Digital Foundry Article Technical Discussion [2017]

Surprised they don't use more of that dual wield technology.

 

Maybe they do but it doesnt yield as consistent of results in other games, like open worlds or larger levels with more variances?

I hadn't seen a write up on it, but I imagine something like the XIM mouse/keyboard to console adapter layer but they have a USB broadcaster/splitter hooked up to all consoles at the same time.

What would be really cool is if they have ability to record and replay inputs. Or maybe thats all that this is, an input recorder and playback device. If they can record and playback then it could be nice to use when comparing patches on games. They could even use it when trying tweaks of the platforms like having downloads going in the background or playing background music or swapping out different external drives or even performing internal upgrades.

 

Overall running near same settings, 4Pro running higher settings on LODs and shadows.
1X holds 60, very little tearing
4pro about 55, tearing is a high occurrence
both use checkerboard rendering.

Overall, Richard feels the 1X isn't pushed. Fair enough. But this is what happens when you backwards patch games. They spent a lot of time and money on the 4Pro version, and probably didn't feel the need to invest many more hours into the 1X version unless it guaranteed sales. It happens, that's part of why these comparison videos are important, hardware is only half the story. I have doubts after this video that CM will go back to make more enhancements on 1X.

A year from now, I think 1X games will really hit their stride, but until then, I expect a lot more of these types of situations, as we've seen with other titles that pushed their 1X versions out to line up with launch.

 

A year from now there might be 1m XBX out of 100m consoles so they might care less without that MS push and MS money.

 

that's possible too.

It really depends on how much you're willing to invest to scale, if it's a little bit more work, then it's worthwhile.

ie. All consoles now support HDR (excluding OG XBO)
ie. All games would benefit from dynamic scaling (from 720->4K)
ie. Not all games would benefit from CBR necessarily, so this is a ?
All games could have enhanced feature set (especially if you intend to release a PC version)

So not seeing a hard lift here, seems like most of the important bits will be in the pipeline. It's all a matter of optimization.
PS4 will be lead, 4Pro and 1X will benefit from having the same architecture. API differences and tool differences will cause some optimization overhead, so it's not entirely that straight forward I guess.
XB1S will suffer likely here.

I also have no reason to believe MS has been funding companies to make X1X patches.
Perhaps the one's they marketed for (1P, Witcher, ROTR etc), but I doubt they can afford all the titles.
At the very least, if they funded the optimization, I don't think they would sign off on a result like this.

 

Pro has higher settings compared to X ?

How do you know they spent a lot of time and money on the Pro version ? Pro version is too good compared to X version ?

Wait, maybe there is some sort of bug on X like in some previous XBX versions ?

 

Please turn down the snarky attitude. This is a technical discussion thread, not some playground sandbox.

 

Up to 20% higher frame rate and no tearing (across the whole of the screen, all the time, no less).

It's clear the PS4Pro is being pushed beyond its performance envelope, while the X1X is sitting inside it. And that's despite lacking the hardware chequerboard support.

It would be interesting to know if any of the dips on PS4Pro are due to the CPU, to see it that extra frequency and those IPC optimisation are amounting to anything measurable on the X1X.

 

Game was released in August. 4Pro had been out nearly a year. X1X SDK kits clearly came out middle of 2017. At the earliest and it wasn’t available for everyone. So having seen X1X release in Nov, we see more developers have time to go back to work on the X1X version. It happens. Priority is should be the platforms that are already released and have a user base. X1X is still in its infancy. Logical business decisions dictate that it should be worked on the least.

It’s not a leap to make this logical assumption. We’ve seen a lot of developers try to put out X1X versions very quickly without adequate testing. I wouldn’t call it a bug. It just needed more time baking.

As for higher settings, yes they noted LOD and shadows. Not a bug. It is no longer a rare occurrence to see lower settings on X1X and then later patched again with proper settings and frame rate.

There’s nothing that 4Pro can do better than X1X. From a hardware perspective we see it, and from a benchmark perspective we have seen it. It’s an “ok” guideline to go by now given the delta in games where we are seeing over 100% more pixels and higher settings. I don’t think we are necessarily exaggerating it’s performance over 4Pro, though a good argument can be made about X1X often pushed past its limits as well trying to reach PC settings.

In this case we are seeing the same settings or less than 4pro. Checker board and all. A lot can be said about that, maybe the game can’t handle more? But we know the hardware so it’s just a matter of learning about how they coded it. I’d be tempted to chalk it up to just lack of effort.

As for how do I know they spent more money on 4Pro. Well time is money. It’s not like they allot so much money per platform. The more time their guys spend on optimizing it for they platform, the more it costs.

The game is checkerboarded and likely uses 4Pro tech. That’s an optimized solution compared to straight resolution. Willing to think that there are more AAA games today that have dynamic resolution over a checkerboard setup. And this is a pattern that speaks to the difficulty of a good CBR implementation and not a knock on 4Pro as hardware.

Since they used the same checkerboarding onto X1X which doesn’t have similar hardware it’s probable that they didn’t want to invest the time in making dynamic scaling run, which is likely the profile that X1X prefers (given the liberal usage of it in the other titles)

 

The shadows were not explicitly noted as an advantage on the pro only that the Pro has sharp almost stencil like shadows while the 1X has a more diffused shadow while looks better close while the Pro shadows look better in the distance. If there is a PC version perhaps it will be better to have a look at the shadows settings.

