[00:04] edition of the DF DirectX Q&A show. To quote Arnold Schwarzenegger from questions and they want them answered through a bunch of those questions, first of all, John Lynman. Hello. [00:16] >> Oh, that was a good one, Rich. I have to admit I liked it. And of course, Oliver McKenzie. Hi. >> Yes, always a pleasure, Rich. And I interesting Q&A with some very, very good questions. And so I have to [00:29] good questions. And so I have to congratulate our supporters on those. >> You've done it. >> Um >> but uh let's crack on. But first, let's take a look at this. This video is [00:43] sponsored by World of Warships, a free-to-play massive online multiplayer naval combat game. With over a decade of constant development, it has established itself as a premier tactical shooter with continuous visual updates [00:56] delivering impressive water effects, dynamic lighting, and changing weather conditions. You can command over 850 historically accurate ships across 13 nations, all designed using real historical documents and actual [01:09] blueprints from the first half of the 20th century. There's a huge variety in gameplay to suit your play style with five distinct ship classes to master. destroyers, cruisers, battleships, aircraft, carriers, and submarines. The [01:21] game receives ongoing monthly updates introducing new ships, events, and cosmetics. You can jump in solo or team up with friends using the Division system for highly tactical team-based battles. If you register for the first [01:33] time using the tracking link at the top of the video description and use the invite code impact force, you'll get the biggest new player offer ever. This unlocks 30 days of rewards, including premium account time, resources, and the [01:47] premium account time, resources, and the German premium tier 7 battleship Shanho 43. If you prefer playing on a console, the game is also available via World of Warships Legends. For more information, check out the link in the video [01:59] check out the link in the video description or the pinned comment below. [02:12] actually relates to um an interview that was given by uh Sha Leaden uh to the podcast The Trophy Room. Um basically making a very very simple comment about the future of consoles. He's saying that they should um uh instead of making a [02:27] PlayStation 6 that costs $1,200, they should consider making another PS5 for out a couple of the quotes here. The innovation needs to come not from making PS6 for $1,200. It's to make another PS5 for $300. Now, that would be true [02:42] innovation. This is the first platform that I've seen that has gone up in price over time. I mean, PS2 at the end of the PS2 era, it was $99. You could buy a PS2 for $99. Um, yeah. So obviously the cost of [02:57] consumer electronics and consoles and PC hardware in particular has been a hot topic um for our supporters and for basically everyone in the gaming space much the whole year if we're being brutally honest about it. Um Oliver, [03:11] this is a really interesting question. Um the question is the viability of it. >> Yeah, I I really question this idea. I guess just the idea that it's feasible [03:23] guess just the idea that it's feasible in the first place because a $300 PS5 would still need a couple of important spec points, right? It would still need 16 GB of RAM and it would still need at least probably a 500 GB SSD. Maybe you [03:37] could go lower than that. I don't think you'd realistically want to. I think 500 GB would would probably be the minimum you could go to. And it also needs a you could go to. And it also needs a chip that can handle PS5 games. um and a [03:49] chip that handles PS5 games. Obviously, the PS5 SoC does that. I don't know, maybe you could make a more modern, smaller processor to to possibly achieve that through some alchemy, but uh that wouldn't necessarily be much cheaper [04:02] process node. It would require a whole redesign, and who knows how that would go. Um, conversely, there isn't a really good path to shrinking the PS uh 5's SoC down to a smaller process node. newer process nodes are are pretty expensive [04:16] and the PS5 has a 256-bit memory bus which wouldn't necessarily miniaturaturize so well. So, I think it makes sense to continue to find ways to cost reduce the PS5 and to keep selling that system, but I don't really know how [04:31] feasible it would be to get that at $300. Alternatively, I kind of think the way you'd want to go with this is probably just going straight to the cloud where you can take a console that maybe has, [04:44] you know, 2% uptime in someone's home and bring it closer to 100% uptime in the cloud because obviously people use their PS5 systems at different times. That could be a good way to go. I don't know. But obviously the PS5 cloud [04:58] streaming experience is not very good. there would have to be, you know, a rise in new technologies to make that a better experience and to actually have buyin from users. But I totally take the point that, you know, a PS6 that costs [05:11] over $1,000 is not going to move the needle on making a wider console That's not good. And I would like to see a $300 PS5, but um I don't know if [05:23] all the signals are heading in the other direction in terms of pricing. And there's not too much meat to chop off a PS6. You know, maybe you chop down that SSD, like I said, maybe you simplify the design a little bit in terms of the uh [05:38] mechanics of it and whatnot, but there's not too much you can do with just the way the technology has gone. >> Um the the idea is pretty straightforward and has much to commend it, right, John? the idea that basically [05:52] at this point we probably don't need um innovation in uh you know more delivering more horsepower for a console. Basically we just need to make what's available out there. But what do you make of the whole idea? [06:07] >> I mean that's not that far off from