• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

NVIDIA DLSS 4.5 Dynamic Multi-Frame Generation and 6x Mode Officially Arrive

DukenukemX said:
But it actually doesn't. The image quality gets worse because there is artifacting and blurring, while also increasing input lag which means it should also feel worse.
Click to expand...
But it actually does. Sorry but on my system, seeing things actually running on my hardware, it looks better and feels good. I'm not talking about what it looks like on Youtube or pixel peeping on screenshots because I don't care, I'm talking about how it is in actual gameplay on actual hardware. It looks and feels better than lower FPS. The artifacts are extremely hard to notice outside of the UI, and those are very minor in Cyberpunk. Things look more clear and smooth when I track something across the screen with my eyes (like a car going past). If it is a game where the artifacting does look bad (Hogwarts Legacy was that way when I tried, though supposedly it has been improved) I don't use it.

DukenukemX said:
Going from 40ms to 60ms is going to feel like there's a rubber band in the controls.
Click to expand...
But I don't. It looks like it adds 5ish ms. I don't notice that. The controls feel the same with frame gen on vs off.

DukenukemX said:
Making a lot of assumptions there. First, is it the shit Youtuber guy? Here's another. Satisfied yet? Probably not. Second, latency is king. The reason people go for higher frame rates is because of lower latency. So in my opinion, going for a higher frame rather that increases latency is counter productive.
Click to expand...
I disagree, I don't think latency is king. I think lower latency is nice, and I'll take it when I can get it, but I think the visual smoothness is worth it even without the lower latency.

DukenukemX said:
Third, the extra frames aren't artifact free. You are losing image quality in the process.
Click to expand...
The quality I gain, in terms of smoother, clearer, image is more than I lose in terms of artifacts. Our vision is far, FAR from perfect, we don't notice everything.

DukenukemX said:
Also, it's not a hatred for new technology, it's a hatred for new tech that comes with down sides. If there were no downsides then I wouldn't have a problem.
Click to expand...
Nothing has no downsides, tradeoffs are inherent to everything. Name a technology, I can point out downsides. It is about weighing them. On a balance, I feel frame gen is a net positive on high refresh displays.

DukenukemX said:
To me, this is a telling sign that Nvidia doesn't want to make notable improvements to their graphic cards, and have instead turned to things like frame generation.
Click to expand...
Ahh yes, the old "They could make faster GPUs if they REALLY WANTED TO!" bullshit. Please then, with all your expertise, tell us what companies could be doing that would make GPUs with so much more raw horsepower than what they have now. Also tell me why neither nVidia, nor AMD, nor Intel chooses to do so if they could if they just wanted to hard enough.

Come on man, that's BS and you should know it. They are throwing everything at the wall to make GPUs as fast as possible. There's no magic wand to wave that would make GPUs faster if they really wanted to. If you think you have one then go ahead, make your billions doing it.

DukenukemX said:
This blame also falls on game developers and their lack of optimizing games. There should be no reason why modern games have such high frame latency to begin with.
Click to expand...
...base on what precisely? What makes you think that game engines could be so much lower latency? Is there evidence that engines used to be low latency and now they aren't? Are there engines out there that are both super low latency AND have fancy graphics? What is your evidence that is game devs just optimized harder they could make super low latency game engines?
 
Last edited:
I find it odd that an individual that obviously doesn't have an Nvidia card, and therefore hasn't experienced DLSS or DLSS+FG for extended periods in order to form an opinion on the tech, has such a strong opinion about the implementation based solely on second hand and often incorrect information.

And at this point in time, any claim made by the individual that they do in some way have an Nvidia card is largely doubtful.
 
Mazzspeed said:
I find it odd that an individual that obviously doesn't have an Nvidia card, and therefore hasn't experienced DLSS or DLSS+FG for extended periods in order to form an opinion on the tech, has such a strong opinion about the implementation based solely on second hand and often incorrect information.

And at this point in time, any claim made by the individual that they do in some way have an Nvidia card is largely doubtful.
Click to expand...
I believe said individual has claimed to be gaming on a Vega 56 fairly recently. :)

EDIT: Didn't look at said user's sig... I guess he has an AMD Radeon 6700xt too, now.
 
Last edited:
Mazzspeed said:
I find it odd that an individual that obviously doesn't have an Nvidia card, and therefore hasn't experienced DLSS or DLSS+FG for extended periods in order to form an opinion on the tech, has such a strong opinion about the implementation based solely on second hand and often incorrect information.

