"Fake Frames" Tested | DLSS 4.0, MFG 4X, & NVIDIA's Misleading Review Guide

[@GamersNexus]

Prebuilt reviews in the works! There are some more 50-series tests coming up, but we're starting to also work-in some other component coverage! Watch the Paper Launch video! This one was fun: kzbin.info/www/bejne/rX7HY4p-gNOWm6s Watch our RTX 5090 review here: kzbin.info/www/bejne/jIi2nYJ5lM-cgrM Support us here! store.gamersnexus.net/ Or here: www.patreon.com/gamersnexus

[@MrPruske]

steve yells a lot in this one, and even this comment with the links are all with exclamations with a couple exceptions XD

[@AwesomeBlackDude]

How do I RMA Nvidia fake frames?

[@robainscough]

DLSS and all the frame generate garbage is for those idiots that think they can tell the difference between 75Hz and 240Hz … how ironic, fake frames for fake user perceptions. :)

[@wahidtrynaheghugh260]

Fan tester video?

[@boburanus69]

I am coming to the conclusion older hardware is about to be more stable/valuable for gamers, as are older games. We are less than a year away from a massive implosion of the media entertainment industry, where it just becomes unfeasible to run a company in the manner these AAA studios and GPU makers want to. When Intel, the CPU maker, has a better budget/mid tier GPU offering for gamers than team green/red, that's a sign the market is turning into vaporware/paperware. On a planet where the 3xxx series never had less than 12gb ram, the 4xxx series never had less than 16, and the 5xxx series never had less than 24, where the performance and generational uplifts are clearly worthwhile, where the price points are truly competitive and you don't have to sit here and split hairs over game title/settings/FramGen to decide what card to buy, people would have the confidence and comfort of buying Team Green again. But, we live in this world where Nvidia would rather sell an AI card to a business or streaming service to an end user, where the hardware is no longer being valued from the enthusiast perspective. The hardware simply isn't being valued appropriated anymore. Company releases GPU, sets price to $2000, has almost nothing for stock at launch, then also markets the card as if it isn't just a slightly better 4080 (ti). Can't trust the marketing, can't trust the hardware, can't trust the industry right now. That's gonna boost the value of older hardware again and have people buying 9xx/1xxx series cards from the used market again, which is why Nvidia is stopping support of those generation suddenly (speculative).

[@zurielsama]

No fake frames for me. Only the finest quality, organic, grass fed pastured raised real frames for my family and their games!

[@jeff230394]

You just poor n can't afford fake frames like me. I am elite and special

[@CD-vb9fi]

@@jeff230394 yea, but why waste all that money for fake frames when you can just close your eyes and use your imagination?

[@justbob8294]

You are what you eat.😂​@@jeff230394

[@AlexanTheMan]

This and fully uncompressed pixels. Being the highest quality pixels you'll ever see.

[@willjohnsonjohnson]

@@jeff230394 You're special alright ;)

[@lerkzor]

nVidia has investigated nVidia and has found that nVidia's changes to the best measuring tool have resulted in better results when measuring nVidia's fake frames.

[@Faith_Walker]

Yup bullseye 🎯

[@Senthiuz]

nVidia now states that their investigation has revealed that generated frames, in fact, are more real than "input" frames.

[@rasta77-x7o]

We investigated ourselves and found nothing wrong... Story of everywhere lately!

[@sujanrages]

No wrongdoing was found

[@Grandmaster-Kush]

The human eye can actually see more of Nvidias generated frames than FSR

[@hakzource8660]

Honestly using AI as a crutch for the underwhelming generational upgrades and still increasing the price is crazy

[@g2jxGhF5G8z1gL7S]

that's what this is all about. the gaming vertical for Nvidia is now the afterthought. they will be an AI only vendor in 5-10 years I'm predicting.

[@ElijsDima]

Tbf AI is the upgrade, gaming is just a side-hustle for Nvidia nowadays.

[@HumbleDude46995]

AI & GAMING is bad. Have you seen Reflex 2? competitive FPS shooters will suffer

[@HumbleDude46995]

@@ElijsDima sad truth

[@InfernusdomniAZ]

Crazy? No. Capitalist? YES!

[@OxyShmoxy]

The most infuriating thing about NVIDIA is how they want to go all in on fake frames but at the same time refuse to give us more memory.

[@phillgizmo8934]

and all this while even nvidia engineers admit that those technologies use more VRAM

[@kekistanifreedomfighter4197]

Remember when Nvidia's shills said that AMD was just adding more vram to make their cards seem better? Pepperidge farms remembers. Hats off to AMD and Intel for making 8gb the absolute minimum and 10/12gb the norm on even the low end.

[@ChaosSwissroIl]

If team green wasn't aggressively anti-consumer, would they still be team green? By deliberately having the cards have insufficient Vram, they artificially push more customers to the more overpriced 80's and 90's. It's the same thing that Team Blue does with their CPU's.

[@strelatronics]

@@ChaosSwissroIl Most people can't afford 80's and 90's, so I doubt it's an effective strategy

[@jpteknoman]

Its called planned obsolescence. Without enough RAM and with DLSS allowing games to be completely unoptimized, you will HAVE to get a new card every new gen if you don't have the xx90.

[@DaniIhzaFarrosi]

That opening is the most concise, short, and informative about the current Nvidia situation

[@aclausenyt]

Kinda Steve's "thing"... I'd argue.

[@hansdoreen]

100% agree with Steve! Dickvidia.

[@AuspiceGwan]

"As simple as possible NVIDIA is being -- a deeck"

[@idindunuphenwong]

Not just the Nvidia situation. A technological industry situation. Fake FPS didn't exist 10 years ago.

[@TheGalacticIndian]

NVIDIA’s market is already collapsing, but they think it’s still running smoothly thanks to reality frame generation😎

[@kazem-playz7742]

Hope they can 4x ai generate stock too

[@GamersNexus]

ooooooof

[@sergiocamara93]

They are too busy trying to ai-generate stock exchange rates atm

[@Aggrofool]

They will 4x the trump tariffs

[@TheTwoTwoForReal]

@@GamersNexus I'm still waiting for the day I can use AI generated money to pay for a 50 series GPU.

[@kazem-playz7742]

@@GamersNexus6 of those o’s were ai generated from a single o

[@Grumpy_ol_Gamer]

MFG is like insurance, it works great when you don’t need it.

