NVIDIA RTX 5090 at 575 Watts, RTX 5080, 5070 Ti, & 5070 Specs

[@GamersNexus]

Watch our AMD RDNA 4 coverage -- because apparently AMD let everyone else announce it: kzbin.info/www/bejne/jHSoZHSLhMaUmJI Get the LIMITED 2024 Disappointment Shirt in Cotton: store.gamersnexus.net/products/disappointment-pc-2024-oxidized-design-cotton Or in a highly breathable Tri-Blend! store.gamersnexus.net/products/disappointment-pc-2024-oxidized-design-triblend

[@perpetualprocrastinator]

I miss Jensen's kitchen

[@Prestiged_peck]

lobster courtesy of ZTT🦞🦞🦞🦞🦞🦞🦞🦞🦞🦞🦞🦞

[@AndyViant]

You better create a 2025 CES disappointment t-shirt after this cluster f***. Perhaps you could do an XXS logo on it and then upscale it by 1500% to a normal sized t-shirt. Add some glitching and sizzle, muddy out the textures... it'll be a hit :D

[@manitoba-op4jx]

i was looking on a userbenchmark alternative i frequent when evaluating PC hardware and their list has an RTX 5050 THE 50/50 MEME IS REAL the site is called technical city if you need a source

[@dwarfer1]

GET YOUR ASS TO CES ALREADY!!

[@Bottlecap]

I’m extremely confident that the 5070 will not *actually* be better than the 4090.

[@Sunedosa]

Can't say that i belive that either BUT the 5070 TI will definatly be on par with 4080 super apples for apples for less money than i paid for my 4070 TI. That's a win for me atleast, i was expecting way higher prices.

[@MrVvulf]

I'm looking at the whole 5000 series and wondering...why? Games have been so trash lately there aren't even any that justify the prices of top end video cards.

[@RafiYagami]

Not at traditional raster fs, but for dlss frame gen titles it will be as it can do 3x fake frames. They say latency and quality is improved as well

[@WardenOfSouls]

im pretty sure that meant "with DLSS4 multi Frame Generation enabled" since thats only supported on RTX 5000 Series as far as i understood. When it spits out 3 AI generated Frames per real Frame it could end up somewhere around 4090 with number of frames but who knows what the Quality will be with that many "fake" frames

[@hasiktak1544]

​@@MrVvulf its for the rich folk that dont care about the money

[@MikhailZuev-k1x]

I wish Nvidia could take my 100$, upscale it to 1000$, and frame generate second one up to 2000$ in total

[@rapgod9447]

dude wtf I am dying 🤣🤣🤣

[@gibster9624]

And then you can have a 5090

[@HaasSpitta]

Basically bitcoin chart 📈

[@vlad1529]

My as well use monopoly money too

[@alexandercheskis]

get that reality upscaler

[@virescenticious]

I know it’s a dead horse at this point, but first party benchmarks pushing frame gen and upscaling stats instead of actual raw performance is insane.

[@tejendrachettri5970]

Get used to it lmao

[@D1amond131]

Nowadays game devs and nvidia don't care about optimising stuff. They just rely on dlss and the unoptimised tools unreal engine uses. They need justify their prices with poor optimisation so better gpus have more performance

[@garydiamondguitarist]

Agreed. People testing GPUs need a dual column setup, with the first featuring all the gen/scale stuff on, the second with it all turned off. Fake frames aren't real frames, they don't count.

[@T33K3SS3LCH3N]

Obviously they will pick the settings that make them look the best. But I still think this is fine as long as the frame gen+upscaling have good enough quality. I'm running quality-mode upscaling and frame gen in nearly every game that supports them since the artifacts are so minimal that I'm fine with using them even just to lower GPU load/power consumption. Seeing GPU load drop from 100 to 70% feels nice.

[@edgargranados1676]

@@D1amond131But DLSS in itself is an optimization strategy. It uses ML to generate neighboring frames.

[@SairaMay-c6k]

The more of a GPU you buy, the shinier my jacket gets" (Uncle Jensen, Circa 2025)

[@Lee-SR71]

His jacket is made from the missing vram chips from the 50 series gpu's

[@MrMega200]

Reddit found out his jacket is worth 4.5 times the amount the 5090 is MSRP for. Jacket is "worth" $8990

[@mchonkler7225]

​@@MrMega200That's just his walking around jacket. He saves the really expensive ones for admiring himself in the mirror.

[@Jlomb436]

This is such a sick burn. Love it.

[@Squilliam-Fancyson]

Nope. Those are the 256bits of mem interface on his jacket, cut away from the 5080. 5080 has less mem bandwidh than a 3080ti with older GDDR 6x

[@scottb4579]

@@MrMega200 Start buying Nvidia stock and soon you too will be able to afford $9k jackets. Just from investing in Nvidia 5 years ago, my stock can pay off my house this June.

[@4Wilko]

"5070 notably has 4090 performance at $549." What? No way, that's bullshit. "This is impossible without AI." Oh. So it's bullshit. Thanks Steve for all the updates.

[@garydiamondguitarist]

I really enjoy how this comment makes AI a synonym for BS. Seems about right. I'm still wondering when that bubble is going to burst.

[@FafthriechRalofson]

literaly the inner monologue going through my head while listening to him deliver that line

[@-PVL93-]

​@@garydiamondguitarist apparently the token query prices are crashing and the training progression has slowed down,so much so that all the AI model are now being designed to do introspectives, plus the entire IT industry raced into the segment. It's about to burst

[@SmileyXY]

I mean... DLSS and FG are also AI technologies and they're certainly working pretty well, so it's not that absurd.

