posts we debated on the reasoning, that such a simple reason evaded us.

 

[Deleted member 11852]

for all the posts we debated on the reasoning, that such a simple reason evaded us.

I speculated the reason was simple but it seemed like as many people were denying a performance difference between OG and S performance, which was contrary to early DF tests and your post on your 'replacement theory' of OG Xbox owners 'upgrading to S' yet others are saying there is no upgrade. The Xbox owning-community here seem divided on whether S is an upgrade.

I voiced my curiosity and John has clarified so I'm good.

 

[Garrett Weaving]

Nice to see you posting here John, huge fan of your work on Digital Foundry, especially the Retro series, keep up the great work!

 

[Deleted member 11852]

Uh oh.. pixel counting groupies.

 

[Karamazov]

so does the game downsamples on PS4 pro and 1080p TV ?

My console is very quiet with this game, contrary to recent games like spiderman.

 

[novcze]

so does the game downsamples on PS4 pro and 1080p TV ?

My console is very quiet with this game, contrary to recent games like spiderman.

you need to enable system wide supersampling for that ... it's really weird approach, like there is no in-game support for 1080p output

edit: awaiting some comparison pictures from you ;-)

 

[Karamazov]

i saw comparison pics on resetera but was not really convinced !