[@EatingMachine23]

This ^

[@rarespetrusamartean5433]

this is the best analogy known to man lmao

[@gmonkman]

lol, best comment

[@Seacle14]

What? When you don't need it insurance just costs money. Insurance only works when you DO need it.

[@rarespetrusamartean5433]

@@Seacle14 the comment implies that insurance works great when you're just paying them with no incident but when an incident happens they find some niche contract clause to not cover your damages

[@Macronicol]

We are so lucky to have you on KZbin. The amount of high quality educational content you give us is incredible and your honesty is invaluable.

[@JonoSSD]

Nvidia: I heard you like frames, so we put fake frames in your fake frames while you wait for a real frame to be queued and upscaled.

[@kevinforget549]

I feel like their time would have been better spent trying to find a more efficient way to rasterize frames

@@kevinforget549 YES! They took the lazy way out to make a quick buck off the backs of gamers that don't really need a card like the 5090 to play games.

[@rj7250a]

Now in the yo dawg meme format: "Yo dawg, we heard you like AI frames, so we puta AI frames in your AI frames, so you can render while you tender"

[@LordNecron]

And then discarded because all the stupid sheep buying our overpriced, scalped cards won't know how the actual game looks anyway, due to them being so brainscrambled by our 'invent half the screen's content' tech, that we can just show a lawyer friendly yellow mouse and claim this is Duke Nukem.

[@JgHaverty]

@@kevinforget549 They... did. The power drop on this generation is massive ... You kids just really look for any reason to get mad. The native performance/watt of these cards is huge compared to the 4000s. You're too busy getting sucked into this crap virtue signaling about DLSS to realize that. And before you "get mad" at me; you need to understand, these are objective numbers; not an "opinion" you get to disagree with. I am fully aware my reply wont change your mind; but I do just want to say you are objectively wrong and this focus on parroting inciteful youtuber comments is pretty egregious.

[@mybaIIz]

16:29 FRAPS!!! What a throwback.

[@mark6302]

oh snap

[@Romess1]

Only 90s kids remember

[@ian_dot_com]

teamspeak 3

[@LupusAries]

Xfire...

[@jonbeargenx]

I had a real flashback when Fraps was mentioned. 😊

[@Probablyacowtbh]

We'll have to start noting RAW(true) FPS.

[@CptJistuce]

USDA organic frame rate.

[@mryellow6918]

In engine frame

[@charlesleonitol.iringaniv8320]

Are these frames free range?

[@Vladislav888]

Just write the projected actual framerate in a big, bold font and DLSS framegen FPS as grey and small. Same for the resolution when it comes to upscaling.

[@tremapar]

We should just start using a system we already have for part real and part imaginary measurements. Complex frames. Instead of being stuck with crappy 30 fps, I can use DLSS4 to get all the way to 30+90i fps!

[@vulcan4d]

AMD seriously needs to advertise "Real Frames" and focus on that. We need a new GPU generation, not frame generation.

[@Razor4884]

Given the Nvidia 5000 series GPU quantity situation, I think AMD could even advertise "Real GPU's."

[@joshuaprobst5380]

AMD isn't even in the high-end GPU market anymore. Nvidia has full control and consumers are going to suffer because of it.

[@Karibanu]

@@joshuaprobst5380 If they can kick the 9070 up to somewhere vaguely competitive with the 5080 but not kick the price up, then they're in the high-enough end gpu market. Nobody buying 5090s is going to buy anything else anyway because they either care about productivity stuff, or they just want a halo product & don't care about anything else.

[@chasetheswift]

Hell yeah brother

[@aroulos]

Instead they will cram more useless vram into their GPUs and you idiots will eat it up. 20+ GB on a strictly 1440p card that will never under any circumstances be in a situation where more than 12 of them are used up, lol

[@audiogarden21]

Steve slowly turning into Geralt of Rivia.

[@21stcenturybohemian]

I was thinking Gandalf, but Geralt works too. :D

[@zivzulander]

Thanks, now I cannot unimagine Steve in a certain tub scene 😅

[@thepowerlies]

Does he accept contracts?

[@htrajan]

@@thepowerlies he’s taken down the baddest contracts from NZXT to Artesian Builds and EKWB.

[@EtaCarinaeSC]

Gerald of hardware Trivia

[@zdrux]

"Hi, I'd like good 3D performance" "Best I can do is 3 AI frames for 1 real frame"

[@GameBacardi]

hahaha

[@AFistfulOf4K]

Have you used it?

[@Shieftain]

"While upscaling from one quarter of your target resolution"

[@Bunnibunnibunni]

@@AFistfulOf4K There were like 300 50 series cards sold, no one has used it

[@AFistfulOf4K]

@@Bunnibunnibunni how about any form of framegen then. Has he used it?

[@Odin029]

"They have their own Steve... his name is Brian" I had to wipe some tea off of my screen after that one. Thanks Steve

[@ND_USR]

+1 hahahahaaha

[@ian_dot_com]

thanks steve

[@rainbowkrampus]

Turns out, Steve was just the hair all along.

[@TheHunnerbass]

Patrick's "Im so fucking tired of redefining this every time a new DLSS version is released" energy was fucking hilarious even if unintentional lol Every engineer on earth felt that Patrick. Every. Single. One.

[@Cinnabuns2009]

This is like when Cypher explains the matrix encoding to Neo: Neo - "Do you always look at it encoded?" Cypher - "Well you have to, the image translators work fOr the construct program" I can see an era where Nvidia just displays static on the screen and we're expected to translate it visually but at least it will be 3000fps.

[@TeaTotalTemplar]

Who would have guessed all those years playing classic ASCII dwarf fortress would prepare me for the "best" graphics of the future

[@sammiller6631]

"Nvidia is being a dick" should be put on a t-shirt. Hundreds of people wearing "Nvidia is being a dick" could walk into CES next year as a flash mob.

[@DavidGotelli710]

Lmao I'm sure Jensen would even laugh

@@DavidGotelli710 Jensen does not care, he got your money.

[@peyton_uwu]

+bump would love to see this

[@garypinholster1962]

100% buy asap if Steve does this haha.

[@clearlisted]

a very long and hard one and customers are being bend over

[@averyparr6910]

I really appreciated the segments hosted by Patrick. I know it’s been done a few times before, but it feels like a really nice way to break out the credit for the individual pieces of work comprising a piece. I would love to see this more (to the extent that whoever has done the work would like to be on camera, of course).