[@Thegooob95]

So is it bullshit when cyberpunk runs at 40 fps native on my laptop but runs at 80+ with max setting and psycho ray tracing and LOOKS BETTER than when it runs native? Is that bullshit? Only bullshit I’m seeing is these ridiculous takes on dlss and ai. Obviously ai is the future of graphics and games. I mean come on man, the way nvidia is doing it will mean LAPTOPS will performs like a desktop 4090. The way you want, with raw power only guarantees the barrier of entry to be insanely high, and guarantees that none of this will scale to laptops.

[@mr.fuffeyguy]

My favorite part was when he said “let’s talk about AI.” after talking about nothing but AI for 30 minutes straight

[@Real_Mick3y6]

I hate how broad the term AI has become

[@bogdanharmandzhiev]

@@Real_Mick3y6 these days its all about the AI and the sweet sweet money it can provide! Just say it has AI and BOOM! YOu can put twice the price tag.

[@devilmikey00]

@@bogdanharmandzhiev Problem is, outside of people like Nvidia who are selling all the AI hardware, AI doesn't make dime. Every time someone uses ChatGPT it costs openAI money. There is literally no profit in any of this and it doesn't look like anyone has a plan on how to to change that. It's a giant money pit propped up by investors believing conmen like Sam Altman. A year from now the whole tech industry is going full 2008 economic crash.

[@Stevenpwalsh]

@@Real_Mick3y6 It's not really broad, AI is just running everything now. Using it every day at work, it's like the digital transformation of the 90's.... but it's happening 10x faster.

[@Stevenpwalsh]

@@devilmikey00 My company makes money from AI... LOTS of money. We're not selling the AI, we're using it. I think that's the key, all these models are essentially the same, they're commodities.... and priced like it. But what we can do with the AI is worth a lot.

[@Alxium]

None of this changes the fact that I will ONLY buy a GPU based off of native rendering performance. Please keep native rendering a priority Steve, those numbers matter a ton regardless of how much these companies want to ram AI down our throats.

[@eightymopar]

Yes. Upscaling and frame gen is the biggest scam both amd and nvidia has done. And they have everyone convinced and eating out of their hands, that this is the "future". If this is the future, im just gonna get a different hobby 🙄

[@lucariotrainer16]

@@eightymopar Upscaling has been a thing for games even on a software side. There is nothing wrong with DLSS as JUST an upscaler, especially when DLSS makes games on average look much better than TAA does. Anything past that is stupid yes

[@Stonewall999]

This

[@davidy22]

Graphics have long been hitting the cap of how much they can push hardware, when LLM powered dialogue starts hitting games that's going to start being the new bottleneck on your GPU

[@fus132]

@@davidy22 _if*_ LLM powered dialogue starts hitting games

[@HussPOE]

Prediction...That "get used to it" clip is going to be HEAVILY used by GN this year

[@SirCamWA]

I hope so

[@nibelungvalesti]

Go play some Path of Exile.

[@haukikannel]

We are not worthy!

[@FernDaawg]

“Let this sink in”

[@vensroofcat6415]

@@FernDaawg nah, leave that sink outside

[@evilmadness8052]

They are going to drop a 5080Ti in here at 1500 bucks with 24GB's you can see it coming a mile away

[@Prophetian]

Wow, such a bargain price, only $100 less than 4090!

[@504Trey]

That's a great price 😎

[@wilburtshnookums]

@@Prophetian cant wait to only pay 9/10 of a kidney instead of a full one!

[@IAmHanYolo]

That's the one I'll be getting.

[@decree72]

@@wilburtshnookums how many kidneys left after the last gpu

[@frangipani-dreams]

Nvidia are going to get shredded for that 5070=4090 claim - but they won't care.

[@jenkeized]

Why care when you already won

[@NameForgotten1]

People will buy the 5070 and believe it has 4090 performance. That's the Nvidia mindshare right now. It's pure brand loyalty.

[@islandfireballkill]

No, they won't. The 5070 will be flying off the shelves. I'd bet 100:1 odds on this.

[@jojothepro15]

@@jenkeized Because legacy is far more important than short term gains.

[@Clooger-]

@@jojothepro15 Tell that to the ldiotic bootlickers who buy anything Nvidia spits out.

[@jontanneguy4960]

Remember when they said lower prices right before launch, everytime? And then they jack up the prices and claim "shortages"? Yea, me too.

[@Vincentos1109]

I am very excited to see the niche case in which the 5070 is going to be as powerful as a 4090

[@dragontales1999]

Reminds me of them lying and saying the 4070 Ti would be 3x faster than the 3090 Ti

[@RyanFennec]

@@dragontales1999 yeah hahaha i remember that meme

[@Renk1]

Maybe the AI will generate more VRAM. You know, 1 real GB of VRAM and 3 fake GB's.

[@60000L]

on all of the NVIDIA graphs they state "DLSS SR (Quality) and DLSS RR on 40 Series and 50 Series; FG on 40 Series, MFG (4X Mode) on 50 Series" MFG = Multi Frame Gen = the new DLSS4 Generate-3 frames-from-1 frame thingy not really a good comparison from NVIDIA as usual lmao

[@Subliminal_Panda]

😂 it'll be like three titles that support the new 3:1 frame gen. Fake frames don't win games, Nvidia!

[@smeezer]

The jacket perfectly sums up Jensen Huang's attitude towards gamers... "Get used to it."

[@DDMSakb48]

At least, he was having fun🤣🤣🤣

[@eliporter3980]

I've already accepted our new Nvidia overlords. I'll be getting a 5090 day 1.

[@billytk1225]

@@DDMSakb48 Not sure he even knew where he was. 🤣

[@PeacefulYakuza]

He forced nothing to anybody. I dont blame him. People needs to know u dont have to buy any new product every God damm time

[@PyromancerRift]

You know, people can chose how to spend their money. Nobody is traped

[@bliglum]

YO DAWG... WE HEARD YOU LIKE FAKE FRAMES. SO WE PUT FAKE FRAMES, INSIDE YOUR FAKE FRAMES

[@ericwheelhouse4371]

For real

[@GramDraws]

😂😂😂

[@gosuprime]

Just imagine, DLSS4 + lossless scaling, *infinite* frames, 20fps to 20000fps.