And at this point in time, any claim made by the individual that they do in some way have an Nvidia card is largely doubtful.
Click to expand...
I rarely look at someone's system in their sig, even if they have it so I didn't realize that but ya... Sounds like AMD owner sour grapes :p. Which is silly because hopefully AMD will improve their FG technology and have something on par with nVidia's. Right now it sounds like it has real issues with VRR which limits its usefulness and also means more latency if you need to use it with vsync. However, while AMD might not be good at pushing the boundries like nVidia does, they do usually catch up as FSR 4 has shown so I think there's hope that AMD's FG will get an upgrade, particularly as they start building more tensor math on to their GPUs.

You know Sony is going to want to work with them on a good FG solution for PS6.
 
Sycraft said:
I rarely look at someone's system in their sig, even if they have it so I didn't realize that but ya... Sounds like AMD owner sour grapes :p. Which is silly because hopefully AMD will improve their FG technology and have something on par with nVidia's. Right now it sounds like it has real issues with VRR which limits its usefulness and also means more latency if you need to use it with vsync. However, while AMD might not be good at pushing the boundries like nVidia does, they do usually catch up as FSR 4 has shown so I think there's hope that AMD's FG will get an upgrade, particularly as they start building more tensor math on to their GPUs.

You know Sony is going to want to work with them on a good FG solution for PS6.
Click to expand...
Agreed.

People need to understand that what we're experiencing in this modern age regarding gaming, graphics technology and performance is compounded by virtue of the fact that really capable hardware is, quite simply, too damn expensive and out of reach for most - To release a card capable of RT, not to mention PB RT, at native resolution with no DLSS and/or FG is quite simply cost prohibitive when the card will cost as much as the cheapest GWM motor vehicle. Therefore, technologies such as DLSS, DLSS FG, NV Reflex, and DLSS MFG are allowing users to enjoy RT and PB RT in a way that doesn't send them bankrupt - Which is in no way a bad thing considering such technologies are improving all the time.

Perhaps people are all pissy because their older AMD card quite simply can't handle ray tracing with any form of acceptable performance - period.
 
Mazzspeed said:
I find it odd that an individual that obviously doesn't have an Nvidia card, and therefore hasn't experienced DLSS or DLSS+FG for extended periods in order to form an opinion on the tech, has such a strong opinion about the implementation based solely on second hand and often incorrect information.

And at this point in time, any claim made by the individual that they do in some way have an Nvidia card is largely doubtful.
Click to expand...
If you've observed him for as long as I have, you'd know that he just blanket hates absolutely any proprietary tech in both the hardware and software side, and thinks that everything should be open-source and vendor agnostic. In a perfect world where there's world peace and everyone is rich and healthy, that would be great. And that is why he's running AMD hardware on Linux. But that's not how the world works and think that's the basis for him never conceding to anything and his unrelenting hate on tech like this and consoles.

The best thing about proprietary tech though is that it will eventually drive competition to deliver something competitive to it and usually a decent universal standard is born of it. Just like Havok was born from PhysX, which itself was eventually open-sourced. It just allows for the pioneers who developed it first to capatilize and profit from it until that happens, which if I were Nvidia or whoever making it, I would feel that is earned. Nvidia definitely deserves much of the criticism it earns as well. But you have to admit if not for them AMD and Intel wouldn't be working on their upscalers, RT performance, and other tech like FG nearly as fast as they are already.
 
Last edited:
Sycraft said:
But it actually does. Sorry but on my system, seeing things actually running on my hardware, it looks better and feels good.
Click to expand...
Kinda think you're biased.

Name a technology, I can point out downsides.
Click to expand...
Tile based rendering. The very thing that skyrocketed Nvidia Maxwell cards.
Ahh yes, the old "They could make faster GPUs if they REALLY WANTED TO!" bullshit. Please then, with all your expertise, tell us what companies could be doing that would make GPUs with so much more raw horsepower than what they have now.
Click to expand...
Crash the AI market.
Also tell me why neither nVidia, nor AMD, nor Intel chooses to do so if they could if they just wanted to hard enough.
Click to expand...
Because they make more money? That's a silly question.
Come on man, that's BS and you should know it. They are throwing everything at the wall to make GPUs as fast as possible. There's no magic wand to wave that would make GPUs faster if they really wanted to. If you think you have one then go ahead, make your billions doing it.
Click to expand...
They already do make faster GPU's, they just charge more money for them. The RTX 4060's have 8GB of ram and 128-bit bus. The 3060 had 12GB and 192-bit bus. Do circuit board wires and 4GB of ram really cost that much more, in pre-AI craziness?
...base on what precisely? What makes you think that game engines could be so much lower latency? Is there evidence that engines used to be low latency and now they aren't? Are there engines out there that are both super low latency AND have fancy graphics? What is your evidence that is game devs just optimized harder they could make super low latency game engines?
Click to expand...
Call of Duty Black Ops 6, Doom Eternal, and Cyberpunk 2077. I would even include Doom Dark Ages, though I wasn't a fan of the slight increase of frame latency.