[@jbrashfx]

It's absolutely insane how different this review is from every other one out there so far, which I mean in the best way possible. Obviously everyone who understood what MFG loosely meant during or shortly after the event knew not to expect 2x the performance form the 5090, but the degree to which this breaks that apart is SO MUCH GREATER than any other video in existence. Props to steve & the team. Love this kind of reporting.

[@RabidBadger_]

AMD and Intel: Our upscaling tech supports nVidia cards from 9 years ago. nVidia: Our upscaling tech doesn't support the $4k nVidia card you bought last week.

[@or1on89]

I love how there is even an Open Paper about Ray Reconstruction from AMD…but still many reviewers and communities keep championing the closed APIs from Nvidia rather than asking why no work is done towards enhancing the standards that everyone can implement on their hardware. It feels pretty clear to me that Nvidia is not that confident and doubled down on the gatekeeping to maintain an advantage…

[@HyBlock]

it's actually the opposite. they will bring it to older cards, in comparison amd isn't going to support their previous (latest gen) for frame generation when they drop the new model

[@or1on89]

@ AMD is not going to support the ML upscaling...it has nothing to do with Frame Generation and AMD already supports older cards with Frame Generation. As far as we know FSR4 is all about the upscaling algorithm and will support RDAN4 cardss and RDNA3 and 3.5 cards later in the generation.

[@maskharat]

@@HyBlock Dude, FSR works on my GTX 970, enabling me to play stuff that card would have been too weak to play. FSR enabled me to use a -14- 11 (misremembering on my side. It was released 2014) year old competitor card of AMD for a generation longer. What doesn't work on it is *any* version of DLSS. General service advice, don't snort while brown-nosing, that's not healthy.

[@derbigpr500]

Are you delusional? Nvidia literally released DLSS4 to older cards a few days ago and gave a huge boost to people with old cards. You AMD fanboys are pathologically delusional, it's like you look at reality, and then decide, NOPE, I'm gonna say the exact opposite of reality to cope with it easier.

[@kenryoshi]

As a Linux user, "Nvidia is doing Nvidia things" resonates so hard lmao

[@Xerpocalypse_]

Wish I could upvote this more than once. I dual boot Linux and Windows as an Nvidia user and sometimes the difference is staggering. Stares at the Nvidia driver that arbitrarily boosts to P2 power state and consumes 50W while watching KZbin.

[@monk3y206]

same here +1

[@lubossoltes321]

Yeah, that's why I have only AMD HW at home, full time Linux user ....

[@formbi]

Novideo

[@anonymist-n8z]

I would not have gotten my used nvidia gpu if i didnt have a steam deck as my linux PC. When rdna 5 rolls out im picking amds highest tier GPU and forever saying goodbye to windows for good

[@WayStedYou]

Fake it till you make it Wait.... that isn't a Jensen quote

[@grillinman84]

Only the gamers will get that one. Huh...that's not right either.

[@SinonStrike]

Nobody likes my captain nvidia shield 🛡 !?

[@666gene]

@@grillinman84 its "make it, so you can fake it"

[@trith72]

No but its a way of life for his company now.

[@vicente3k]

The more you buy, the more you fake 🤔

[@DaGuunOSHO]

“Being the last guy in town using candlelight , walking outside and saying (I have a great idea, what if we switch to electricity?)”. GM. 😂

[@Chiongster101]

Finally someone says it out loud that the naming convention for frame gen falling under super sampling naming scheme is disingenuous

[@FuelX]

Maybe they meant sub sampling ;)

[@Krushx0]

and finally reveal it that it doesn't matter because nearly every body do not know what that is and in fact doesn't care for it. They just use it and happy with it and never read comment and write about like amd fan boys do. Just like in a car you do not need to understand the several underlying technological marvels inside of it to drive and enjoy the features that it provides. If you perceive it as net positive there is no harm for you. Sure you have to pay the price for it but thats how business and cutting edge works whoever invest into the future will dry ra*e like no tomorrow those who doesn't. Numbers do not lie, just do not even argue whoever you are trying to this comment.

[@pkpnyt4711]

@@Krushx0 tell me you don't know what you're talking about and be an obvious shill by simply talking. It does matter if it doesn't work for fast paced games. The most popular games are mostly the fast paced games which frame gen is completely useless for. What's marvelous about a tech that can't be used for all games? Stop it.

[@mechanicalmonk2020]

It's just a brand

[@LiamNajor]

@pkpnyt4711 while also failing utterly to: A) Convince me they are not a literal, paid shill B) Demonstrate even the slightest understanding of the technologies they are bashing and shilling C) SPEAK ENGLISH PROPERLY

[@Dawnbandit1]

The "Probably real percent" line is hilarious.

[@sjolson68]

Misleading naming and description of features + doesn't support the industry standard benchmark solution which is open source = dubious intent clearly

[@andersjjensen]

Anything that is proprietary to Nvidia = Dubious Intent(TM). Remember PhysX? Remember how Nvidia pushed a driver update that would disable PhysX if an ATI card was present in the system too? Yea... I bought a cheap Nvidia card to get PhysX acceleration, but Nvidia decided that I didn't give them ENOUGH money, compared to the money I had spent on my existing/primary card, to get PhysX despite the box saying "PhysX acceleration". That was 20 years ago. That's when Nvidia lost me as a customer forever. And nothing has changed since... Neither with Nvidia, nor with me.

[@Netist_]

"DLSS is kind of the opposite [of super sampling]". That's what kills me. It's funny how people have forgotten that DLSS WAS originally marketed for super sampling. Use AI to scale up to a higher resolution, then super sample back down to native resolution, in order to get improved image quality. But nvidia completely memory holed that idea when developers realized they could just use it to upscale instead of optimizing their games, and nvidia realized they could make bank selling cards that magically let you play games at 4k (even though it looks like trash).

[@EarthIsFlat456]

DLSS looks like trash at 4K? Need to get your eyes checked.

[@kalnik47]

Isnt that the DSR or DLDSR?

[@ewitte12]

There are quite a few decently optimized games. May not be optimized for performance the trend is to optimize for graphics and let DLSS bring performance back up. It is still optimization just not what your looking for. Saying something is not optimized because it is slow is not true.