[@BrwrFan806]

wonder what the input lag will be

[@BeanieKing]

YO DAWG... YOU NEVER BOUGHT ANYONE ELSE, SO NOW YOU'RE STUCK WITH US. WELCOME TO THE FUTURE YOU HELPED CREATE 😂😂😂

[@RiyaConley]

It's ridiculous that the 80 class is now only half the size of the complete chip. The 780 was essentially a Titan with the same bus and rop count but a small tpu drop.

[@defranken]

They figured out people will pay anything for the top card during the GPU shortage. We are never getting back the good old days of 80 class cards for 500 bucks

[@ChoJun69]

@@defranken 70s 500, 80s 700, 90s 1300. Would still be super expensive for my poor ass, but i would definitely jump to 5090 if it was like that, now?? Fuck that lmao. 4090 still costs 2.3-.2.6k here in Lithuania, 4080super ~1.3k..... these cards will be obnoxiously expensive .

[@enzo1st]

bro, you posted using the scam account

[@Directx11cool]

@@defranken You will never return it because of inflation. Have you ever heard of this?

[@defranken]

@@Directx11cool no

[@DoctorMandible]

Reminder: The last time the '70 series of the new generation outperformed the previous top card was all the way back with the 980 ti vs 1070. Nvidia (and board partners) did NOT like this because they had lots of 980ti left to sell. Don't be surprised when the 4090 outperforms the 5070. That's by design. The new gen sales can't completely cannabalize the previous.

[@devilmikey00]

I mean we already all know that claim is only true with all the frame gen stuff turned on. They had it in the small print on all those charts. With all that off I doubt it even out performs a 4080.

[@matth7118]

Nvidia stopped making the 4090, as well as other 40 series cards, back in October. This is not an issue.

[@Osprey1994]

From the polls I've seen, people don't upgrade as fast as a lot of you guys. A lot of regular people like myself wait 2-3 generations. In my case, I would have liked to wait for the 60 series, but gaming tech has made some relatively decent jumps, and it's causing performance issues for me already, so I held off touching the 40 or 40 super, and I'll hopefully pick up a 5070 or 5080.

[@22Epic]

Nobody said that the 5070 would outperforms the 4090 overall. It will with some specific conditions. Also it doesn't really matters as both cards don't target the same audience and nobody will ever say "Oh I was thinking about buying a 4090 but I will buy a 5070 instead" (and if someone does you'll know he is just lying).

[@sznikers]

It didn't really, similarly to current situation there were some caveats to that claim although not that shameless. 1070s outperformed 980Ti reference blower with stock 1000-1100MHz clocks and assuming games use all 10 series new tricks. While 980Ti were clocking like crazy, custom models that people bought were released with 1250-1350MHz clocks and 2 slot proper coolers, those were little better than 1070s in most titles. Nvidia took care of it over few years of driver updates thou 😅 I had 980Ti back then, it was 1350MHz out of the box and could be pushed to 1415, was also cheaper than 1070. Bought it months before 10 series release and it went up in value after 1070 reviews arrived so i wasn't alone in that opinion. Swapped once the real deal arrived in the form of 1080Ti.

[@DjPolo-f2d]

I remember Nvidia eluding to the fact that a 4070 was three times faster than a 3090, we all know how that turned out. At this point I don't believe a thing they say.

[@FatherLamb]

Yep! I had the 4070 ti super & a 3090 in hand to test my self. I can say.. no. It was only more powerful in "SOME" things where you could turn on everything known to man to TRY to help the FPS, other wise? Without all the cruch, it was behind by a TON.

[@Knives7777]

You should never take any corporation at their word. Dont trust and verify to confirm

[@DjPolo-f2d]

@@FatherLamb For sure, I remember most reviews basically said a 4070 was functionally the same as a 3080.

[@bofasofa9399]

They also said the 3070 was as good, if not better than the 2090ti. They were correct.

[@alonsogarrote8898]

so if 4070 is close to a 3080 then the 5070 will be like a 3090?

[@LynxBite]

Them calling native rendering "brute force render" is actually insane

[@BombaJead]

It's double the insanity when we consider that basically the way "AI" works is through brute force training.

[@Ober1kenobi]

Gotta hack those polygons

[@Carl_dev]

Yeah, they'd rather you look at a shittified, upscaled version of the same game, because it's much easier to crank high fps that way and it's easier for them to improve DLSS tech than actual rendering performance.

[@Scytian1]

Yeah, with new Nvidia technology we will get textures with upscaling artifacts, upscaled to native resolution and then 3 frames generated out of that one upscaled frame, so we will get artifacts upon artifacts upon another artifacts. Nice, you can already go back 7-8 years and see how much sharper and cleaner games looked back then, everything is going to shit.

[@Cinnabuns2009]

Can't wait to see the latency induced by MFG. FG is already bad enough for Dark Souls to become even darker souls. MFG + Super ULTRA Low Latency mode = still laggy?

[@Stinksatthis]

Love the videos, so much deeper than the majority of channels

[@davidquiram7339]

"I thought I was worthy", I feel, is a clip that will get a lot of mileage this year.

[@SanctusBacchus]

I hope not, it's doing exactly what they want us to do.

[@rezaimran98]

I highly doubtful scamming people would make him worthy. But I don't know the rules.

[@Bunster]

@@rezaimran98 nobody is forcing people to buy these luxury goods

[@sarjannarwan6896]

@ Where's the scam? Ripoff you can argue about but you know what you're buying, something being expensive isn't a scam.

[@supertuberful]

You know, he was suspiciously static in hope that people will get to snap the image for meme. The corporate planned that. It wont go far unless paid

[@RavTokomi]

The difference in cores from the 90 class down to the rest is insane. You could have a whole lineup of cards just between the 90 and 80.