View: https://youtu.be/b8QehJIgFOk?si=lli5NzWMqzZAWNwB

View: https://youtu.be/eXn574aUmjE?t=442
Mazzspeed said:
I find it odd that an individual that obviously doesn't have an Nvidia card, and therefore hasn't experienced DLSS or DLSS+FG for extended periods in order to form an opinion on the tech, has such a strong opinion about the implementation based solely on second hand and often incorrect information.

And at this point in time, any claim made by the individual that they do in some way have an Nvidia card is largely doubtful.
Click to expand...
This is a frequently odd statement. Particularly from here as I don't see this outside of this forum. I must own it and use it in order to have an option on it? Having reviews is just not good enough anymore it seems. Just wait for Hardware Unboxed and Gamers Nexus to review DLSS 4.5 6X. That's when the shit show will really start.
GoldenTiger said:
I believe said individual has claimed to be gaming on a Vega 56 fairly recently. :)

EDIT: Didn't look at said user's sig... I guess he has an AMD Radeon 6700xt too, now.
Click to expand...
Since 2024. Don't worry, I'm sure you'll point it out in the future as I don't plan to replace it anytime soon.
T4rd said:
If you've observed him for as long as I have, you'd know that he just blanket hates absolutely any proprietary tech in both the hardware and software side, and thinks that everything should be open-source and vendor agnostic.
Click to expand...
Yes, but that isn't the reason for my hate on DLSS 4.5 FG 6X. It's just a genuinely bad technology. I even have empirical evidence and that doesn't fly with you guys.
In a perfect world where there's world peace and everyone is rich and healthy, that would be great. And that is why he's running AMD hardware on Linux.
Click to expand...
You do know I've promoted Intel GPU's? But yea, being an AMD and Linux makes me a total loser. How's it go? Slander becomes the tool of the loser?

The best thing about proprietary tech though is that it will eventually drive competition to deliver something competitive to it and usually a decent universal standard is born of it.
Click to expand...
Everyone has Frame Generation. Intel has XeSS 3 Multi-Frame Generation (MFG) and AMD has AFMF. They all suck because they all have the same problems.
Just like Havok was born from PhysX, which itself was eventually open-sourced.
Click to expand...
What? Havok is older than PhysX. Havok was used in Half Life 2, a game released in 2004. PhysX was open sourced because better tools were released and replaced it. Havok was here before PhysX, and still here after PhysX.
 
DukenukemX said:
This is a frequently odd statement. Particularly from here as I don't see this outside of this forum. I must own it and use it in order to have an option on it? Having reviews is just not good enough anymore it seems. Just wait for Hardware Unboxed and Gamers Nexus to review DLSS 4.5 6X. That's when the shit show will really start.
Click to expand...
When your constantly posting second hand memes and video's that are blatantly incorrect or exaggerate some form of latency not present running DLSS MFG I'd say yes, some form of personal experience with the technology is required when you're so outwardly opposed to it.

Right now, your postings are no more than confirmation bias.
 
Mazzspeed said:
Agreed.

People need to understand that what we're experiencing in this modern age regarding gaming, graphics technology and performance is compounded by virtue of the fact that really capable hardware is, quite simply, too damn expensive and out of reach for most - To release a card capable of RT, not to mention PB RT, at native resolution with no DLSS and/or FG is quite simply cost prohibitive when the card will cost as much as the cheapest GWM motor vehicle. Therefore, technologies such as DLSS, DLSS FG, NV Reflex, and DLSS MFG are allowing users to enjoy RT and PB RT in a way that doesn't send them bankrupt - Which is in no way a bad thing considering such technologies are improving all the time.
Click to expand...
Not even just that, but flat out impossible. Like I have a 5090, which is way too expensive for most. Even it can't handle everything cranked up. If I take Cyberpunk, crank up all the nice Path Tracing and such, and then try and run it at native 4k with DLAA it drags ass. 30fps or less kind of thing for a lot of the game and that's with a GPU that draws 600w and is so large it is stupidly expensive. There's just no option other than things like DLSS. I have to use super resolution to push the base FPS up to a playable level. Balanced is what I settled on in this particular case. Then if I want nice smooth high FPS I need to kick on frame gen. There's just no other option, there isn't a bigger, better GPU for me to step up to and as a practical matter even if there was I don't know that I could. Like if there was a 1200w GPU that was twice as fast as a 5090... well I'd have to rewire my house to get a dedicated 20A circuit (or maybe a 240V one instead) because a 15A plug flat out couldn't deliver enough power to run a system with that, even if dedicated. So even for those of us who are silly and blow too much money on our setups, we STILL need things to improve FPS when it comes to the high end of rendering.