[@alles_Freaks]

nvidia got the market so confused we still believe dlss looks better than native. 😂

[@Netist_]

@ DSR and DLDSR are both super sampling technologies, which is what makes this so ironic. It's right in the name: DLSS is Deep Learning Super Sampling, yet the modern version of it isn't actually super sampling at all

[@LonSeidman]

I got the same pitch at CES on this frame generation stuff.. This confirms that what my simpleton mind thought about at the time - they have hit the wall on getting significant year over year performance gains and developed a crutch to continue making the case for new hardware. And of course they make a lot more money focusing on AI and to a lesser extent crypto miners so graphical performance is not a priority for the company at the moment. My 3080 is still driving adequate 4k performance for the games I play to my 60hz OLED TV so I haven't felt the need to upgrade yet.

[@christophermullins7163]

Yep. I was running a GTX 1080 on a 4k screen for a couple years. 1) there is plenty of other benefits of using a high res screen like content consumption and real estate and 2) upscaling to 4k still looks better than native 1080 or 1440p.

[@eawblablatron9161]

they have hit the wall in general renovation from cell phones to gpus

[@eawblablatron9161]

"used to make or hoped to make more money on AI. at least until DeepSheep release"

[@EmanuelHoogeveen]

It seems to be a process node thing more than anything - they got a larger than average boost after switching from Samsung back to TSMC for the 40 series, but now TSMC's next process node isn't quite ready yet and so the 50 series got a smaller than average boost.

[@hammerheadcorvette4]

I don't think they hit the wall, more than they just decided to not invest into a line wide Hardware Architecture revamp and push it out a generation to save money in R&D.

[@highcue]

Did you consider doing a blind test where you and your team have to guess if they are playing a 4090 without fake frames and a 5070 with fake frames (once it comes out)

[@AFistfulOf4K]

GN would get too much hate from people who've never used it. I've done that experiment IRL with several of my friends, playing Deep Rock Galactic at 240 native or 120 interpolated using Lossless Scaling's frame generation. Nobody can tell the difference. As far as I can tell this entire controversy is manufactured by influencers who need something to be angry about for engagement. The only other explanation I can see is people using it for something it's explicitly not made for, like trying to go from 20 FPS to 40 and rightfully saying it's not an improvement at that performance level... like, no shit.

[@RoyalMudcrab]

Huh, good idea.

[@Demigirlboss]

Patrick is a really good host, and I really hope we get to see more of him in that role in the future!

[@CasualGamers]

Great video!Looking forward to Latency testing! Very few outlets doing it and we need the reliable data GN provides. Also, great to see Patrick with more screen time. Cheers

[@wallnuts7]

0:04 Thanks for the fast break down Steve! lol

[@Tanmay0607]

Lmao doesn't even care about being demonetised

[@daviddavis7255]

This man said " modern trend of wood accents". The world has come full circle lol.

[@yugdesiral]

70s station wagons are making a comeback

[@claudiobizama5603]

​@@yugdesiral I saw a Rivian with (factory iirc?) options of wood accents on the interior We are slowly coming back

[@MazeFrame]

Design is a big circle, I tell you!

[@ancientflames]

@@claudiobizama5603they finally realized wood is classy af if done right. The future is not plastic and chrome.

[@MikeOHara1479]

The candlelight/electricity statement had me laughing more than it should have.

[@GamersNexus]

YOU GUYS WERE ALREADY USING ELECTRICITY? WHEN WERE YOU GOING TO TELL ME?

[@JuniorJunison]

​@@GamersNexusin the next windows update.

[@shinobi-zensei]

The 50 series in a nutshell: We can’t make our GPUs any faster, lets pretend like they are then convince everyone.

[@MultiOlimac]

on raw performance the uplift from the 5090 to the 4090 is barely 8%, ridicolous, still cant run cyberpunk 2077 at 4k with at least 45 fps. a shame to be honest

[@davefb]

Isnt it more that, they CAN make the gpu faster, but they're already capped due to peoples cpu's ? So in order to sell to someone with a 'perfectly okay cpu', they need to be able to 'show' a gain, even if its potentially "fake" ?

[@ctrlectrld]

0:02 "Nvidia is bein a dick" Thanks for going straight to the point and for respecting our time.

[@snarfbot]

how about "interpolated frame" as it was called forever until nvidia invented it.

[@Sir_Robin_of_Camelot]

That too accurate and not misleading

[@RobertleeEtn]

Motion interpolation is like Early 2000s tech now its Revolutionary because Nvidia? lmao

[@opticalsalt2306]

I’ve said this in multiple comments and people always shut me down about how much more complicated it is… suuure If my smart tv from 2017 can do it or my 6600xt can clearly it’s not alien tech (bringing that up makes me an amd lover or Nvidia hater, conveniently also told it’s so much worse than REAL framegen).

[@bosstowndynamics5488]

​@@opticalsalt2306I hate to defend Nvidia here but generating the fake frames with deep learning models is going to result in better visuals than the terrible motion interpolation hardware in consumer TVs, which also add a ton of latency. There's a degree to which it's going to be inherently worse than real frames but it is better than previous techniques

[@MadLustEnvy]

Frame interpolation might be more understandable, but it's inaccurate. AI generated frames are much, much better than interpolation between two frames.

[@ElderGamerX]

So, Nvidia wants to standardize a metric that's more favorable to mult gen, while simultaneously not contributing to the open source code that forked off from? I'm thoroughly shocked...

[@garydiamondguitarist]

Not sure why, this has been happening with open source emulators for years, parts of the code base sometimes turn up in paid retail releases of older games on newer consoles. In Nintendo's case it's likely profited from open source emu projects while having other emu projects closed down, if looked at a certain way it's hypocritical.

[@Vegemeister1]

There are open source licenses that prevent this behavior. Intel chose to release PresentMon under the MIT license, which does not. If they don't like it, well, maybe they shouldn't've listened to Google-brained influcencer-programmers.

[@markwestcotts]

I've been an investor in Nvidia for the past six years now because I strongly believe in the company and in the stock, but now I've been thinking of redistributing my portfolio of about $560k and buy other stocks. Maybe buy GE or Tesla? The decision can be confusing sometimes

[@byrrongardner]

I think the next big thing will be A.I. For enduring growth akin to META, it's vital to avoid impulsive decisions driven by short-term fluctuations. Prioritize patience and a long-term perspective most importantly consider financial advisory for informed buying and selling decisions

[@markwestcotts]

I'm intrigued by this. I've searched for financial advisors online but it's kind of hard to get in touch with one. Okay if I ask you for a recommendation?