[@amunak_]

...that's probably where the actual line up would / should have been if there was any competition.

[@ying0878]

I think 90 series is targeted towards corporations while the others are for consumers.

[@goldenhate6649]

@@ying0878kicker is, its not. The A series is for corps. These cards are horrific for corps and AI use because the differences in how the cards work. Gaming cards guess values, they don’t give real values

[@cr0ss414]

Probably because in that range is where they're gonna shove the 5080 Ti/Super/Ti Super when they need a mid generation refresh

[@imitenotbe]

@@ying0878 the 90 series replaced the Titan line, so yeah it's always been that

[@petroslinna]

Thank you so much for fast turnaround on this video!

[@GamersNexus]

Thanks for saying that! Long haul today, love this kind of work!

[@cs-cl9qs]

@@GamersNexusIt’s greatly appreciated

[@Flameancer]

When I didn't see an immediate upload I thought you'd wait till tomorrow. Glad I stayed up, its 130

[@vovap9257]

@@GamersNexus For real, it is shocking that you've managed to get this video up so fast. I just watched the keynote with my friend and we were laughing at the absurdity of it the whole time.

[@sevenofzach]

I suspect the real heroics are testing right before launch when results aren't making any sense

[@HaoHaki]

You explained this so well, thank you!

[@Skater6453]

Very fitting that Jensen went from real leather to fake leather. To go along with the 50 series.

[@CptJistuce]

It's DLSS leather.

[@dawiepieterse9703]

@@CptJistuce with the help of A.I.

[@Skobeloff...]

So not wearing a corpse is somehow a step backwards to you? Did he diss your cave?

[@fdevaa]

@@Skobeloff... I'm selling fresh grass if you're interested

[@CahyoPrabowo]

It's shiny ray tracing leather

[@rcvillapando]

4090 perf for $549? We have the January Disappointment Award 2025 locked in already!

[@theskiff]

Technically February but yeah

[@Boss69292]

Its with framegen and dlss 😂

[@JerryFlowersIII]

Looks like that claim is only because of MFG. Like for like will be very different. Personally I haven't played a game where Frame Gen was really worth it aside from CP2077 with Path Tracing.

[@artursobkow2680]

It's New frame Gen, probably more prerendered frames Are generated from one real frame.. By the look of it, 5070ti is only better by 20%-25% than 4090 in raw real render generation. If you look on only RT comparison in left side of the 2 Games so no dlss on only raw power of the card.

[@captainlooney4356]

5070 is in Feb... only 5080/5090 is January 30th

[@yasuchika]

I'm not a fan of compromising hardware with software solutions, 12GB VRAM on the 5070 remains unimpressive.

[@HelloImCrimson]

It doesnt matter. If they lower VRAM requirements of DLSS and FG like they promised, 12 gb will still be enough to play anything maxes out at 1440p

[@honkhonk8009]

True. But lowkey the gaming industry does need to be more open to AI stuff. Im mostly scared how video game devs are gonna use this as an excuse to churn out slop-ware. Already with Nanite coming out, we see bad LOD's and a complete disregard for basic optimization. Same with DLSS and TAA.

[@GewelReal]

12 is more than enough for anything outside of the most unoptimized games at 4K Ultra (if you need more just get a 2nd hand 3090 Ti)

[@nipa5961]

Just another scam generation.

[@skyscapedrones2485]

100% agreed, I feel many would lean towards the 5070 ti for this reason, it could actually be a 4k card with all the new AI technologies

[@vasheroo]

Looking forward to "get used to it" clip being added to list with "thanks steve"

[@Raytheus]

I sincerely appreciate you rounding up the price to the nearest multiple of 5. XX99 does seem somewhat superfluous compared to XY00.

[@GamersNexus]

Thanks for noticing. We've long had a policy of rounding to the closest whole number and dropping the silly 99 act. We actually made that change on our own store a while back for the same reason.

[@Edgedick]

I don't comment often, but thank you for being reliable and consumer-oriented in a sea of exploitative shitters. You guys are truly the good ones

[@BladeCrew]

@@GamersNexus I mean it's sounds more simple, less use of number and waste of space. Also I love watching your video Steve. I am from South Africa.

[@flyingping]

I thought .99 was a Jewish merchant invention.....

[@ShadowRune]

I do the same thing when I'm selling stuff you'd be surprised how often it breaks people's brain over here in the USA because they're so used to the 0.99​@@GamersNexus

[@Just_An_Ignacio]

The 5070 > 4090 is the same we got with 4070Ti > 3090Ti. Obviously using agressive upscaling and the locked frame gen technology of each generation. But somehow people still believe it.

[@Skelath]

Yeh on a 4090 the Frame generation is case by case, on some games it's fine and on others such as Starfield it produces a "glass ghosting" effect on movement.

[@DimkaRofler]

3090 Ti and 4070 Ti both have 40 Tflops. So objectively we can say that we get a cheaper card without losing performance. Even without DLSS and generating frames. But... 4090 80 Tflops vs 5070 with 30...

[@garydiamondguitarist]

I think maybe performance metrics need to have a column for all upscaling/frame gen stuff ON and then another where it's OFF. Then people might see how much of the money they're spending on a new GPU isn't worth the premium.

[@raikaru0]

The 4070ti is in the same performance bracket as the 3090ti

[@user-wq9mw2xz3j]

Nope. 4070ti was actually close to 3090ti. 4070 though wasnt. and 5070 wont be either.

[@ForzaE2]

This comment is not Ai unlike CES

[@GamersNexus]

Seems real!