For FG there's also the other side of it that if we want to have ultra-fast monitors, which we DO want to have to completely eliminate motion blur and to give us a smooth, real-life, presentation we need it regardless of graphics because of CPU limits. That was the point of my Tempest Rising demo but you can find plenty of tests out there like that, Hardware Unboxed has done lots of CPU tests and for most games even with the best CPUs and in a non-GPU limited setup you aren't getting much more than low 200s of FPS, often less. So even if you are on the "screw high end graphics, I want FPS" bandwagon, you run into limits. You can find plenty of people talking about how if we want super high FPS it is going to have to be through some kind of frame generation or reprojection technology, it just isn't feasible to brute force it.

DukenukemX said:
Because they make more money? That's a silly question.
Click to expand...
What? AMD and Intel would 100% make more money if they were able to release GPUs that were faster than what nVidia has... but they can't.

DukenukemX said:
They already do make faster GPU's, they just charge more money for them. The RTX 4060's have 8GB of ram and 128-bit bus. The 3060 had 12GB and 192-bit bus. Do circuit board wires and 4GB of ram really cost that much more, in pre-AI craziness?
Click to expand...
Even those aren't enough to max out things with high end graphics. I have a 5090, it doesn't get faster than that. It is stupidly large, stupidly power hungry and stupidly expensive. Even then high end graphics at high frame rates are out of reach through brute force.
 
4saken said:
/me periodic check in to DLSS thread.
/me sees same 4 people going back and forth.
/me goes back to gaming with DLSS.
Click to expand...
Me too, me too.

slow-clap-hell-yea.gif
 
Mazzspeed said:
When your constantly posting second hand memes and video's that are blatantly incorrect or exaggerate some form of latency not present running DLSS MFG I'd say yes, some form of personal experience with the technology is required when you're so outwardly opposed to it.

Right now, your postings are no more than confirmation bias.
Click to expand...
We've been down this road before with Frame Generation. Gamers Nexus did a video on it, and that was at 2X and 4X. Both had image quality problems like ghosting. You telling me that DLSS4.5 has fixed all these issues? Keep in mind that neither Gamers Nexus or Hardware Unboxed had a chance to look at DLSS4.5 frame generation. I'm going by what one YouTuber found and it doesn't look good, and a small group of you, the usual group, are losing your minds over this. What are you going to say when Gamers Nexus and Hardware Unboxed gets around to reviewing this? Most likely Hardware Unboxed since Gamers Nexus isn't big into reviewing this kind of tech.

View: https://youtu.be/3nfEkuqNX4k?t=955
Sycraft said:
What? AMD and Intel would 100% make more money if they were able to release GPUs that were faster than what nVidia has... but they can't.
Click to expand...
Intel already does but nobody cares apparently. AMD has been intentionally avoiding to compete with Nvidia because... I really don't know honestly. Though AMD did have some good ideas to reduce cost by using LPDDR5 memory... that we're never going to hear about thanks to AI buying up all DRAM.
Even those aren't enough to max out things with high end graphics. I have a 5090, it doesn't get faster than that. It is stupidly large, stupidly power hungry and stupidly expensive. Even then high end graphics at high frame rates are out of reach through brute force.
Click to expand...
I blame UE5 and Ray-Tracing. Even a game like Doom Dark Ages is going to have problems on a RTX 5090 at 4K without DLSS. Doom Eternal can do it easily without DLSS, even with Ray-Tracing enabled. Notice the frame time of 3.3ms for this guy? Path Tracing just murders Doom Dark Ages performance.

View: https://youtu.be/pyPBRVypTtk?si=ZDKooysQ-JgX61U_
 
Well I'm going to call it here and stop responding because you seem to be unable to understand the principal thing I'm saying which is that to me, the game looks better and feels better with it on. Your argument seems to be "Nuh uh it doesn't." I like how games look with Path Tracing, and I like how those Path Traced games look with frame generation even better. Hence I'm going to recommend them to people who have the hardware to try it. Your argument seems to just be "Stop liking things I don't like."
 
DLSS really only works if your right on the edge of decent frame rates without it. But I am not a fan of anything beyond 2X, after that it gets noticeable input lag and artifacts start become far more noticeable. After that you are far better off to just turn a setting down that is killing performance on your card.
 