[@byrrongardner]

My FA is Gabriel Alberto William, a renowned figure in his line of work. I recommend researching his credentials further. He has many years of experience and is a valuable resource for anyone looking to navigate the financial market

[@markwestcotts]

He appears to be well-educated and well-read. I ran an online search with his name and came across his website; thank you for sharing

[@BarackLesnar]

lol bot bot bot bot

[@MMOStein]

GPUs in 2005: 50 fps. GPUs in 2015: 200 fps. GPUs in 2025: Lets pretend that 50 fps is 250 *FidelityX SUPER* fps.

[@wolfkane1986]

Gpus in 2015 never hit 200 fps,maybe in 480p

[@IcecalGamer]

@@wolfkane1986 Quake 3 arena

[@Saturntime33]

I love this comment 😂

[@wolfkane1986]

@@IcecalGamer yea legoland graphics 🤣🤣

[@derbigpr500]

You when asked to tell the differnce between real and "fake" frames: **confused pikatchu face** "looks the same to me"

[@antonvillalta9205]

The Steves talking about people talking about their hair, not a thought I thought I would ever have.

[@Hauntear]

Loving the energy in the latest videos, keep it up!

[@GamersNexus]

Thank you! Having a lot of fun with the content we're working on!

[@wastanley734]

​@@GamersNexus so what is your take with gnca of NVIDIA using the public as a ai pool and then pay wall consumers to the results of their own work like the have done with 3000 to 4000 series? Isn't that illegal due to Sherman Act says that's illegal from monopolies doing that exact thing after the great depression?

[@milowadlin]

On your first chart, your blue bars are labeled "Control (No FG)" You could get in the spirit of nVidia's innovations by shortening that to "Control (NFG)".

[@GameTimeWhy]

Frfr nc

[@NoneRain_]

I prefer the term "fake-frame" so: FF and NFF

[@ChrisM541]

Nah, it SHOULD be "Control (No Fake Frame)" ;)

[@IAMSEYMOURMUSIC]

CNFG

[@nulshift]

@@ChrisM541 So CNFF.

[@nomarzenun]

NVidia is selling software(dlss 4.0) not hardware for 5000 cards. Why even bother with 5000, just keep making 4000 cards and just add TI at the end.

[@distrusts]

Replace the "T" with and "A"

[@gaminginstilllife9429]

Plus locking down options behind pay wall

[@TatsuyaWow]

a ti should be like 30% or more improvement 50 is less lol

[@scarx4181]

@@TatsuyaWow Eh, TI performance improvements have always been all over the place. The 3090 TI for example was less than 10% or so and coincidentally matched the 5090's pricetag at 2000. The 5090 is also around a 30-35% increase over the 4090 for about the same amount more wattage lol. The rest of the 50 series... well at least the 5080 overclocks decently, we'll see how the others look but probably more of the same.

@@TatsuyaWow They way they were talking before the 5090 came out it was more like 50%.

[@herihelke729]

Patrick missed the oportunity to say "Thank you Steve".

[@JPEaglesandKatz]

Where would we be and what would we do without you Steve and your team!!! You provide an invaluable in depth technical analysis where all other just skip it fail to grasp it!!! Thanks for doing what you are doing!

[@Snarfnado]

Makes me happy to see Nvidia under the microscope recently. Good work Steve

[@paulosouza449]

Still way too high

[@derbigpr500]

He's a hater spreading false narrative about Nvidia, you realize that? Because he knows the internet is full of delusional AMD fanboys like yourself who love looking at videos of someone unreasonably trashing Nvidia. Even when they do amazing things.

[@innos_god]

@@derbigpr500 give us the evidences or stfu

[@derbigpr500]

@@innos_god What evidence, I've been playing it for the past 2 days testing in every game I have. So have plenty of people, look up some forums, plenty of videos on youtube, plenty of side by side screenshots to find as evidence. The experience people are having is overwhelmingly positive because DLSS4 finally does what it promised, more performance without any visual degradation, and gives new life all older RTX cards boosing their performance massively. So how is any of that a bad thing? You people would have having parties on the street if AMD did 5% of this, yet when Nvidia does it, it's still somehow BAD? But just by asking to show you evidence, it's clear you DON'T HAVE AN NVIDIA CARD, so, you stfu, you have no right to publicly speak about it, becuase you CAN'T talk about something you never tried, have no experience with, and no knowledge about. That's how dumb people behave. Or AMD fanboys. Same thing really, nobody smart would buy an AMD card to begin with.

[@Fiendir]

​@@derbigpr500huff just a liiittle more copium and I'm sure Jensen will return your calls to confess that your undying love for him is mutual 💖 Seriously thou, you shouldn't think any GPU company has the consumers best interests in mind right now, especially not Nvidia.

[@timweydert3490]

Love the paper boxes representing the paper launch on your desk haha!

[@Apokathelosis]

nvidia frames are also like cardboard props

[@garydiamondguitarist]

Good spot! You're right, decent visual gag, didn't notice until I read this.

[@oceanmangg]

THANK YOU for getting the super resolution part clear. It's crazy how nvidia has messed up the perception of what super imposing is for the masses by upsampling instead of supersampling

[@Uni1187a]

This is the first youtube video I have ever DECREASED the speed on to .75x (for just the first 2 minutes). Steve, you were talking like an auctioneer. Lots of awesome information though, so appreciated!

[@sylvie7868]

frame-gen tech feels like snake oil

[@InfernusdomniAZ]

Not just feels like it is.

[@tonyd7164]

one real snake and three fake snakes.

[@hochhaul]

The situations where it would be most useful suffer from increase input lag. The situations where it's unneccessary, it adds artifacts. It isn't a replacement for raster performance and I don't buy the excuse Nvidia gives us for why this is supposed to be the future. I think Jensen is chasing the more profitable enterprise AI market and he wants gamers to just accept that their GPU's designs are now prioritizing increased AI performance over raster performance, and DLSS is the bandaid to mask the weak generational performance increases.

[@Fetchdafish]

As someone who has actually tried it, it actually works very well under the right circumstances. What feels disingenuous is how hard they're leaning on it. Framegen should be a cool bonus feature in your marketing, not what you lead with. That just shows your product hasn't really improved much.