the Nintendo strategy of 2006 with the Wii. part of the Wii, but the Wii was essentially just adapting GameCube [06:19] hardware, and that started the cycle of Nintendo essentially being a generation Nintendo essentially being a generation behind. Um, but I think things are different from that era in that the hardware is good enough. Uh, and I I [06:31] genuinely don't think we need more hardware right now. That's not going to solve anything. um the the the changes that have occurred within the console development space. I've said it before, it's down to [06:44] just manpower, development time, budgets, that kind of stuff. That's what's taking longer. That's what's eating up resources as far as releasing software. The hardware is good enough for now. I mean eventually obviously you [06:57] want to see evolution but I think the speed at which hardware was improving didn't really match the reality of what can actually be made. So it's a shame though that we're at this point where this actually sounds like a good idea. [07:12] hardware more affordable? I don't know how you do that. I mean AI basically screwed us all in this regard. The the ram goblin has come and there is no end to this in sight. The suggestion of the cloud is quite possibly the worst idea [07:27] I've ever heard. I hate that idea. If that comes to pass, I guess I'm out. Like, I don't want to play games via the cloud. Like, even if GeForce Now is interesting as an option, it's not it's not something I want to [07:40] engage in. And I know how internet is these days, no matter how fast you go, uh there are issues that still crop up. I face them enough that it it absolutely down and watch a film and you're using streaming and for whatever reason at [07:56] way you want it to work or there's some kind of DNS problem or some some server along the way is having a problem there's slowdown and it doesn't work right. It's extremely frustrating. You [08:08] don't want to deal with it. It would be so much worse for games and that's not where we should go. So, I don't even like suggesting this as an option because it's the kind of idea you don't want to put into people's heads. Not [08:20] >> right >> I I don't I don't like that. Cloud for gaming stupid. I don't do it. >> It's it's an option, but it's not it that's >> it should only be an add-on thing. [08:35] Something for convenience and specific use cases that it takes advantage of your existing library, but it should never be the only option is my point. I'm pretty sure I'd always want to play games through local hardware, even if [08:49] the cloud experience does advance considerably on PS5 to the point where it is more like GeForce Now. I like local hardware a lot and I I have I have a very strong affinity for owning my own local hardware playing games seeing it [09:02] out heat that's very satisfying but if you are looking to like if hardware does that we need to find a way to subsidize that across many users of that hardware [09:14] effectively then that's the way that you have to go I don't know but I ultimately personally want right >> let's take the take the supporter question here from Wadeki. Um, over the last three generations, consoles have [09:28] hardware perspective with Helix on the horizon basically being a PC and price cheaper, more youth focused box that ships simpler, cheaper to develop games, or is that space fully occupied by the Switch 2? With cost of living issues [09:43] deciding elections all over the world, is taking consoles into super premium price tiers a smart idea? Alternatively, could the home hardware be lower spec and fancy rendering techniques be handled by a subscription to AI or cloud [09:55] compute and streamed in only when needed? Um, right. So, I've got some thoughts on this basically to follow up on what you guys were talking about. Um, concept of a a $300 PlayStation 3. I mean, basically Sony has got to innovate [10:09] mean, basically Sony has got to innovate in a way that basically um uh overcomes the best of what you know um what they've done previously and what the people like TSMC are capable of delivering. It just doesn't really seem [10:21] particularly viable. Um cost reducing the PlayStation 5. I mean Sony has been doing it since 2020. You know, basically they've moved to a um um a more costefficient process node. Um there's been like slight reductions in [10:37] power consumption. Um they've reduced or rather they've increased and then decreased storage. Um there's only so much more they could actually do with an existing PlayStation 5 to to you know to reduce the cost and um as mentioned [10:52] earlier the pricing of the key components in this case being memory and storage um isn't really going to amount to enough to produce a $300 console. I think the other thing in terms of viability is the the fact we had a $300 [11:07] console. It was called Xbox Series S and um Microsoft lost money on it and it and it, you know, basically lost the the argument to the $400 PlayStation 5 digital. So, I just don't really see how this is all viable. Now, let's talk [11:23] this is all viable. Now, let's talk about the cloud uh just quickly. Um about the cloud uh just quickly. Um the problem there is very very simple. Um, imagine that we were in a cloud scenario right now that all of the Xbox [11:35] consoles and the PlayStation consoles were cloud uh derived. You know, you just had like a very very cheap receiver box um uh that that you or just an app on your TV. What's going to happen when Grand Theft Auto 6 comes out? The answer [11:49] is that there is simply never going to be enough capacity to run um Grand Theft Auto 6 at that point. any AAA game I would venture to suggest um we will just simply have uh too much of a demand on the infrastructure. So I just can't you [12:04] know if you can't guarantee service I mean John you've been talking about quality of service from a cloud >> uh