Speaker 3
i think it's like not really it's's, it's, you're comparing along different axes that's Apple and oranges to me, but like, I do think that having a display that you really like, possibly a larger format display or a display that can deliver really strong contrast ratios, good HDR experiences and good brightness. Those are all things that I'm kind of looking for. I think that can greatly improve experience of playing a game. And like, I would definitely say, you know, I would definitely prefer, let's say Alan Wake 2 on PS5 on a really, really good display that I liked a lot versus Alan Wake 2 with path tracing on a display that was kind of mediocre or poor quality. I think that's definitely the case. Um, so I think having that ideal display for you matters a lot and could possibly matter a lot more than each additional margin of performance. But I think there also is something is like a display that's good enough for your purposes. And perhaps like that, the margin between like a mid-range television that might be, let's say, mini LED based and the very end to mini LED or OLED display might not make that much of a difference, right? It's about finding that a bit of that compromise point. But yeah, for sure. I would, I would tend to agree actually that the, that your choice in display can make a, definitely a bigger impact on, on your gameplay than a, some margin of a couple of generations on your GPO. Yeah classic example
Speaker 1
of that from my perspective is if i go um on vacation which what do i take the switch oled at one point it was switch oled versus steam deck lcd and i took i just took the switch oled right the screen was definitely a big part of that decision of course things change when the Steam Deck OLED came along. Now I'll take that because you're getting more performance and you're getting like a really awesome screen. So yeah, screen is definitely part of the graphics experience, no doubt about it. As to whether it can be quantified in terms of the amount of GPU generations, that's a bit trickier, I think. Okay, final question. This one from Ildar Nurizlamov. One could say that by developing a bunch of proprietary technologies, NVIDIA is pushing the industry forward. But in some ways, it's also kind of sabotaging it. Let's talk about reflex. I think tech like this shouldn't be proprietary. Its approaches should just be part of graphics APIs and most games should be using it by default. Or am I missing something? It's a software thing, right? You have a video about Reflex, but it's still not clear what kind of magic it actually does. General terms like optimizing and reducing latency don't really explain the how, and why this can't be just the standard for all games. It would be awesome if you made a deeper, more technical video about Reflex, maybe with some guest devs, because right now it's a total mystery to me. Alex, what does Reflex do and how does it work? So there is a great battle
Speaker 2
nonsense came out of hibernation at one point to do a video about like when when When AMD tried to launch anti-lag and it was messed up, and he explained very, very well in that video how Reflex works and what makes it unique. And the thing that makes it unique is that it's doing like a whole bunch of things at once. One thing, it's capping the frame rate below your monitor's full refresh rate so that it never goes into VSync and then frame queuing, which occurs with VSync. Because the GPU is going to keep working behind the scene and queuing up frames, which increases input latency right when you get into VSync territory, which is at the max refresh rate of your monitor. So it prevents that. Then it also reduces the amount of frames that are being queued in general to just basically be like one so it's like double buffering almost in that aspect uh but it's basically just reducing as much on the engine side as possible by reducing the amount of max frames that can be queued up that's something you could technically do beforehand in your control panel, at least on the NVIDIA side, and then AMD eventually added as well, too, with this, like, I forget what they call it. They have these new names for it. It's like auto low latency plus or something. It used to just be like a max queued frame option. But that technically can cause issues in games, by the way. It can make your frame times more erratic, just so you know. That is not like something that is just only good. That's one thing Reflex does, though. And another thing that it does is it tries to prevent your GPU from reaching max utilization. The reason for this is because, is a little bit confusing, but there's like a back pressure component to having the GPU being completely utilized. It'll actually increase latency if it is. So it wants to also keep your GPU from maxing out. And in which case, if it is below the refresh rate of your monitor, it will automatically reduce your frame rate below max utilization in your GPU. So your GPU will be sitting maybe around 90 to 96% instead of being full 99 or 100. And it's doing these combination of things in real time to make sure like the automatic GPU utilization frame rate limiting is actually probably the magic sauce of Reflex, which makes it feel so good. And that is the one thing that is proprietary more than the other technologies. AMD had to come up with their own solution for that. I still haven't looked at Antileg 2. It deserves a look from us in a latency piece. But I presume it's going to be trying to do something very similar, which is the whole purpose of it. Otherwise, why would they make it? But that is something that is proprietary. And arguably, though, I agree with you. I don't think it should be proprietary. I think DirectX needs to not be so slow. And it's just like a snowy landscape. It is constantly like little bits are moving here and there, but it's like not advancing too much over time. The landscape looks very similar. DirectX feels like that a lot. And, you know, these are things when Reflex came out, Microsoft should have really, I mean, maybe they did. They should have been like, wow, you came up with an API to make it so that developers can easily reduce the latency in their game with a single click option. This should be part of DirectX. And I really think it should. I think that's with a lot of NVIDIA invitations. Maybe they're keeping it on for themselves because they want this proprietary money machine, essentially. But also, I feel like at that point, it should be really the, like, Klaxons should be going off at Microsoft. Hey, we need to really push forward on our side, on the API side to make a generic version of this then through our own research. Because man, it's like right now, we're only getting cooperative equity support, like maybe at the tail end of this year, I guess, which is machine learning directly used in the graphics API. 2025. NVIDIA was already doing this in 2018, right? Like, it's like crazy timescale lag there. So yeah, I do agree with you fully, though, that I really wish things like Reflex were just straight up DirectX and not tied to NVIDIA.
Speaker 1
Oliver, what do you make of this? I mean, it is basically somebody's got to drive standards forward, right? And it does seem to be used as a point of differentiation to sell GPUs, which is kind of like normal. But at the same time, you do want to see DirectX improving.