[@ChuddAnon]

Based avatar

[@ZinoAmare]

Steve "We're only scratching the surface of this topic today." Me "It must be 40 min long then." **Looks at timer **38:58****

[@GIjoelly]

He could add some fake frames to make it even longer

[@ChiekoGamers]

Can we pay the leather jacker boi fake money too?

[@EatingMachine23]

Sure, as long as 1 in 4 notes/bills is real 😂

[@Akkbar21]

Trump coin

[@Sky2I]

Sure, just don't buy

[@psybertao]

This video is fantastic. Only halfway through and it confirms my bias for playing ten year old games on my laptop's integrated graphics and simply enjoying them.

[@LaymensLament]

i appreciate this: "it cant predict the future, it can interpolate from the past" is somethig you would argue in sociological systems theory on e.g. how conciousness projects futures :D. core description is "conciousness is a process that operates with its back to the future" (because you only ever react to stuff that has already happened in there, like 'experiencing' a thought can only happen after the thought somehow appeared/got established as a thing). it might sound stupid but its actually correct and relevant (imho).

[@sinnwalker]

Going on a tangent to prove we have no free will as an analogy to a GPU matter is actually hilarious. But yea I agree.

[@zeus1117]

Many people forget that most games we play now are a bit older. NONE of those will receive any DLSS update or RT update ever!

[@AFistfulOf4K]

You can use the nvidia app to force DLSS transformer model on anything back to 2000-series and even people who've been malding over DLSS for the past 5 years agree it's good enough there's nothing left to complain about anymore. Oh but yeah besides that green company bad. 😴

[@meffmeffmeff]

@@AFistfulOf4K I was searching for way too long for one sane comment like this. It's incredibly infuriating how people can hate on something they don't understand

[@alexpkeaton4471]

@@AFistfulOf4K You seem to have confused "games" with "video cards".

[@alexnorth3393]

True

[@MooseCastle]

​@@AFistfulOf4KThe difference being that you don't need to do that on older games for them to run just fine, and if you want to use it to upscale older games to make them look sharper then you don't need a 50 series.

[@atherasia]

"What?! You mean a multi million dollar company is lying about their product performance?! This can't be happening!" I'm so glad everyone can just walk past the "free frames". "Oh, average consumers won't care. It's free performance". Well, I care. Nvidia treating their gaming customers like dumbasses. This shouldn't be the trend in any market whatsoever.

[@anonymist-n8z]

Oh but it is and will be, the reddit gaming culture is mindless consumerism and its incredibly strong especially in pc hardware circles. Consooooom

[@bansheezs]

Most gamers are dumbasses

[@ashkebora7262]

_Every_ corporation treats their customers like dumbasses. The people going, "lolol you care about that bro?!" are the true dumbasses. Lemmings forking over money just to be spat on. Such a pitiful lot.

[@sebigamer]

*trillion

[@rabb1tjones921]

I love the new get to the point intros to your videos. This one especially. We were all thinking it.

[@denvera1g1]

I raised these concerns on launch day about how the tool used to measure these frames, is closed source fork and maintained by the company that makes the cards and would benefit from no one being able to properly inspect the performance metrics, and possibly fudge the numbers of reflex on+ OMFG off, vs reflext & OMFG both on to make it seem like frame gen doesnt add latency, and apparently in some reviews, reduce latency over native with reflex

[@huntari]

I vote we use the term 'organic' frames vs 'artificial' frames. This also lets us make endless jokes about raw, free-range organic frames compared to nVidia's factory farmed artificial frames.

[@zivzulander]

I only consume pasture-raised frames 🧐

[@zacharyharwell351]

​@@zivzulander As our lord Gaben intended

[@-Devy-]

This is such a 0 IQ comparison. Organic doesn't automatically mean good any more than industrial automatically means bad.

[@dead-claudia]

@@-Devy-it's a joke

[@xypha85]

​​@@-Devy-0 iq response to a clear joke lol

[@luismovil5341]

Wait, you mean the company supplying the industry copying intellectual property to pass it off as their own with hardware isn't willing to work with the rest of the industry, by copying older versions of the software as their own in a closed off way that might lead to misleading results favoring them?

[@JJFX-]

Say it ain't so!

[@-Devy-]

"to pass it off as their own" Literally at no point did they do that. Learn to listen.

[@GreenBean6570]

What is bro talking about 😭Noone stole anything

[@nogravitas7585]

@@GreenBean6570 "AI" or LLM are trained on copyrighted IP books/art/music there are like a dozen lawsuits currently ongoing about it, assuming this isn't a bot commenter given the use of emojii it seems unlikely.

[@luismovil5341]

@@-Devy- This thing called OpenAI, sounds like you should learn to listen to news.

[@JesseNewcomer1]

this was long, but full of fantastic information. ya'll do an amazing job and thanks for this and a great website that isn't full of ads

[@ariprabowo85]

So... Can I say that when GN's Steve met Nvidia's Brian they mostly talk about Hair Works? .... ... .. . Sorry about that. I'll see myself out.

[@GamersNexus]

We sort of did though!

[@MaaZeus]

​@@GamersNexus Lamenting about the fact that despite the advancements of technology still the games cannot render accurately your flowing locks?

[@tek_lynx4225]

Should compare it against Lossless scaling app, since it can work on any modern gfx card and any program, and do alot more then 4x multi framegen.

[@GamersNexus]

We've heard about that one! Might look into it.

[@ryanspencer6778]

Don't forget the ability to run it on a separate GPU to avoid performance hits to the game.

[@robotmechanicalwhizkid2521]

@@GamersNexus yes please do

[@tek_lynx4225]

I've been using it with 30\60fps locked games as of late, and its really impressive. The input latency does not feel as bad as thought it would be. I'd love to see tests of what the actual input latency % increase is.

[@03chrisv]

It has more latency and more artifacts. Nothing exciting or surprising. It's not bad by any means but DLSS frame generation is superior.

[@CattyRayheart]

These deep dives are definitly your best content.

[@th3R0b0t]

I enjoyed deadpan Patrick with the "I've been looking at this garbage for over a week and just barely understand what could possibly be going through the minds at nvidia... please.. send help." With Steve's "how INSANELY STUPID IS THIS?" interruption... I was giggling and had to rewatch Patrick twice because of his 1000yard stare.

[@zivzulander]

2:00 I'd call them "synthetic" frames. In the same way that synthetic vanilla is not vanilla, or synthetic rubber is not rubber.