device I'm talking about something more fundamental which is access if you can't access your games [12:16] >> it's it's a it's a complete non-starter. So we and Rich to add to that I mean not only if everybody's doing it there is the access problem right there in terms of like maybe you can't get access games that are multiplayer already have still [12:29] have issues with this today just getting on their servers and this would be so much more bandwidth but it also creates problems of again if you're in places where you want to enjoy this without internet or with poor internet and those [12:42] everywhere. >> Yeah. So I think the the answer to this question has to come through >> great over overprovisioning relative to then what you would do with that overprovisioning level presuming you own [12:57] all the hardware and it's in some low latency data center somewhere is then you run that hardware on other workloads when it's not being used for gaming and I think in this case although I mean not specific to PS5 but in general at the [13:10] of some form or another uh either using that internally or renting that out. Um, hardware that is is quite lucrative and actually you can run lots of interesting hardware and maybe with a number of systems multiplex together. There are [13:25] some interesting things you could do. I don't know how longive that phenomenon could do at the moment. Or maybe you alternatively, I don't know what kind of arrangements might exist in the future, but maybe there's some leasing [13:37] opportunity or renting opportunity during times of very high provision where you could dip into, you know, the AWS. I don't know how feasible that would be from latency perspective, but maybe they could cook something up. I [13:49] don't know. But yeah, overprovisioning is a is a difficult problem when it is a is a difficult problem when it comes to very uh heavy peak load. uh very very popular games, things like that. Um so yeah, you'd never you never [14:01] that. Um so yeah, you'd never you never really approach that ideal point of just neatly as you would in other cloud workloads. Yeah, I think the other thing of course is that if you're overprovisioning um uh [14:14] I would suspect certainly in the current climate that there would be uh competing workloads that would probably earn them more money than gaming there. I think the, you know, the ultimate problem with um uh the cloud [14:29] and it's kind of a problem that has lessened over time is that um Microsoft, Sony, whoever, they have no control over the infrastructure um beyond a certain >> So, you know, there could be a problem with your um uh infrastructure in your [14:46] It kind of relies, you know, nothing is going to beat um a cable connected to a TV in that regard. So, you know, the cloud, you know, I guess to a certain extent, [14:58] we've already seen a cloud in the form of like multiplayer servers and stuff like that. And, you know, we time and time again, they've failed as well. >> I just don't really I don't really see the the cloud as a future. Um and I [15:11] the innovations that Shaun Leen is calling for here um would be applicable not just to gaming but to every other consumer level uh electronics device [15:23] at the moment in terms of like affordability and whatnot. So, you know, if these innovations are going to be coming, I would suspect that they would coming, I would suspect that they would be happening anyway for other devices, [15:36] for other use cases. And I'm not sure, you know, that we're actually going to the previous solutions we've seen for cost reduction, which is basically silicon and memory and storage becoming cheaper over time, isn't isn't happening [15:51] anymore. And that's the sad sad reality of it. Um, yeah, so grim times. I mean, it's a fantastic idea, but ultimately um I can't see a way out of it. I can't [16:03] um I can't see a way out of it. I can't see a way to make this possible. Um and see a way to make this possible. Um and uh that's uh a bit of a shame. Um let's move on. Got a question here from Shetty 1982 and [16:15] it's kind of like adjacent to what we were just talking about with the cloud. Salutation Jets exclamation point since FSR 4.1.2 2 is now being utilized in series consoles along with QSSR, PSSR in the PlayStation consoles. Is there a way [16:28] to prolong these consoles with something like cloud RAM to allow for a quote unquote increase in RAM allowing for more memory to extend the lifespan of those consoles while RAM prices come down or is it impossible as usual keep [16:41] up the good work exclamation point? I guess um my point is on on um uh the interesting discussion to have with the power of the cloud because obviously [16:53] we've heard it all before um principally during the launch of the Xbox One. Um Microsoft realized they were sort of outgunned by Sony on the horsepower stakes and um the pivot was to talk about the cloud and all of the [17:07] opportunities that came with the cloud because that was an infrastructure plus point that Microsoft had that Sony didn't have and nothing ever came from it really. In fact, the the demo, which really did look quite impressive uh for [17:21] this point with some Gamescom footage that was on IGN, um that never happened. We got something quite different in the in the end instead. Um and I guess the only cloud situation we've got right now is multiplayer surface to to a certain [17:37] extent and of course the the limited streaming service. We've seen the concept of RAM being sort of uh added on via the cloud. The problem there is latency. RAM needs to be as low latency as possible. And of course there's the [17:49] whole concept of um uh quality of service in terms of the connection. So um yeah unfortunately it is impossible I think. Um Oliver is there is there any [18:01] sort of thing you know the the concept of the power of the cloud >> it's been