[@RyanFennec]

what was the total "AI" count by the end of the keynote, i gave up counting once it was around 120 times that it was used

[@Tubofmayo]

@@RyanFennecNvidia was 205 or something

[@BeastOrGod]

Yeah? Then what does this captcha say then? J̵͚̪̘͍͕͊̍͋̀̊̄̍̉̎̕e̴̡̛̺͙͈̼̫̮̰͉͓͇̻̼͙͗̆͒̋̒́͊͛͠n̶̛̬̮̫̦̾̒̈́͐͗̿͌̀̈̚͘͠s̶̬̓̔̂̔̎͋͒̚̕ẹ̷̢̠̦̳͍̻̠̯̀̀̑̎n̷̛̪̩̖̙̗̟͍̗̊͒̓̌̀̇́̌̊̓̕̕͠ͅ

[@Altamyr_]

'Father'

[@timothyvandyke9511]

Imagine trying to rebrand traditional rasterization as "brute force" 🙃

[@kekeke8988]

It was a ray tracing to ray tracing comparison, no rasterization to speak of.

[@Fluskar]

<a href="#" class="seekto" data-time="42">0:42</a> we got captain taiwan bruh

[@Toast10i]

I hate the fact I looked at this comment right before the time stamp happened.

[@Techandgamingreviews]

Lmao ☠️

[@deadangel03]

LMAO

[@listerinekiller]

LMFAOOOOOOOOOOOOOOOOOOOOOOOOOOOOO this comment so underrated

[@DagobertX2]

Captain Cringe

[@StatsGam4er]

Nvidia comparing RTX 5070 as having same performance as RTX 4090... with 4 x the fake frames...

[@alexaka1]

And 0.3x the RAM

[@Xurikyo]

yeah me thinks the 4090 will atleast be a 5070ti/5080 performance or the same caliber excluding vram shenanigans. I hope to see price drops but we all know nvidia.

[@emma6648]

How else is he supposed to scam people into spending a ton of money? He needs that new jacket man

[@some-replies]

I hate dlss and frame generation.

[@RyanFennec]

i love input lag and artifacting bro wym!?

[@reindeerdashie]

Only 1 in every 15 jackets were actually real and rendered on that stage

[@flyingping]

"Rendered" is a great choice of words....

[@neilranson4185]

Hopefully, why be so inefficient?

[@selectionn]

I couldnt even see his jacket because of all the DLSS ghosting and artifacting, guess I need to upgrade to the 5090 for DLSS 4 to see it, maybe frame generation too.

[@zivzulander]

The jacket is real, it's Jensen who is rendered. That's just a robot on stage. The real Jensen is in Tahiti. 🙃

[@reindeerdashie]

Rumor has it that the next generation of kids following Generation Alpha will be called the Frame Generation.

[@HisOka-h8c]

Great video! Really enjoyed it.

[@FrencisWhite]

maybe in couple of generations we will have to render only the first frame when launching the game and the rest will be ai generated ^^

[@Slav4o911]

The games will be AI generated, no "normal frames" at all, just a text prompt.

[@alexanderpopov3587]

A robot will play the game for us

[@jayb2705]

Bots playing bots, just like Fortnite

[@on-quest]

Already been done. Doom played on a neural network simulating Doom game engine in real time: kzbin.info/www/bejne/pme1q4aVotZgpqs

[@ipodtouchiscoollol]

Honestly I wouldn't even care, as long as it plays like a game and looks like a game it is a game. How the game runs under the hood has nothing to do with the quality of the game itself.

[@migiola]

NVIDIA 2028: 'Our RTX 7090 GPU using 0.03 nm technology which we managed to fit in just a 5 slot card, with AI will display for every 1 real frame another 99 hard coded preinstalled frames, reaching 400% of the performance of a 6090, for only 19'000 $. An absolute bargain for everyone who paid 15k for the 6090'.

[@devynj6924]

thanks steve for all you do im still fairly new to the channel and i love learning about pcs much love man

[@GamersNexus]

Thank you! Awesome that you love learning about hardware!

[@spladam3845]

You came to the right place quickly.

[@darrenwilson921]

Exciting! Hopefully your testing will uncover if it’s really worth upgrading too 👍🏻

[@jrp335]

I’ve never before watched a CES presentation until the NVIDIA one last night. I don’t think I’ll be doing that again.

[@jason11wyatt]

All tech channels should review every single gpu from here on out with no upscaling or frame gen whatsoever. Publish raw raster performance, no bandaids. No mention of those features at all. See if the lack of media on those features will cause Nvidia and AMD to focus on real generational uplift rather than new software locked behind hardware.

[@LucidStrike]

Tbf, AMD has focused on raster and gotten laughed at for it by a lot of folks.

[@garydiamondguitarist]

I think it should be all the scale/gen stuff on and then a second column with it all off. If Nvidia are going to be dishonest about specs, the tests need to put the honesty back in for them, maybe stop a few characters wasting so much money on these things.

[@arha-z1v]

but that is what reviewers already do? and if they do not - you should change your viewing habits to the ones that do

[@LOLOLOLOLOLOLOLO7636]

Raw performance isn't relevant though. Why are you stuck in the past?

[@GoalOrientedLifting]

@@LucidStrike thats cause theres fanbois acting like desciples of a company. its litterally brainwashed people

[@real-raiden-ei]

The more money Nvidia makes the shiner Jensen's jacket becomes.

[@defro125]

Jensen brought out the pathtraced jacket this year.

[@arc00ta]

tbh you can't hate on him for it, hes built this company into a crazy giant in the industry and everyone buys his stuff no matter how overpriced it is.

[@TheAmpsych]

"Oh hey Jensen, you get a new hair or something?" "Look closer Lenny."

[@katosjoes]

Jensen has long since been uploaded to a Super GPU. What we are seeing on stage is just a ray traced AI model of him.

[@dpjohnjones432]

Pretty soon we’ll be blind

[@titaniummechanism3214]

I'm mostly looking forward to seeing what cool engineering solutions the guys at Nvidia came up with for the heatsink, the fans and the PCB. If the PCB is in the middle, but the PCIe slot is further to the rear, they are probably going to have to adapt that somehow. Very interesting!