[@GamersNexus]

That's pretty good.

[@ElijsDima]

All frames are synthetic though.

[@zivzulander]

​​​​​@@ElijsDima I'm just talking about relative to conventional frames. If you take "synthetic" to just mean "synthesized", then yes, all frames are generated or synthesized, but in this case the goal is to imitate or replicate what the traditional raster pipeline produces. "Synthetic" sounds like a better word to use to me because the negative connotation of "fake" may not hold up very well as AI technology improves, while many "synthetic" alternatives are often but not always respectable alternatives to the original for certain properties, and/or have their own emergent uses (synthetic oils, fabrics, music, etc.)

[@Emma-i9x]

The thing is, synthetic stuff can be bioidentical to the normal stuff, unlike AI generated frames are.

[@Emma-i9x]

​@@zivzulanderthen conventional vs ai?

[@Rain1]

man i hate that frame gen has become the crutch instead of optimization

[@JayceCH.]

Look at Monster Hunter Wild's recommended specs. Its crazy..

[@Rain1]

@JayceCH. That one makes me quite sad. I played the beta and oh man how badly I wish they just cared about optimization instead of making money. Game is fun but it runs like "crap" on my 3080 and 8700K

[@RobertZ1973]

If you want to optimize a game for your toaster there is a wonderful section devs include called graphics settings, maybe navigate over there and adjust settings to match your toaster.

[@3dcomrade]

​@@RobertZ1973 maybe you should play some og crysis 3 to appreaciate optimization

[@RobertZ1973]

@3dcomrade Pffftt.. been building professionally since the days you were nothing more than an afterthought and a tragic mistake, sit down.

[@imakuniaw]

Ty for the summary, its 1 AM and I really need to go to bed. Thx steeve

[@sammiller6631]

Do you really need to go to bed? Don't you remember the "Just Buy It" article from the editor of Tom's Hardware where he says "When your whole life flashes before your eyes, how much of it do you want to not have ray tracing?"

[@no-trick-pony]

I love seeing Patrick/your staff more on camera!

[@paincake2595]

Nvidia has set a new low with barely any uplift, making partners suffer, artificially limiting gpu supply, so basically helping the scalpers and they lost a lot of stock.

[@bass-dc9175]

The stock market does not care about those. What is more likely is the Tarifs on TSMC which caused the drop.

[@psychickumquat]

​@@bass-dc9175 Possibly Deepseek as well, apparently they released a report on how they trained their AI more efficiently on lower end cards, so now the 5090 isn't as enticing as it was prior.

[@Wobble2007]

Anyone with any sense would totally ignore Nvidia's 5 series anyway, you're a sausage if you fork out for such terrible hardware, vote with your wallet.

[@OdinAlgeron]

@@Wobble2007 for non normies the 5080 is a very good OC card

@@Wobble2007 That's why I'm sticking with my 4090 that does all I want. As far as I know this will be my last card, until it or I die.

[@Edward135i]

10:50 see Steve's hair is more like a 90s swedish death metal band where Brian's hair is more like the Beatles in the 1960s on the Hippie trail to India searching for a guru.

[@zivzulander]

This is what we call an unassailable comparison

[@Maco_im_Sacko]

​@@zivzulander you had me at ass

[@brocklewis7624]

Finally. I think this in depth review is the most important part of evaluating this new generation.

[@martixy2]

An interesting idea for examining frame gen through the format of a youtube video: For 2x, for example, you could cut out every other frame of a capture and then edit together only the real or fake frames, and see how different the video looks like. That would be cool to see IMO, if at all possible.

[@redaffix7320]

Love when Steve and co. do some in-depth research into the facts that confirm my opinions... so I don't have to. Also, Cyberpunk looks so nice with DLSS off or even set to performance. Would love if TAA could catch up. If you want to see how bad TAA can get, play the Guitar in Cyberpunk and watch the little "noise shadows" that appear when V's fingers move around the frets and strings. If you do look at it, though you'll start to see it everywhere. Turning quickly in games with no motion blur, watching corners while turning, going up and down ladders in first person, etc. Another great example is the latest Indiana Jones game. It's EVERYWHERE. It looks like an SCP trying to sneak up on you.

[@sunderkeenin]

There just isn't a way to make TAA good. It is fundamentally incapable of not blurring because blurring is the mechanism which defines how TAA works.

[@magicalchicken5667]

Rotational camera movement is actually very easy for TAA to handle, a lot of games just seem to not have it setup properly. First person meshes like hands moving can get a little tricky depending on the data given to the shader, but I'm pretty sure that's solvable too. TAA is quite advanced but not everyone does it right, the method itself gets blamed for bad implementations sadly. DLSS has the advantage of AI to help its temporal aspect so it's not really easy for traditional methods to catch up.

[@borealeone]

DLSS is TAA , it is just devs dont bother not just with alternatives for people who prefer motion clarity to anything else, but also just basic optimisation of non dlss TAA algos (because uncle jensen does the job for them, so why bother)

[@magicalchicken5667]

@sunderkeenin "Blurring" is an oversimplification by people who don't know how it actually works. When not messed up or not used purely for denoising (a lot do that one) you can have a very crisp and clean image with TAA, both static and in motion (yes, TAA can actually clean up rendering, thin lines in particular, thanks to the scene jitter and extra data it gets from that before correcting it at the end) Games just need to stop using it as a crutch for denoising, it wasn't designed for that so it gets the worst of both worlds there. TAA is here to stay because it's genuinely good, there are no proper alternatives to what it can do that don't have major downsides elsewhere.

[@krisc1684]

​@@magicalchicken5667 another issue is that TAA just doesn't look great at lower frame rates. In my personal, anecdotal, unscientific experience, I don't use it unless I'm getting at least ~80-90 fps. But I legitimately love how TAA looks at its best

[@VysesDarkheart]

Im 16 seconds in and I can already tell its going to be a BANGER!

[@2012ArsenalMega]

And I just finished and wasted all that time.

[@mr.martian7065]

appreciate the hard work but i think the data on this video could've been a short slide at the end of the video for quality comparison, because we are talking about artifical frames, it doesnt matter how many frames are on screen if they look bad, so i think the most important part is how they actually look, cant wait for that video

[@chincemagnet]

Multi frame generation is probably the worst marketing tactic Nvidia has leaned on in a long time. It’s pretty much a refresh, only it took twice as long and the performance boost is disappointing.