mooted several times and in fact thinking back to that FCC leak the Microsoft leak they were talking about it even then. [18:14] >> Yeah. I mean, you could do some things that are relatively latency insensitive. Crackdown case, I think they were using it to compute like large scale destruction in um large multiplayer maps where [18:29] obviously you're going to need an internet connection anyways. So, maybe we could do that through the server. Or alternatively, in a lot of games, they will do their um online bot AI connections through the cloud through [18:44] some server CPU. Um that's important in some cases for performance reasons, important in some cases for anti-che reasons. Um, but beyond that, it's it's difficult to envisage. And in this case, you know, if we're talking about a [18:59] server connection, you're going to have an enormous deficit in latency relative to a local RAM connection, like an almost unfathomable difference. And then be dealing with even on a gigabit connection, you know, thousands of times [19:14] slower transfers. It just not really practical in that sense. So, I don't actually think there's a a really neat clean way at the moment to integrate a local and cloud workflow in such a way [19:29] that you get a really compelling enhancement for game visuals or graphics or certainly something that would allow you to offset the high cost of local hardware. I just don't really see it happening. I think you either go all in [19:42] happening. I think you either go all in on local or all in on cloud. Uh John um there's no 32x solution or or even a connect solution this time around for this generation to kind of [19:56] extend it right. >> Uh yeah I mean I don't think I think the way doesn't really make sense and doesn't you know and the idea of the question is doesn't make sense either because as you said latency is is key. [20:12] you're not getting like 16 nanoseconds or whatever out of an internet connection necessary for this to work. It's just not possible. And even if I I It's just not possible. And even if I I don't think you should use it for low uh [20:26] less latency sensitive tasks either unless that task can fail and it's no problem you know cuz like you just cannot fundamentally you still cannot rely on internet being always available in some select areas and where where the [20:43] luckier in that regard but that's just not the true for most people. There's you just never know what it's going to happen and you just cannot have a situation in which it's always reliant upon a connection like this. uh for [20:59] games especially I mean with video it still happens like YouTube randomly has buffering for no reason on a gigabit connection because of some some DNS server along the way or like some routing issues you know things like that [21:15] and that sucks but it's like okay I'll just wait you know or whatever but for just wait you know or whatever but for games that can break in spectacular ways I think and right now for multiplayer it's not a big deal cuz it's well I mean [21:28] break in the same way as actually like using it in a way that they're >> Yeah, I guess there is Microsoft Flight Simulator which is a very limited use case scenario because you're not going anywhere particularly quickly. So in [21:42] terms of like streaming that world and texture data in the background, you >> Yeah, >> unless your unless your Cesar achieves warp speed, uh it's never going to be overwhelmed, I don't think. Yeah. Yeah, [21:55] that the uh the the scenario is a kind of limited. Final thoughts, Oliver. Simulator, it can be prepared asynchronously in the sense that they know what general direction you're going in. They can load in that data. They can [22:09] do that reasonably well. Likewise for a video, you know, you can have a a two minute, three minute buffer in a video such that any short-term intermittent in uh connection issue that you're going to have with the server is going to be [22:21] overcome by that buffer that's giving you that video playback time effectively offline, right? um with a game when we're talking about rendering challenges, you know, even the the one or you know, even the 200 [22:34] milliseconds of latency you might have with the server or something is like a a talking about such severe latency challenges when we talk about actually contributing to the real-time rendering of each individual frame. it like the [22:48] the latency problem is is so much worse here that these kinds of solutions can't case of like you know I know in Call of Duty for instance they do like a little little bit of uh streaming of like uh texture data right to the to the game [23:02] you're in you know there's a there's a lot of latency that can absorb and then instance is just not showing you the high quality asset they have an asset to show you right so in that case and in the in the Microsoft Flight Simulator [23:15] use case Mhm. Absolutely. Okay, let's move on. A question here from Inabit Yo, I think I've got that pronounced correctly, but who knows? With the PS6 looking to be very expensive, albeit it cheaper than [23:29] being more powerful enough to deliver path tracing on games that PS6 can't. Do purchase, especially towards the tail end of the generation where although it's up in the air, we can expect the price to be more reasonable? I can just [23:44] imagine all the comparison videos making the PS6 look bad dot dot dot. Um, well, because, you know, I guess we've got an idea of the specifications for each of the consoles that have leaked. Um, but I don't really envisionage a scenario [23:59] where path tracing will be available on one but not the other because they're it's just going to be a case of like if you could imagine in the PC space one PC having um faster CPU and GPU than the other but not to the extent that [24:13] features get locked out. I also think um there's a a use case situation