DukenukemX said:
We've been down this road before with Frame Generation. Gamers Nexus did a video on it, and that was at 2X and 4X. Both had image quality problems like ghosting.
Click to expand...
And I'm sure as a 'tech tuber' it got him a lot of views. I've been running FG, MFG and SM under a vast number of titles, and with the exception of in game HUD's under very specific titles I've seen no ghosting that would put me off the technology - and even ghosting regarding in game HUD's is so rare now it almost a non issue. If you consider LSFG, that implementation has artifacting on stationary screens like in game settings menu's that's immediately obvious.

I was 100% a naysayer of the technology until I actually made use of it, now I swear by it. If I can remove CPU bottlenecks without expensive hardware upgrades - that's a good thing. Once again, your comments are nothing more than confirmation bias when your whole argument is based on second hand video's and meme's.

Edit: GPU bottleneck should have been CPU bottleneck.
 
Last edited:
Mazzspeed said:
And I'm sure as a 'tech tuber' it got him a lot of views. I've been running FG, MFG and SM under a vast number of titles, and with the exception of in game HUD's under very specific titles I've seen no ghosting that would put me off the technology - and even ghosting regarding in game HUD's is so rare now it almost a non issue. If you consider LSFG, that implementation has artifacting on stationary screens like in game settings menu's that's immediately obvious.
Click to expand...
I think part of the misunderstanding from people who haven't used it at high frame rates is that yes it does have artifacts... but you don't notice them in actual use. There's a big difference between what you can see when you frame capture it and go frame by frame and look at them in isolation and what you see when it is being displayed to you at 200fps.
 
Sycraft said:
I think part of the misunderstanding from people who haven't used it at high frame rates is that yes it does have artifacts... but you don't notice them in actual use. There's a big difference between what you can see when you frame capture it and go frame by frame and look at them in isolation and what you see when it is being displayed to you at 200fps.
Click to expand...
Exactly. I'm seeing great FPS, things are happening faster than I can notice artifacts captured by tech tubers in still frames (gawd I hate tech tubers), and I can't notice a latency increase of ~5ms - So from my perspective as a gamer, it's a win all round.
 
Last edited:
Gideon said:
DLSS really only works if your right on the edge of decent frame rates without it. But I am not a fan of anything beyond 2X, after that it gets noticeable input lag and artifacts start become far more noticeable. After that you are far better off to just turn a setting down that is killing performance on your card.
Click to expand...
Your post indicates you confuse DLSS (Deep Learning Super Sampling) with FG (Frame Generation).
 
Mazzspeed said:
Exactly. I'm seeing great FPS, things are happening faster than I can notice artifacts captured by tech tubers in still frames (gawd I hate tech tubers), and I can't notice a latency increase of ~5ms - So from my perspective as a gamer, it's a win all round.
Click to expand...

This is basically the crux of it.

I have a 5090 and 240hz OLED.

Going from like 70 fps to 200 fps is a GIGANTIC improvement in smoothness. Absolutely huge difference.

There is straight up no noticable latency to me when the base FPS is solid like that. I've sat there trying to flick my mouse to see if I can feel any difference whatsoever - and I basically can't.

Artifacts are a complete non issue in 99.9% of gameplay. I think maybe I've seen a UI element slightly ghost once or twice or something weird. It just looks completely an utterly fine the massive majority of the time unless you're literally sitting there deliberately trying to move your mouse in a way that you can find something fizzling on an edge.

It's completely worth the trade offs in my opinion. It's a way better experience.
 
Gideon said:
Not confused at all, that's how Nvidia markets DLSS
Click to expand...
That is something I've been annoyed with them for a long time and it is only getting worse: They call EVERYTHING image enhancement related DLSS. Makes shit confusing. Their version numbering doesn't help either. They really need more clarity on their different technologies, but their marketing department seems to like to be "EVERYTHING IS DLSS!"

socK said:
There is straight up no noticable latency to me when the base FPS is solid like that. I've sat there trying to flick my mouse to see if I can feel any difference whatsoever - and I basically can't.

Artifacts are a complete non issue in 99.9% of gameplay. I think maybe I've seen a UI element slightly ghost once or twice or something weird. It just looks completely an utterly fine the massive majority of the time unless you're literally sitting there deliberately trying to move your mouse in a way that you can find something fizzling on an edge.
Click to expand...
They are also less of an issue the faster your base and final FPS are. It's a combination of two things:

1) The higher the base FPS, the less actual artifacts that get produced because the less difference there is between frames. If you try to take a game at some completely unplayable framerate like 15fps and generate frames there will be so much change between rendered frames you'll get a fair bit of artifacts in the generated frames. Similar to how you get more artifacts the lower the input resolution for DLSS upscaling. However when the base FPS is higher, there's less change frame-to-frame so the generation does a better job and there's less artifacting to be seen, even when you freeze frame and look at it carefully.