 

This is where you are wrong. This narrative is wrong. The facts tell us more complex results. From what I have gathered on plenty of sources the OK guideline is 60-70% more pixels on average. In some cases more, in some cases less, in others rare cases, only 0-10% more pixels even using a dynamic resolution (seen in some scenes on COD WW2 and Project Cars).

If you all wait for 100% more pixels in all games you are going to be all disappointed and all gonna need plenty of those long explanations: "bugs" "engine quirks" "less optimized" "less money" "less time".

 

That may have come across as a translation problem Global. The “it is an OK perspective” I’m referring to is the sentence before, that the X1X wise perform better than the 4Pro in all scenarios. No one expects all titles to have a resolution gap of that size. This is caveated with the obvious, obvious being the developer had sufficient time to put the hardware to good utilization.

As for how close 4pro and X1X that’s great, and I think it speaks to how well when optimized sufficiently, 4Pro reaches its vision of being this high resolution HDR device at a lower price point. And as time goes on I expect ideally the difference wouldn’t be much more than CBR vs native. That would probably be an ideal situation.

 

Discussion on PR Statements moved to this thread: https://forum.beyond3d.com/threads/...or-how-i-learned-to-loathe-pr-spin-off.58143/

 

Also this is one of the rare games that is really optimized for Pro, devs previously stated that this game is using FP16 (with RPM) on Pro. So that's not an unsurprising result retrospectively (how close both games are). DF totally ignored that aspect in their video and explained the small gap by saying XBX isn't pushed. But the devs famously stated that the game would use FP16 on Pro. We know FP16, if used properly, can improve performance of games (even if small). How did they miss to make the connection with their own results ? Very odd incompetence.

Far Cry 5 is another game that is announced to properly use Pro FP16 RPM hardware. It's going to be another interesting comparison, particularly compared to another Big Ubisoft game, Origins that doesn't properly use Pro specific resources.

 

F1 is running at 60 hz with no tearing on X1X. There's no more low hanging fruit to be picked. It's got several Gigabytes of memory literally going unused along with tens of GB/s of BW (do the maths even for when there's a 20% framerate advantage).

I don't know why we'd be looking for it as proof of near parity between two systems that are clearly not within 20% of eachother in terms of pixel pushing.

 

Well, FP16 has nothing to do with this. This is just a raw parity-game. Nothing more. The x is really not pushed. no better Textures (all it would need is to use the ram), no higher res but a much smoother framerate.
FP16 won't even the 40% performance + 4GB memory advantage (+bandwidth). So it seems they invested a lot of time in the Pro-"port" but just let it run on the same settings on the xboX

 

It’s hard to know. FP16 is also for Wolfenstein and we don’t see near parity.

FP16 has its limitations as well, if the bottleneck is ALU, then I can see FP16 having a good improvement. There are some restrictions of course, and a lot of things need to fall in place for FP16 to be a real factor. Hard for me to see a whole game optimized with that in mind, PS4 and XBO don’t have FP16 so there must be some limitations to the optimization.

 

5fps higher fps is much smoother now ? Remember that the XBX is not locked, also has tearing so the hardware is also pushed to its limits in some scenes.

We have recent and old XB1/PS4 comparisons with higher fps difference during stress tests. Tomb Raider DE was running consistently ~20fps higher on PS4 during framerate drops and I think the 2 recent CODs run often ~10fps higher on PS4 compared to XB1 (with higher res and/or settings on PS4).

Well, Wolf runs worse on XBX in scenes featuring the flamethrower, as much as 10fps lower. So there's that.

 

X1X can use fp16, which helps in BW and register limited applications. It's only when ALU is the bottleneck that RPM gives an advantage.

We need a FP16 optimised game for both, where both are running beneath the same fps cap, with no other feature changes, to accurately compute the potency of the sauce.

Because currently all we have is a CB hardware supported, RPM using game on Pro dropping down to 50 where the X1X holds at 60 with masses of BW and 3+ GB of memory going unused.

 

Well we have that game with F1. In framerate stress tests the game drops at ~55fps on Pro with 59 or 60fps but tearing on XBX (meaning the XBX is pushed to it's limit). That's a 10% advantage for the XBX version in those scenes (not oddly those scenes that show the lowest gap have rainy conditions...)

In others scenes think the lowest I have seen is 51fps on Pro vs 59fps on XBX -> 16% advantage for XBX.

In this game, XBX has between 10% to 16% framerate advantage with same res and roughly same settings. Pretty telling, don't you think ? compared to allegedly 100% advantage in others games ?

Sure the extra memory is not used for better textures but the 6tflops should be used to produce faster framerate. And about FP16 (without RPM) I don't think it would help much XBX considering the bandwidth and GPU ALU at its disposal. On the other hand that's a feature that on paper should more help Pro considering its main bottlenecks compared to XBX: mem bandwidth and ALU.

And it certainly does help the first multiplat game that use RPM on Pro seeing the quite small difference between both versions. But we'll have better analysis with others games.