The scalability and evolution of game engines *spawn*

Yeah you are completely right. I guess engines will take also storage to ram bandwidth into account, and lower the quality of the assets for the slower machines.

 

indeed. scaling to Lockhart seems fairly reasonable (as far as we can see for now; UE5 is debatable); not nearly as ambitious as trying to scale to Xbox One S and keep things relatively uniform across the platforms

 

Develop and deploy on multiple platforms, probably is not very mindbending. But QA is probably where you get into trouble. Bugs in frameworks, sdk's and hardware differentials showing up in practice.

 

Well put. I think the only area where Series S could theoretically hold back multiplatform games is the lack of compute. I recall when PS4 was announced and Mark Cerny was talking to everybody in 2013, he absolutely believed that more and more traditional CPU tasks would migrate to GPU compute in the future and that having more compute was critical. I don't know if this was really every a thing that happened this gen - I can't recall any detailed GDC presentations on this subject - or whether the more capable compute units in nextgen make this more likely nextgen but it's pretty much the only area where Series S looks a bit underwhelming.

But as Ive said before, I think on balance that Series S will probably be a more capable and balanced system for delivering 1080p/1440p games than either Series X or PS5 will for 4K content - subject to the final confirmation on Series S memory bandwidth. And I still expect Series S to clean One X's chronometer in terms of performance in most games.

 

Honestly now that folks have caught me up on the rather sad reality of MS "no gens" thing I can very much see the point of an XBSS given that most next gen 3rd party games will drop XB1 support asap and it will most certainly wipe the floor with XB1X. I do wonder though if MS might have to suffer some bad comparisons of early cross gen titles where XBSS may have more consistent performance than XB1X but a rogue low res texture in a screenshot leads to some bad memes.

Good Lord I thought I understood their strategy and it's not as if I'm not engaged with this stuff but so much of it is smoke and mirrors and needless foot shooting. Like why announce things that are quite generous to the consumer as some kind of platform initiative so that it sounds hollow when it sort of sneaks out that really they are policies restricted to your activities as a publisher alone? Why not stake a flag on "Hey this is what publishing should be" instead of sort of pretending it's a platform policy so that it looks like you're the bad guy causing confusion when others don't step up to the plate. At the beginning of 2020 I was quite impressed with MS message discipline but from when they first started their MS Directs the wheels have come off at an alarming rate messaging wise.

 

Yeah, everyone's plans for 2020 were thrown into a blender. Some came out more consistent than others. If rumors are to be believed, the Series X would have been released in August or September and the potential Lockhart in November timeframe. It's still entirely unknown if Lockhart will even be a released product! Maybe Lockhart just becomes an xCloud runtime target platform for average internet connections.

With neither side wanting to blink or flinch first, it seems like a real shit show for us enthusiasts watching so closely. I imagine the viewpoint from Joe consumer will be different than ours. I hate this slow drip of information with so much held back by both sides.

To get this back on topic, I think MS is trying to leverage the PC development process, where engines need to support scaling either statically or more advanced with dynamic self-scaling options after being roughly dialed in by devs.

 

It's perhaps not entirely clear what sort of fundamental impacts to the renderer there would be between a 4TF and double to triple that considering the gulf in resolution targets. Volumetric global illumination techniques scale volumetrically, so it might be moot.

IMHO, it seems somewhat unlikely for developers to still target 1080p for the next gen, although if they did, I'd hope they'd consider some bloody anisotropic filtering.

At worst, Lockhart gets a sub-1080p target, and the mass market doesn't give a shit because they're the mass market - not B3D denizens.

 

I'm really hoping for a 1440p target with downscale to 1080p, this tends to produce decent results. DF did a piece back in 2015 on the lack of anisotropic filtering in this gen's console games and later updated it with some input from a dev which explained its absence in some cases.

I remember reading the original article but didn't spot that there was ever an update.

 

As a early PS4 adopter, I looked jealously at the image quality of early Xbox One games because of proper texture filtering, even when the resolution was lower. 50% more TMUs, but worse textures.

 

Yes, it is important to actually enable different filtering modes.
Those poor Ps4 TMUs must have been fiddling their thumbs waiting for something to do.

 

That was a common mistake of API misuse it seems. I do not remember such cases from 2016.

 

I'm really curious as to how dynamic resolution will play out for Lockhart. In principle, could they leave the top end unbound? Would a jump between the top and low end resolution be too much? Also does Lockhart's reduced memory put the cap on top end?

 

I think that the fact that products (controller) is going out and referencing xss means that the decision has been made now.
Or they would need to rush through a DD xsx to compete on price with the PS5 DD.

Although would only be able to run a single instance of Lockhart on the xsx blade due to cpu requirements, the power savings alone would make that a reasonable target for mobile and sub 10" screens.
TV's and PC's can run the full xsx target.

The number of full xbox exclusives is going to be extremely small, and then a small subset of that would have problems scalling if any at all.
As you say they need to do it anyway for PC, even if they didn't, graphical elements to an engine is the part that is easiest to scale.

There's a huge difference between scaling for Lockhart compared to XO/1X.

 

use quarter size textures on your PC games and see if your loading times cut in half.

 

Apart from resolution targets it's not like PS4 Pro and XBox One X offered much more really.

 

Thought we were comparing the XB1X and the XBSS, not the vanilla last gen versions.

By all accounts, we have already factored in the new tech by setting the new 4TF as roughly equal to 6TF old tech.
CPU is a step up for sure, but I doubt that's going to do much if games are to run on Xbox One X/S too. Maybe more stable framerates.

Not sure why people are asserting the XBSS as having significantly more performance. 1080p isn't a luxury that the last gen consoles can't target. If devs don't do specifically target 1080p for XB1X now what's forcing them to do that in the future?

 

Exactly. Sometimes with few reduced settings.
This is the same for next gen XSS and XSX

 

So effectively what you are saying is that the XBSX will be impaired by Devs having the XBXS as the lowest common denominator. Great.

 

Sony's CPU is already doing that. LOL

Tommy McClain

 

Really? A mere 200Mhz of difference is doing that? *facepalm*