2) Every frame is on the screen for less time so your brain doesn't notice things as much. You have to remember you can't actually "see" at 200fps. Like if I presented you with novel images, each being different from the last, at that speed you wouldn't be able to process those images and tell me what you are seeing. Our perception is strange, not 100% understood, but it isn't something where we see individual images at a high speed. Rather parts of our vision update and blend with our memory of what we were seeing. Net effect is that at high frame rates, minor issues are less perceptible.

Hence why I call it a "win more" technology. It won't take 10fps and make it 60fps, or rather it will but the result will be garbage and unusable. Game will look bad and still be unplayably laggy. However it will take 70fps and make it 210fps and it'll look much smoother and still feel good.

It's also why you need to be extra careful with looking at frame grabs. Not just because seeing the image static is not the same as seeing it as part of motion, but because often those are done at 60fps since that's easy to capture. So you limit FPS to 30, turn on 2x FG, then capture that. Works fine and easy to do... but you are going to get frames with more artifacts than if you were capturing from a higher base FPS.


Speaking of stills being different than in motion, I'd put Ray Reconstruction and to a lesser extent Path Tracing in that same category. When you see still image shots, they aren't always super impressive. Like I'll see a back and forth wipe shot and it's like "Ok that looks good, but is it that big of an improvement?" With RR in particular you can notice a loss of texture detail if you pixel peep, particularly compared to the new 4.5 DLSS models. However when you fire it up in game, it just looks better. When you see the light and reflections interact with the environment they just look so much more correct. It is a thing you at least need to see in a video, and really I find experience it in actual gameplay, to truly go "Oh ok, I get why it matters, ya this is awesome." It isn't about how it looks in a screenshot or what you can find when you peep pixels, it is about how it looks when you are playing the game.
 
DukenukemX said:
Not a review on DLSS 4.5 Frame Generation but he does talk about how it's not really a performance increase. Also yes, the Lego game require FG to reach 30fps.
Click to expand...
I don't get how companies can make games that garbage, particularly with simplistic graphics. UE5 isn't a reason either. Satisfactory is UE5 with reasonably simple graphics and it runs quite well. Even with unoptimized software lumen (the devs allow you to enable it but straight out say it wasn't designed with it) it runs fine and with that off it runs very well.
 
MavericK said:
That's gross. I was looking forward to a new LEGO game to play with my kid, we've pretty much done everything in DC Villains and Skywalker Saga.
Click to expand...
I mean we can hope they are just being dumb with how they listed recommended settings... but ya if that's real like what the fuck?
 
Sycraft said:
I mean we can hope they are just being dumb with how they listed recommended settings... but ya if that's real like what the fuck?
Click to expand...
Yeah...I mean, LEGO Horizon and Skywalker Saga basically look about as close to real LEGO as I've seen, I don't think they need to go too far beyond that.
 
DukenukemX said:
The 3060 had 12GB and 192-bit bus. Do circuit board wires and 4GB of ram really cost that much more, in pre-AI craziness?
Click to expand...
Yes.
Each of those bits in the bus is a physical connection back to the GPU. Those are some of the largest parts on the GPU itself and the outgoing lanes for memory plays a larger part in determining the rest of the GPU than anything else. They also don’t change size with node shrinks so the more advanced the node the more expensive they are in comparison.
 
Lakados said:
Yes.
Each of those bits in the bus is a physical connection back to the GPU. Those are some of the largest parts on the GPU itself and the outgoing lanes for memory plays a larger part in determining the rest of the GPU than anything else. They also don’t change size with node shrinks so the more advanced the node the more expensive they are in comparison.
Click to expand...
Furthermore, when we compare the RTX 4070S to the RTX 3090, the 4070S with it's narrower memory bus and less vram trades blows with the 3090 even at 4k. When it comes to modern GPU design, like modern CPU design, larger onboard cache is proving to be a notable determining factor in relation to GPU performance.
 
Sycraft said:
I don't get how companies can make games that garbage, particularly with simplistic graphics. UE5 isn't a reason either. Satisfactory is UE5 with reasonably simple graphics and it runs quite well. Even with unoptimized software lumen (the devs allow you to enable it but straight out say it wasn't designed with it) it runs fine and with that off it runs very well.
Click to expand...

Nvidia introduces Ray tracing:

As it becomes more available, developers use it as an excuse to save time and money on optimising for performance, creating an abjectly worse experience.