here. Now maybe plans have changed but my um impression of them of the Xbox project [24:25] helix design as it's been leaked is that it's not just competing against Sony. In fact, it's probably more competing against PCs and it wants to be um uh it PC almost. You know, it's got high levels of uh memory, which it's going to [24:41] need more memory apparently than PlayStation 6, but PCs need more memory. It's got um more capable CPU because it's going to be competing against desktop um computers with um dedicated CPUs. And similarly on the GPU side, um [24:58] the PlayStation does seem to be more of a traditional console style box in terms of its uh hardware makeup, but it is fundamentally using the same technologies. And in the past, we've talked about how you know things like [25:13] dynamic resolution scaling and certainly u with machine learning based upscaling that the the differences between different classes of hardware can be quite minimal. And it probably even be more so if developers are actually [25:27] tailoring the experience for each device. Um Oliver, do you think anything has particularly changed there? I I I don't know. I don't really see it. The comparison videos will be on YouTube which kind of like nullifies a lot of [25:42] typically. >> Right. Right. Right. I think that I mean the differences between the two consoles as far as we know both are RDNA 5based [25:55] both have Zenbased CPUs. There's a bit of a difference in terms of the uh CPU of a difference in terms of the uh CPU because PS6 uses Zen 6C where I think uh because PS6 uses Zen 6C where I think uh Helix is a mix of Zen 6 and Zen 6C. Um [26:09] and then Helix has 40% more GPU compute according to the leaked uh figures. I believe around 40% more GPU compute and then it should have about 20% more then it should have about 20% more bandwidth presuming they have the same [26:22] memory speed pack a similar memory speed because the Helix has a a wider memory bus according to the leaked spec. Um but because they have a similar GPU architecture and should have very very similar features you know I don't know [26:35] that path tracing would necessarily be be on the cards for Helix if it's not on the cards for PS 6. Um, my thinking is that if lighting was transformative, a technique like path tracing, it would also be present on PS6 because there [26:52] isn't this like fundamental gulf and acceleration like we see for PS uh uh PS5 Pro where some kinds of workloads are just much more suited to PS5 Pro. In difference between a lower-end GPU and a higherend GPU. effectively a GPU that [27:08] might in practice, you know, be 20% 30% faster doesn't usually spell a whole lot techniques that you approach that console hardware platform with, know, one console platform that is the minority, probably a minority share of [27:24] the of the console market and one that is is a is a significantly larger proportionate console market. That's my expectation at least. Um, I mean, we already have a microcosm of this currently on PS5 and Series X and uh, [27:39] you know, there's there's I mean, I can't think of a single example where a lighting system was was hugely different qualitatively between those two notionally a a significant power difference between them. So, yeah, I [27:52] at all. If anything, I just expect to see resolution differences because that's the simple, easy, cheap way to do things. maybe some frame rate with CPU limited performance, that would be an interesting area where there could [28:05] be meaningful performance differences that would be hard to overcome through changing graphical settings. But yeah, not not the kind of differences that this uh user is contemplating, >> right? Yeah, I mean I think you've hit [28:18] the nail on the head there. There could be and there probably will be resolution differences. Um, if um, Helix is targeting more of a PC market, you would expect it to be able to do things like address a 240 Hz monitor that maybe a [28:32] PlayStation 6 probably wouldn't be doing, that sort of thing. Um, so, you know, I don't know, uh, the the difference between a 5070 and a 5080 might, you know, that's it's basically about a a similar gulf in terms of [28:45] The question is the extent to which it'll actually be tapped into by developers and what the use case for these consoles actually is. Um, anything to add there, John? >> Um, I think this is kind of one of those [28:58] questions that's kind of getting lost in the weeds a bit about where gaming is at these days. And I just don't think any of this matters in terms of the success of a machine. And I also think that this coming generation, based on what we kind [29:11] of theorize, nobody's really going to be competing in the traditional sense. They'll be competing for time, but beyond that, look like a traditional console market where they're like directly competing [29:25] machine. We got the best graphics kind of stuff." Like, I just don't think that works anymore. >> And it's it's there's so many other factors that determine where people are going to go. And I think saying like, [29:39] not on the other one, I just don't think it would matter for most people like at all these days, it's so yeah, >> you know, like I really do not I don't think Nintendo's still out there proving it right now [29:54] >> with their Switch 2 model. They're sticking very closely to the traditional sticking very closely to the traditional game console uh like market and it's working for them. >> Mhm. To be fair though, I do think that [30:06] with PlayStation 6. you guys said. I don't think there's going to be like a gigantic difference, just think because of the nature of what they're trying to do next generation. [30:21] it's not going to be a traditional console in that sense anymore. And it's the same space. >> Yeah. So, >> I also think that what we mean by path tracing in a real-time rendering context [30:36] gets