[@th3orist]

i think its telling when he says "if you all object to this, well, get used to it". He played it as if he meant the jacket but we all know what he means xD

[@bazzy5644]

I wouldn't have imagined that FPS would one day become something you can not trust any longer. But here we are. Nvidia made sure of it with a double hit even. I have no words for this. Lets see MFG introduce 3 times the lagg too. :D

[@A_GWAKwork_Orange]

lmao there was also that passive aggressive sikk burn to amd when he said something like "thats like if you came to nvidia and there was no gpus" disguised as a joke about asking for michelob in the michelob building but they only gave him water.

[@th3orist]

@@bazzy5644 tbh i really dont care if the frame is natively generated or through ai. The higher fps number is really there. A game running with 150fps native or 150fps where 50fps are generated makes no difference in terms of how fluid it looks. The only difference can be in the image quality and/or input lag, but i believe there will be continuous work on that too. The issue simply is that we arrived at diminishing returns in terms of how much raw raster performance you can squeeze out of silicon while keeping r&d and production costs reasonable. So if one frontier is reached you set your sight on the next. The time of pure raster performance and native rendering is exhausted. I for one fully embrace this direction.

[@WenMare]

😏

[@or1on89]

@@th3oristthere is difference. And when there will be no difference most of the games will be locked behind proprietary API making hardware monopoly easier. In the meantime here is 400$ more for the real generational uplift.

[@EmoEmu]

575W is pretty wild. I don't like forcing performance improvements by letting computers use more and more electricity.

[@TheInfinityMaster1]

We suffer from HIGH ELECTRIC BILLS greatly in my country....

[@enok7934]

@@TheInfinityMaster1 Just a heads up in case you didn't know, you should clear you channels 'about' section where you say your old account got 3 strikes and terminated. It's against TOS to evade bans. Keemstar is an example.

[@slice0fpi934]

"Get used to it" - Huang, 2025 For real, though. Dennard Scaling and Moore's Law are broken down entirely. There won't be a way forward for traditional non-AI performance without bigger, more power-hungry cards.

[@willmaud2359]

@@slice0fpi934 Silicon could be replaced with carbon nanotubes.

[@gabeschoenfeld317]

For any professional user the increased energy bill will be inconsequential compared to increased work flow. For a gamer in a hot room it sucks, but you can always underclock your GPU and play on lower settings if energy or heat is a concern

[@Kizzster]

Using only DLSS 4 in a benchmark is shameless from Nvidia. FPS is now tripled because of AI generated fake frames from DLSS 4 and what makes it worse is that DLSS 4 is exclusive to ONLY RTX 50 series GPU's. We all know a 4090 can handle DLSS 4 which is more software than hardware... Shameless. Nvidia are at it again with the tricks.

[@HELLF1RE9]

i ask you, what is a fake frame? the frame shows you a moment from the game adding overall smoothness. why the hate on framegen, i didnt see this tantrum when dlss 2.0 was launched, have you ever played with framegen? please do, its a massive performance gain along with dlss upscaling. leaving those on the table doesnt make you a purist, youre just stupid

[@garydiamondguitarist]

Withholding an upgrade that's possible on a previous generation product so people buy the new generation product with it instead is pretty standard tech company malfeasance. If consumers wised up to it and stopped buying it'd be less of a problem. But they won't.

[@gavinbad2371]

To be fair, the 4090 really might not be able to generate that many frames per second

[@TrackmaniaKaiser]

to be honest, fake frames are not a selling feature! If you play anything online. You only get a smoother experience, but you don't get any additional information until a real frame gets rendered again. An enemy player might have walked to the left for a frame. Now Nvidia givs you 3 more frames of the player walking to the left, but the player decided to walk to the right again. All you shots miss, because you have a bullshit feature! If you can't render at an adequat framerate, drop the settings!

[@Kizzster]

@@gavinbad2371 It will if the 5070 can, 5070 natively won't match a 4090.

[@idknothin9]

Thank you guys for being transparent and direct with your reporting. The clarity is great!

[@Artista_Frustrado]

that's the closest Jensen has sounded like non-scripted talking... the man REALLY loves that new jacket

[@tabkg5802]

Can't blame him it goes kinda hard

[@viscountalpha]

You mean the jacket of +20 douchebaggyness?

[@DBTHEPLUG]

​@@viscountalphaYou sound butthurt.

[@Skobeloff...]

@@tabkg5802 🤡

[@Y0Da77]

That's the AI hard at work.

[@fleurdewin7958]

Everytime I hear Jensen mention the word AI ... it means Artificial Inflation ( AI ) in order to fund his purchase of new leather jackets

[@dragonic_rage]

Artificial Idiocy

[@scottb4579]

I'm invested in Nvidia stock so it's also funding my retirement.

[@eddwar8597]

AMD Also mentioned the AI word so much that it blow away its GPUs.

[@triv4555]

​@@dragonic_rage 'Authentic Idiocy' is more like it

[@raybod1775]

Locally run AI is real.

[@noondlyt]

Thank you for the quick turn around and good luck with the AI testing.

[@dubzillaaaa]

Kinda want a 5090 but at the same time, my 3090 still runs all the games I play I just fine.

[@emanuelpellizzaro3681]

So keep it. My 7900 XT is a bit faster, like the 3090 Ti, and I'll not replace it so soon...

[@user-so8pf2gl5v]

It’s never been more important to wait for real reviewers to test the actual performance before buying.

[@DiddysBBOil]

Lmao don't buy this trash

[@alexh7513]

We allready knew how the RTX4090 performs. So the RTX 5070 will perform like the same.

[@americantacos7618]

@@alexh7513 You sweet summer child.

[@ecoista1373]

​@@alexh7513Go away Nvidia bot

[@ecoista1373]

Yes... Keep buying this trash

[@NF-pk5mo]

Remember boys, hold fast until we see the native resolution benchmarks

[@jeffwads]

The native vs fake is hilarious. It depends on what it looks like. The perception of it. Amazing that people that get that simple fact.