Nvidia introduces DLSS:

As it becomes more available, developers use it as an excuse to save time and money on optimising for performance, creating an abjectly worse experience.

Nvidia introduces Frame Generation:

As it becomes more available, developers use it as an excuse to save time and money on optimising for performance, creating an abjectly worse experience.
 
Mazzspeed said:
Furthermore, when we compare the RTX 4070S to the RTX 3090, the 4070S with it's narrower memory bus and less vram trades blows with the 3090 even at 4k. When it comes to modern GPU design, like modern CPU design, larger onboard cache is proving to be a notable determining factor in relation to GPU performance.
Click to expand...
Cache certainly helps, just as the extra cache on the x3D CPU’s reduces calls to memory so does the extra cache on the GPU. It doesn’t reduce the amount of memory needed but it can decrease the number of times the CPU needs to ask for data from that memory so it can make better use of the lanes it dos have.

Cache also doesn’t scale well with node shrinks beyond 7nm. So extra cache is also increasingly expensive at smaller nodes.
 
KazeoHin said:
Nvidia introduces Ray tracing:

As it becomes more available, developers use it as an excuse to save time and money on optimising for performance, creating an abjectly worse experience.

Nvidia introduces DLSS:

As it becomes more available, developers use it as an excuse to save time and money on optimising for performance, creating an abjectly worse experience.

Nvidia introduces Frame Generation:

As it becomes more available, developers use it as an excuse to save time and money on optimising for performance, creating an abjectly worse experience.
Click to expand...
I’ve said from the beginning, Ray Tracing was never about more realistic graphics, it was about automating a huge portion of the Art Department so developers didn’t need to hire teams to manually tweak lighting effects and material properties.

All the effects Ray and Path Tracing do, trained professionals can do as well, they will look nearly identical. But one requires the end user to spend an extra $500 on a GPU, the other requires a team of a few dozen artists working for years at California wages…

All those games where people were turning on Ray Tracing and then saying “aside from the FPS decrease, I can’t tell the difference” what you were really saying is, “the image quality from the automated system looks the same as the manual work from the Art Department”

For developers it’s easy math, save $30 million in wages by telling the customer to spend a little more on their next GPU.
 
Lakados said:
Cache certainly helps, just as the extra cache on the x3D CPU’s reduces calls to memory so does the extra cache on the GPU. It doesn’t reduce the amount of memory needed but it can decrease the number of times the CPU needs to ask for data from that memory so it can make better use of the lanes it dos have.

Cache also doesn’t scale well with node shrinks beyond 7nm. So extra cache is also increasingly expensive at smaller nodes.
Click to expand...
The problem is: If we consider the 3090, despite it's outright shader units and TMU's, it still doesn't appear to have the HP to drive 24GB of vram.
 
Frame Generation relies on your CPU/GPU to perform calculations in real time. If your system is already under heavy load—whether due to background tasks, outdated libraries, corrupted system files, or driver issues—the quality of generated frames can suffer. This often results in more artifacts, blurriness, or inconsistencies in the output. Additionally, compiled shaders (if corrupted or outdated) can also introduce artifacts, so deleting and regenerating them may help.
In some rare cases, the only solution is a clean Windows reinstall.

Some videos showing FG issues may be misleading just because system problems = not all “problems” in those clips are real.

Yes, using more and more "fake frames" can cause blurring or visual inconsistencies when objects move quickly (moving the mouse left <> right), especially on monitors with a low refresh rate.
 
I actually see slight artifacting on still screens in rare scenario's, like menu's or screenshots on 'tech tube' channels, I see no artifacting on scenes of high movement.
 
Last edited:
AVATARAT said:
Frame Generation relies on your CPU/GPU to perform calculations in real time. If your system is already under heavy load—whether due to background tasks, outdated libraries, corrupted system files, or driver issues—the quality of generated frames can suffer. This often results in more artifacts, blurriness, or inconsistencies in the output. Additionally, compiled shaders (if corrupted or outdated) can also introduce artifacts, so deleting and regenerating them may help.
In some rare cases, the only solution is a clean Windows reinstall.

Some videos showing FG issues may be misleading just because system problems = not all “problems” in those clips are real.

Yes, using more and more "fake frames" can cause blurring or visual inconsistencies when objects move quickly (moving the mouse left <> right), especially on monitors with a low refresh rate.
Click to expand...

All of that can be said of just general gaming, no mention of frame gen required.

Frame gen is a smoothness filter. Not a performance booster.
 
KazeoHin said:
All of that can be said of just general gaming, no mention of frame gen required.