very fudged because what that term means in this context, especially when advertised to have path tracing like, you know, Phantom Blade Zero recently, I I I think that fundamentally the key rate tracing technologies that go into [30:52] making a game look like it does at its best will also be on PS6. You know, I that path tracing on PS6 or path tracing on Helix or path tracing on PC or what whatnot. I I I think the core of what makes a game look the way it should will [31:09] be on PS6 and will be on Helix. Maybe Helix will have a one or two little ray traced effects or something in the mix, but I I I think what this user is really think that's going to be be the case, right? It's going to be it's going [31:24] to be good on everything. M yeah interesting. I mean the other thing of course is that um um Microsoft has announced that FSR diamond will be used on project helix and you know the FSR project is also basically uh all but [31:41] a partnership with Sony with project amethyst at this point. So you know there's a lot of um homogenization of technologies across both consoles. It really does depend on what this horsepower difference actually means on [31:54] shipping games. And yeah, this is one of the reasons why we didn't really do too much in the way of pre-launch um specs comparisons between PlayStation 5 and and Series X because um uh you know ultimately it's the it's the final games [32:08] that that matter. Um I think it's the bottom line. Um which kind of leads into the next question thinking about it, but let's move on. Okay, question here from Scud. Hey chaps, I'm curious if FSR4 and Xbox is going to make console [32:21] you again. We might not have any new consoles for a bit, but I'm guessing comparing PSSR, QSSR, and FSR4 might things might might make things more variable on a game by-game basis. What do you reckon? Well, it's quite an [32:35] interesting question, isn't it? I think we've kind of discussed uh the topic before within Direct, but um let's sort of crystallize our current thinking on this. I mean, QSSR came out um John, we took a look at it and um it was good um [32:50] in Wolverine at least. The differences were far less pronounced in Ghost of Yote, but fundamentally it's a machine learning based upscaler on a console that doesn't have accelerated machine learning functions. And um it it looked [33:04] kind of good. Now we've got FSR4 coming to Xbox and um any kind of like advantage that we might have uh suspected PlayStation might have in that regard is probably going to be uh obviated by this because you know [33:18] technologies. This is another one of those wait and see scenarios, isn't it? We got to see shipping games compared on both. >> Correct. Yeah. We simply don't know yet. But it does have the potential to make [33:31] it more interesting from our perspective because there could still be some minor differences or even major differences. Who can say yet? But that's kind of been one of the issues as far as like why platform comparisons have been largely [33:45] uninteresting. I think is because they usually just use the same technique across the board. And Xbox Series X and PS5 are very similar hardware-wise, so you usually don't get much of a difference there. And this is this has [34:00] don't think it's going to be necessarily game-changing, per se, but it will potentially give one version an advantage over the other in a way that we've not really seen before. even though they are fundamentally derived [34:14] from similar technologies, uh there should be some differences. So, I'm interested to see what they do there. >> Yeah, absolutely. Um Oliver, >> my thinking on this is that because QSSR, PSSR, and FSR4 and Xbox Series, [34:32] they're all based in the same neural network. They're all relatively similar to FSR 4 in some key respects. Um, I am not sure that we'll see very large differences between them. I could be [34:45] wrong on that. I think the key differences will probably come down to perhaps how they're accelerated and how fast that ends up being on the consoles target. >> That to me is is more important. Now, [35:00] there that we could observe in different titles? I would expect there would be. Um, but I also think that as upscalers get better, they're converging on some shared image quality characteristics and [35:13] it's getting harder and harder to spot differences between them that are differences, but differences that are meaningful to the end user. Like obviously, if we compare FSR3 and something like PSSR or DLSS, there [35:27] are huge categorical differences in how it handles different kinds of phenomena, how it's handling discclusions, things like this. But between these upscalers, I think they'll all be very good. Now, might there be one or two things we can [35:40] spot here or there? Sure. But it's not going to be like an FSR3 versus PSSR scenario where we're just like, yeah, one of these kind kind of looks bad and one of these looks good. It'll be they all look pretty good. So, yeah. And I [35:53] also expect some of the failure points of them will probably be, you know, descended from the same work on the same family of upscalers. So I actually expect because there is this homogenization here actually effectively [36:08] I expect that we won't talk about upskilling as much is my expectation because the upscaling will all be pretty good and it'll be pretty similar and we that's that's my expectation but maybe we'll spot something that's more [36:21] exciting than that. I don't know. >> Ultimately it does come come down to the quality of the games right and we haven't seen any yet. We've seen QSSR on two titles. I mean, an advantage I hoped that QSSR would have would basically be [36:35] that um essentially the code path for PSSR is is used. And so I was hoping that, you know, the last couple of years worth of um PlayStation 5 games um that had pro versions that