[@feloniousjones8834]

nvdia 60 series coming soon with "Supercalifragilisticexpialidocious"

[@borisvokladski5844]

It sounds like an avanced Harry Potter spell

[@andersjjensen]

@@borisvokladski5844 It's from Marry Poppins, which predates Harry Potter by a handful of decades.

[@Stefan_Kawalec]

So money-wise 5090 should be no more than $1000.

[@ElderGodDragonFurry]

499$ the most expensive versión

[@iLegionaire3755]

Pigs will fly and cure cancer with their orgasms before NVIDIA lowers the price of the RTX 5090 to $1K ;)

[@DelgaDude]

Why stop there? Just make it free!

[@SnakebitSTI]

Nvidia should be accepting a few fake dollars for every real dollar.

[@hansolo8225]

The $2000 RTX 5090 is a waste of fab capacity that could produce a $30,000 AI Card instead.

[@thisposter]

To best present the fake frame data, you should use a stacked bar chart of the real frames for that model followed by bar of fake frames (which would visually show the cumulative number of frames vs the control). The label to the right should be the total number of frames followed by the percentage of uplift that cumulative total of frames over the control no-fg run. If the bars are very zoomed in like the current video, you could have the number of fake/real frames labeled within each bar. Having that percentage for each run would allow for easy comparison between the models of cards to see if the model is taking advantage of the different hardware in different ways.

[@hochhaul]

14:44 Nvidia just can't help but scheme like this. It's one of the more annoying aspects of the company. Take an interesting or valuable concept or tool that already exists and try to turn it into a proprietary or closed source thing that they can use to their advantage. I get that it's just a part of competition but it's bad for for customers. How long before Jensen resorts to custom power supplies that output a unique voltage their cards need -- but wait -- it has some kind of G___ technology in it and power supply manufacturers need to pay to have a license to use the chip and only PSU's with this G____ technology can properly power the RTX 60 series. The PSU's are more expensive but Jensen says this had to happen and these awesome new cards just wouldn't work right with a standard ATX power supply.

[@briannormant3622]

When AMD motto is "Never miss an opportunity to miss an opportunity" you can get away with anything.

[@hochhaul]

@briannormant3622 No that's just a motto popularized by smooth brains. Nvidia was doing this stuff even when Radeon was doing well and had much bigger share of the market. Jensen has always weaponized the use of proprietary, closed source, closed libraries, licensed tech derived from open standards, etc. That's the benefit of being the brand of choice for default-brained gamers for nearly 30 years.

[@lukeemerson7384]

doin the lords work my friend. keep it up. you and your team deserve an award for the amount of consumer information you have given the common man. thank you all.

[@Vashtiian]

I was waiting for this one Steve! Thank you 😊

[@rainkloud]

Great work here and I will definitely be coming back to reference this as it’s just so chock full of quality info. One thing I would push back against though is your comment that MFG isn’t fixing a problem. I think gamers with high hertz displays want to be able to take advantage of those without waiting a decade for raw gpu power to catch up. 60 fps is the new 30 fps in terms of acceptability. Sure it’s playable but 120 fps is just such a massive upgrade and 240 delivers exceptional motion clarity that has a dramatic effect especially when coupled with bleeding edge graphics. So I would say that allowing gamers to take full advantage of their displays that would otherwise not be possible is indeed worthy of the description of fixing a problem.

[@saleplains]

the tortured look on patricks face really sells his segment

[@Genesisfury]

22:46 I need my frames to be virgin frames. Pure frames. Nothing less will suffice.

[@dead-claudia]

18:16 i hope this benchmark rant gets shared within nvidia internally

[@caschque7242]

I think it's a good video. Though I mostly just watched the conclusion, I wanted to add that you do not just see it in black and white with multi-frame generation. For me, whether it's 2 or 3 extra frames does not really matter. However, I think someone who wants to play with smooth 90-180 frames per second and needed to lower their settings significantly can now raise the quality further to achieve that. He would need to ensure playing without multi-frame generation at around 50-60 frames-or wherever it feels smooth enough-and then activate the new technology. I think that is actually valuable. It’s not a miracle solution, but it’s a useful option you can consider.

[@smiIingman]

The biggest disappointment is nvidia saying we can force mfg into games. But turns out its a whitelist system which defeats the whole purpose of it. I am furious

[@AFistfulOf4K]

profile inspector doesn't let you override?

[@PutYourQuarterUpGaming]

Why are you surprised tho? DLSS, in all iterations, is a whitelist system.

[@PutYourQuarterUpGaming]

​​@@AFistfulOf4Kno, devs have to implement it, but unlike FSR, they also have to pay to license it. Why do you think cyberpunk basically become a tech demo for them? They're footing the bill...

[@DanielSilkowski]

Worth Considering their Geforce 'Cloud Gaming' model in all this. Who cares about hardware when you can sell your own ? They have a vested interest in anti-consumer practices.

[@GamersNexus]

We should probably go back and re-test that. It's been a while.

[@n0madc0re]

With all that debacle, I really ended up thinking that inciting gamers to turn to GeForce Now was the plan all along.

[@clarenceoveur9497]

Loving Patrick's segments, his cadence brings a great balance to Steve's delivery!

[@heavenlyavenger]

Great approach on the topic! I liked how you're going to split in parts, I was really looking forward to your take on MFG and am really really excited on what's to come! When I commented about this in the 5090 review I was not sure how much GN team would care about the feature, turned out a great deal! I'm sure you're going to find fancy ways to get real "shots" on what's going on these magic frames! Wonder if that's going to be possible to reliably capture frames and discern between real & fakes. Maybe if the game is played in "slow motion" it could dump the real frames to be looked one by one, highlighting limitations on what can be "educately guessed" by mr. AI?

[@10secondsrule]

The current CEO’s like Jason, bezos, Zuckerberg, musk are so detached from reality that listening to what they say is like listening A.I. hallucinating. I don’t bother watching any of it anymore as it makes no sense and just gets me kind of angry.

[@_GntlStone_]

0:37 They even used Frames per Joule without attribution!

[@talald342]

Now this is what I watch gamer nexus for! Actual pc hardware journalism and not just influencing. Great video guys.

[@MrCrushcode]

Many thanks! The video actually is exremely educational about current state of "fake frames" situation.

[@woodcutter9]

RTX 50 series ❌ Benchmark Methodology ✅ Benchmark Mythology