Frame gen is a smoothness filter. Not a performance booster.
Click to expand...
When the delay between the fake and real frames increases, you end up with some really ugly artifacts, and I know this from my own experience :D
It’s one thing to have a slight blur that looks like a glare, but it’s a whole different story when you have 2–3 mm while swiping quickly with the mouse...
 
AVATARAT said:
When the delay between the fake and real frames increases, you end up with some really ugly artifacts, and I know this from my own experience :D
It’s one thing to have a slight blur that looks like a glare, but it’s a whole different story when you have 2–3 mm while swiping quickly with the mouse...
Click to expand...
For an extreme example (note: the entire video is sort of a gag; he wasn't expecting what he did to be a viable way to play a game) of framegen going badly wrong, and watch (especially the sky) for about 10-15 seconds.


View: https://www.youtube.com/watch?v=HyCXWI1pNHc&t=870s
 
Lakados said:
I’ve said from the beginning, Ray Tracing was never about more realistic graphics, it was about automating a huge portion of the Art Department so developers didn’t need to hire teams to manually tweak lighting effects and material properties.

All the effects Ray and Path Tracing do, trained professionals can do as well, they will look nearly identical. But one requires the end user to spend an extra $500 on a GPU, the other requires a team of a few dozen artists working for years at California wages…

All those games where people were turning on Ray Tracing and then saying “aside from the FPS decrease, I can’t tell the difference” what you were really saying is, “the image quality from the automated system looks the same as the manual work from the Art Department”

For developers it’s easy math, save $30 million in wages by telling the customer to spend a little more on their next GPU.
Click to expand...

That's not true for ray tracing and lighting.

When they were first introduced ray tracing games had to have both traditional methods and ray tracing and it was actually more work.

Ray tracing was generally just slapped on as a feature checkbox and less effort was put into making it look good and ray tracing was typically just used for a few things, the games didn't use full ray tracing. So early on you could argue it didn't look any better depending on how well it was done in the game.



Now as some games only do ray tracing it definitely does save time. But artists still need to tweak lighting for ray tracing to make it look good.

With full ray tracing artists can move things around and instantly see what it looks like in game. They can more closely make light sources match the actual physical lights placed in the game and look good from every angle. The artists have way more freedom in what they can do and can actually tweak things to look closer to what they want with way less compromises.

The old process involved moving and adjusting the values of a bunch of invisible engine light sources and moving the fake light models for them around to try to match them. It is basically impossible to get those to match up perfectly and not look goofy to any players paying too close attention to it. There area lot of limitations and it's really not a fun process most of the time. It ends up being a battle than a work of artistry. You can basically spend infinite time adjusting these and things will look much better from one angle and so you try to prevent the players from moving to a position it looks bad. And every time you adjusted something you just got an approximation of how things would look in game, you had to hit a button and wait minutes, sometimes hours to actually "build" the lights and see the final results in the game.
 
sharknice said:
That's not true for ray tracing and lighting.

When they were first introduced ray tracing games had to have both traditional methods and ray tracing and it was actually more work.

Ray tracing was generally just slapped on as a feature checkbox and less effort was put into making it look good and ray tracing was typically just used for a few things, the games didn't use full ray tracing. So early on you could argue it didn't look any better depending on how well it was done in the game.



Now as some games only do ray tracing it definitely does save time. But artists still need to tweak lighting for ray tracing to make it look good.

With full ray tracing artists can move things around and instantly see what it looks like in game. They can more closely make light sources match the actual physical lights placed in the game and look good from every angle. The artists have way more freedom in what they can do and can actually tweak things to look closer to what they want with way less compromises.

The old process involved moving and adjusting the values of a bunch of invisible engine light sources and moving the fake light models for them around to try to match them. It is basically impossible to get those to match up perfectly and not look goofy to any players paying too close attention to it. There area lot of limitations and it's really not a fun process most of the time. It ends up being a battle than a work of artistry. You can basically spend infinite time adjusting these and things will look much better from one angle and so you try to prevent the players from moving to a position it looks bad. And every time you adjusted something you just got an approximation of how things would look in game, you had to hit a button and wait minutes, sometimes hours to actually "build" the lights and see the final results in the game.
Click to expand...
They still need people, but instead of 50 people for a few years, they need a half dozen for a few months.

It was always about cutting back on development costs.
 
Lakados said:
They still need people, but instead of 50 people for a few years, they need a half dozen for a few months.

It was always about cutting back on development costs.
Click to expand...

No. When they used both it did not cut back on development costs, it increased them.

So it was not "always about cutting back on development costs".
 
You must log in or register to reply here.
Back
Top