use PSSR, hoping to see a lot of retrofits of QSSR to [36:51] older PlayStation 5 games. We've not even seen a single one at the moment beyond the Sony games. um which does kind of suggest maybe um QSSR doesn't work with older versions of the Sony SDK. Um it's sort of bizarre to me why [37:06] we haven't seen at least one or even a third party QSSR game at this point. Um so yeah, that was one advantage that I was hoping to see, but it quite doesn't seem to have come to pass, which is a bit disappointing. third party games, I [37:21] guess, moving into 2027. Um, we we'll start to see what's happening there. And I'm looking forward to seeing what it all uh looks like. Um, but let's move on. Question here from Taken 86, straight to the point. Now that we know [37:34] that DLSS frame generation runs on 20 and 30 series, why has Nvidia not provided frame generation for switch 2? So to put this question into context, So to put this question into context, basically all of this sort of um reverse [37:47] engineering that's been going on recently, it has seen um Nvidia recently, it has seen um Nvidia technologies move to older versions of technologies move to older versions of um RTX cards and um frame generation, [37:59] multi-frame generation even. I believe it's been shown on 2030 and 40 series. Uh there was a hardware unbox video on it the other day which was pretty good. it the other day which was pretty good. Um so um interesting question. Why has [38:11] Nvidia not provided frame generation for Switch 2, I guess, ultimately, Oliver? Um, there's a a question of computational cost, there's a question of politics, I guess, as well, you know. Um, what do you think? [38:23] >> Yeah. And there's also a question of resources within Nvidia to deploy to such a project. I don't know what those would look like, but for sure there's a question of of politics, I suppose, because this isn't supposed to be [38:35] officially supported. The question of cost is the most interesting to me on this the other day because I was thinking about this a little bit myself and I think it' probably be like I don't know 8 milliseconds or 10 milliseconds [38:49] in 1080p dock mode roughly in that vicinity based on projections from other ampier cards. So I think to make this truly useful you'd want a frame generation light. Maybe they could make a a frame gen light for Switch 2. That [39:03] they were going to do that, maybe you'd want to hold off on that before deploying it widely. But I think there's maybe a a bit of a pressing imperative for someone to do something about this because developers have taken a liking [39:17] to using FSR3 frame generation on Switch 2 lately with the likes of Dragon Stockma 2 and Stellar Blade, which I covered last week. And the results there covered last week. And the results there actually to my surprise not too bad or [39:29] at least when you compare it to the alternatives maybe reasonable. I don't know. seems all right to me. Little bit of extra input lag, but smoother frame rate, 60 fps, whatever. It's not the worst thing in the world, certainly for [39:41] otherwise doesn't have the CPU or GPU hardware to push that uh that higher hardware to push that uh that higher refresh in most of those titles. So, I it. I think Nintendo should do something about it. Um, but maybe they are holding [39:55] least for something a little bit better because I think they could probably do a little bit better than FSR3FG. Maybe quality um quality is okay there but but certainly frame time cost is significant [40:08] >> right okay um yeah it is quite interesting that FSR3 actually turns out to be in any way viable on the switch GP GPU certainly compared to a potential millisecond cost >> but then again thinking about it on the [40:23] PC side of things FSR3 fame generation did seem to be faster than DLSS >> um which which certainly is quite interesting I I think the bottom line is that the reason it probably wasn't there, first of all, um it would have [40:36] required a good degree of investment to get something operational that runs in a secondly, there's the question of anticipated need. Um you know, did uh Nintendo, even Nvidia anticipate that people would want frame generation on [40:51] Switch 2? Uh they presumably took the call that it wasn't. Um any thoughts on this, John? Yeah, I'm still not a big fan of using frame gen at low frame rates. I think it could be interesting if you're taking this Switch 2 game from [41:04] if you're taking this Switch 2 game from 60 up to 120 or something like that, but 60 up to 120 or something like that, but uh I I watched Oliver's video and I don't I see where you're coming from, but I just don't think it it works that [41:16] well in this case. There's just enough artifacting and the extra leg is enough where it just it's not a great solution. And if they could come up with something better, that would be nice. But I still think it's not I don't think there's any [41:28] magic bullet to make like a 30 fps input feel good with frame gen. >> So it's it's just unlikely. >> It's a land of I think bad alternatives of something like Stellar Blade, I think they could have gone a step further and [41:45] just done a a 40 fps targeting or 30 fps tarting mode. that would have been betteract in a lot of respects. >> In a game like Dragon's Dogma, they do give you the alternative of 30 of a 30 fps mode, albeit a poorly frame based [41:58] fps mode, albeit a poorly frame based 30fps, which has its own problems. So, they're given the option of a frame generation mode that they're given an alternative that looks like a locked 30 [42:10] fps or 40 fps with consistent frame pacing or as close to locked as is reasonable given the game's uh technical underpinnings. But yeah, it's it's not a good solution, you know, as compared to something like [42:24] a nice native 60 fps Nintendo game. But with the software that they're