[@saveborg1091]

nah bro, they're just giving you a fake farcry native graph. everything else has the trash dlss creating dusty fps to buff up the chart, and i bet they're not even correct, they're inflactionated cos you know... no info at all = big balls. Frame generate my balls

[@vahnwinkle275]

Hold fast?? I can't afford these GPUs lol

[@SMGMP7A1]

No one does, FOMO kills it for 99% of people. Look at games with microtransactions and Netflix stock price as proof the modern consumer is undisciplined.

[@jlllx]

naw, i'm diving right into a 5090 no matter what.

[@NameForgotten1]

If DLSS4 MFG adds 3 frames to every 1 real frame, that suggests that the 5070 is actually 25% the raw performance of the 4090. That's quite the marketing trick.

[@mryellow6918]

How?

[@NameForgotten1]

@@mryellow6918 Well, we don't know if they ran FG on the 4090. My guess is they did not. If the 4090 had DLSS3 FG on, then the 5070 looks a little better. Still, it's probably about 40-50% the raw performance of a 4090 at best (4090 FG ON), and 25% the performance at worst (4090 FG Off).

[@tek_lynx4225]

yea people are ignoring the framegeneration bit on the charts. I can't wait to see the real tests GN does.

[@Jallerblue]

It's actually more like 50% the raw performance of the 4090 because they're likely comparing 4090 with DLSS 3 (1 real frame : 1 fake frame) vs 5070 with DLSS 4 (1 real frame : 3 fake frames). Now, there's more variables here because of the updates to RT and tensor cores, so more than likely it'll be

[@NameForgotten1]

@@Jallerblue Yeah, I agree with you. I did a some quick napkin math above. The best thing we can say is that the 5070 is 25-50% the raw performance of a 4090. Benchmarks should clear this up. Further, Digital Foundry will likely cover the quality of MFG too. Let's hope that MFG is high quality. Time will tell.

[@Rance53]

Thanks Steve

[@LogieD223]

Back to you, Steve

[@eQui253]

Back to you Steve

[@thesajery]

Thanks Steve.

[@proy3]

Thanks Steve

[@mr.anderson2257]

Steve, thanks.

[@graphical_7431]

2 steve uploads in a day. I never thought i would see this.

[@Shadow_Dragon05]

They said the same thing about the rtx 4070 being as powerful as 3090 but we all know how it went

[@mitchjames9350]

And the 3070 being on par with a 2080 ti

[@thejaegerbomber99]

@@mitchjames9350 the 3070 was indeed on par with a 2080 Ti according to benchmarks. Even the 3060 Ti was breathing down its neck. But the 4070 being as fast as a 3090 is BS though. It was slower than a 3080 at 1440p.

[@SirMord]

Gonna be a fun cycle of testing for the start of the year. Looking forward to what ya'll put up cheers GN

[@GamersNexus]

Can't wait for all the stuff to test!

[@CombatMedic1O]

​@Gamers Nexus please don't test the fake frames/pixels. Make a stand. It's gone to far.

[@Ro_Ball]

Imagine how better things would have been for us if the GPU market had healthy competition.

[@billbeis7691]

More like healthy customers and not fanboys .

[@wumi2419]

Yeah, more like healthy customers. As was said many times, people don't want AMD cards to be good so they can buy AMD cards, but so Nvidia drops the prices. AMD just made the separation "value money / don't value money" official.

[@proggz39]

We can only blame ourselves 😔. Not really though because there’s only one Taiwan factory making chips for the entire world

[@johngaltline9933]

For what it's worth looking at all the comments about fake frames here, The last two generations of AMD cards have been pretty competitive for pure raster performance with Nvidia's cards in the same price class and have been better than Nvidia's cards much of the time. No one actually cares because those fake frames everyone is complaining about here are better on the green card. Outside of the halo tier card, AMD has had competitive product for years for the features everyone says they care about, but no one is buying them.

[@mmmkmkmkmable]

It is healthy, my AMD card has been great and intel looks to be improving. It's the customers who have an unhealthy relationship with nvidia.

[@KellicTiger]

OK so the 5090 will be $2500 to $3000. Got it. No way will you EVER see a 5090 for retail price, ever.

[@AstralviewsUK]

Man Jensen was cringe af at that keynote. The richer he becomes the more eccentric it would seem.

[@selectionn]

unpopular opinion, I love the cringe. His jokes falling completely flat with a dead silent audience is fcking hilarious. its super obvious he isnt funny at all, and is just very wealthy and so expects everyone to laugh when he does something he thinks is humorous, because, well, he is rich, and you are a mere peasant.

[@didamnesia3575]

​@selectionn no, that's the popular opinion. That he's a slimeball and gets weirder every year

[@brettito]

@@selectionn Money makes people think they can do anything because they can get away with it. That's the definition of a corrupt individual.

[@KartKing4ever]

How is he still their public speaker? Lmao

[@EvilMcSheep]

I for one enjoyed the cringe :D Feels like listening to a real human, not some oversanitized corporate PR bot

[@CallsignVega]

So basically the 100% faster than 4090 is DLSS 4 marketing BS, and the chip is 30% larger with 30% more power use for a measly 30% rasterization performance increase for a 28 month wait.

[@WybremGaming]

basically no innovation in pc gpu hardware the past 3 years nearly. Just add more power to it.

[@blackwidow7804]

And cost 33% more 😅

[@selectionn]

on their website they literally compare the 5070 to the 4070. If that claim was true they would be comparing the 4090. its just a blatant lie to generate buzz. Which should be illegal. Sean Murray of No Mans Sky fame was more truthful about his game.

[@AGuy-vq9qp]

Yeah it's mainly being sold to AI fucks with the extra VRAM.