dealing with, you know, it is a way to hit 60 fps. And some of the compromises aren't as offensive to me as they would be on on PC, you say. [42:37] >> Right. A lot of people on handhelds, PC handhelds, you know, have used various frame generation techniques. Uh, and they're quite happy with it. Um, it's basically the the question of perception and and user value on those features. [42:51] Um, I'm sort of I don't know. I I did play the Stella Blade demo on Switch 2. know, I can see that there are going to be some games. I mean, thing about >> That's right. >> It's also just dropping dropping frames [43:07] all the time and also dropping frames in all the time. Like it loves to start throwing in 33 millisecond frames like in that boss encounter I showed in the video where it's like constantly dropping to 33 mills. You spin the [43:20] milliseconds. There's a big explosion in front of the camera, dips to 33 drops in the true sense, although it has plenty of those, too. It's just dropping >> that looks bad. >> I get that it looks bad also if you try [43:36] to interpolate that. But please just do it cuz I want it to still update it 16 milliseconds, right? >> The way stellar blade works, I think I'm extra harsh here because specifically the way Eve is designed, she has very [43:50] specular heavy clothing, right? and her hair is all over the place and she got those transparent armbands and like all that stuff together moving at a really high speed. It just creates this like mess of pixels that looks really ugly [44:04] and then those things like you mentioned where it can't it reminds me of forcing a frame interpolation on a TV when anything gets too fast or close to the screen and it basically drops back to the lower rate and those inconsistencies [44:20] drive me nuts. So, I really don't think so. This could potentially work on a much slower game. I think that doesn't require high input. I'd be more okay with it there. I would say [44:35] video that I uh was quite intrigued by, Oliver, which is that when you were mostly above 30, right? >> It was like 35 to 45 FPS, roughly [44:47] speaking. >> Um, right. I think that a 40fps lock could work okay there. There would be some dips, but I think it'd be better than 60 fps. Or alternatively, I think the game is pretty much rock solid at 30 [45:02] >> as well. I guess the other thing I was I was thinking of is that you actually had to have a resolution drop to accommodate frame generation even though your frame upwards. So um it does kind of suggest that even [45:18] uh e you know even um in this scenario the the cost of frame generation is significant enough that they couldn't run um stellar blade at the same internal resolution. They had actually had to drop it even though [45:31] >> the non-frame gen version is 35 fps upwards. >> So it it does kind of suggest that FSR3 is quite heavy on the Switch 2. Anyway, I think we've played that one out. Let's move on. Final question of [45:45] the show comes from Gregory Hobson. The DF time machine shows you what you were doing one year, 5 years, and 10 years ago. Um, if you could work forwards in time, what would you like it to show you doing 1 year, 5 years, and 10 years from [46:00] >> um I'm going to go to you first on this one, John, but I think ultimately if future, uh we probably would just shut up shop on DF and do something else. >> bank back bank this predictive technology and uh become [46:16] multi-billionaires. Anyway, um let's sort of embrace the conceit and uh what one, five, and 10 years? >> In one year, I predict we'll be playing >> In one year, I predict we'll be playing GTA 6 in the PC. In five years, we'll be [46:29] playing the enhanced version of GTA 6 on the PlayStation 6. And then in 10 years, we'll be playing Red Dead Redemption 3 while waiting for GTA 7. of GTA 6. >> That's true. There might be the double [46:43] enhanced on the on the next next generation. That's true. >> well, in 10 years, I think John should do a GTA 6F retro cuz at that point [46:56] >> Oh, yeah. that point it will be a retro game, right? Cuz 10 years is the is the >> By those rules. >> Absolutely. >> Wan, crazy stuff. >> I don't think we really want to be drawn [47:08] year, 5 years, and 10 years cuz we're also in our minds adding one year, 5 ages, which in my case isn't a particularly uh uh enticing scenario. [47:21] certainly still going to be doing this sort of job in some way, shape, or form. maybe I would have aged out in terms of like the the the video stuff, but um the reason I've been in this game for so long is that it's it's exciting and [47:35] unpredictable and and and intriguing and fascinating and the stories are just uh just non-stop. They just keep on coming. It's never a dull moment. That's why I business. Mhm. >> Um but in terms of like you know actual [47:51] content beyond um various editions of Grand Theft Auto 6, well I guess you know to a certain extent it's beyond u beyond prediction. Uh but it's certainly an interesting question to end the show. >> I guess Rich I I also thought maybe in [48:04] 10 years we'll have our staff will consist of us plus 20 Oliver bots made >> Only 20 >> agent just TW. You don't need to go overboard Oliver. We can get away with 20 Oliver agents just doing doing stuff. [48:20] >> We wouldn't be able to afford the memory >> for 10, I don't think, or even two. I mean, who knows? >> Right. Okay. That was the final show. So, as usual, like, subscribe, share, all of that stuff. Ring bells for [48:34] supporter program though, that is uh very, very serious. So, please do patreon.com/digitalfoundry. Amazing community. early access to DFDirect, early access to other things, um weekly updates from the team, uh [48:49] good stuff. Lots of other stuff going on there, too. But that's all from us on this one. Thanks for watching and supporting Digital Foundry. And uh yeah, I guess we'll see you in the next direct proper.