[@nickyarbrough8392]

Pushing out 12gb VRAM cards at $600+ in the year of our lord 2025 is ridiculous.

[@faceurhell]

It was $550

[@siddharth846]

​@@faceurhellwait till it comes in market

[@iTzSnowPro]

@@faceurhelltaxes

[@chronosschiron]

@@siddharth846 ya double that easy and member USD so in canada with are tanked dollar its literally a 50% add + usualy 100% so you can see it at 1200-1500 canadian watch this is why fuck these shits

[@Aotearas]

@@faceurhell MSRP is without taxes. Including sales tax and/or VAT where it applies, it will be well above 600. And that's not factoring in retailers taking a slice off the inevitable scalper market.

[@CorruptOutcast]

"You sit on a GPU made of AI generated lies"

[@AmigoAmigo-w5p]

Your whole life will be generated by A.I in 10 years from now. So relax or go find a cave.

[@47enterprises]

media.tenor.com/uwxQ3r-1XVEAAAAM/will-ferrell-elf.gif

[@deanconstan481]

This message is sponsored by AI.

[@truthseeker6532]

Power, cost etc... is now facing the public. Be thankful that such things like the AI frames is possible. For old tech of "straight forward" is not possible anymore = Limits been hit to the point AMD cannot go any further anymore due to that fact and why Nvidia can = They have the knowledge needed, the shareholder power and the patents etc....

[@CorruptOutcast]

@@truthseeker6532 Uh no, it's called devs need to learn to optimize their fuckin games again and stop relying on res scaling and frame gen for their games to run.

[@Alex-JGr]

The textures on Jensen Huang's jacket are clearer now.

[@cesarbravo822]

Can't wait to play Cyberpunk with 8x the input lag

[@poka26ev2]

Basically like the ads (game content) in Cyberpunk 2077, ai generated, now is ai generated by an ai generated program

[@miliyn17]

The input lag is essentially the same as using DLSS now.

[@etovsehromka]

Doesn't it seem strange to you that the 5080 only has 16 GB of memory, while the 5090 has increased it to 32 GB? as if this suggests a quick announcement of the 5080 super or ti super with increased memory

[@mitchjames9350]

It’s what I thought as well which means a 5080ti and 5080 ti super.

[@delijamomak1711]

For sure they are saving it for super or ti versions

[@N4CR]

No, it means they are gatekeeping you from running machine learning on it and they won't make a super with more vram. They are not enough ram for 4k already lol.

[@gemelwalters2942]

at this point the 5070 should have 16gb, so no...that wouldn't be very impressive at all

[@jaylapointe1654]

The 3gb gddr7 chips aren't available yet. They will probably use those for the supers or future ti over increasing the bus width.

[@DanielCrapo]

all that power with only 12gb vram

[@hggyuu]

let's say it's X perfomance So 1060 6gb = X (Good) 4060 8gb = 2X (12gb ) 4090 24 gb = 4X (Good) 5070 12 gb = 4X (24gb) That power perfomance with 2 times lower memory

[@Kirinketsu_]

its like people forgot about the R9 Fury X and its 4GB of HBM that out preformed the GTX 980Ti and its slower 6GB DDR5 in 4k gaming for years despite games calling for 6GB+ VRAM.

[@jamesm568]

The more RAM in the 4090 is what going to keep it competitive.

[@RCmaniac667]

​@@Kirinketsu_i remember, but game engines are more reliant on vram now than then

[@deadangel03]

@@hggyuuyeah i’m good at math but you are NOT cooking with the hypotheticals right now 😂

[@InfoRanker]

Nice to know I'll be able to heat my house with one of these and annoy the local power company.

[@meekmeads]

Huh, they'll probably send you a gift basket 😂

[@tooprood]

​@@meekmeads I have market dependent electricity pricing and sometimes the prices drop to negative during the nights, meaning one of these cards would be a GREAT investment for reducing my power bill /s

[@Maxi86inAction]

I hate NVIDIA having not a single realistic benchmark. Ist just apples and oranges

[@jkell411]

Since when does ANY company do that? Name one. Please. Not one has ever released a reliable or real-world accurate chart. This is nothing new.

[@mares3048]

just a marketing scam with those graphs 1.7x* better !!! but then huge asterisk is put * there :D

[@Maxi86inAction]

@ most of them have comparable software setups though. This is comparing DLSS 4 5080 vs DLSS 3 4080

[@avonire]

​@@jkell411 whether or not it's nothing new, doesn't change the fact it's kinda annoying lol

[@jkell411]

@Maxi86inAction , how is this different from any of these companies cherry-picking games that only show the most improvement and scaling the graphs in ways to make them look better? It's all the same. They're all trying to sell a product and boost their stock prices. I'm not defending Nvidia, I'm just pointing out that they are no different than any other of these companies. I have never seen ANY charts released by these companies that match up to real-world testing. They are all misleading. This presentation was no different.

[@umeng2002]

You get fake frames, and you get fake frames... EVERYONE GETS FAKE FRAMES!!!!!!

[@kevinerbs2778]

Fake limited supply too by nvidia.

[@nailil5722]

fake resolution AND fake frames, I love modern PC gaming

[@alexartiles9756]

And if it looks and feels good to use. Doesnt matter if its "fake"

[@mr.goatman4024]

YAY!

[@fever309]

Why is it fake though? Not a traditional rendering pipeline? Sure... But the GPU still creates those frames.

[@papasmorc7130]

Still cant believe that Nvidia is making cards worth $500+ with only 12 gb of vram in the year 2025.

[@Skobeloff...]

As long as people buy them, there is no reason to cut into the profit margins by giving them more ram.

[@owdoogames]

Worth $500+? They may cost that, but are they worth that?

[@bighit20100]

Yes because you need more than 12GB on a xx70 card.../s

[@bsx132]

The 5070 will run anything out at 1440p, the VRAM wont matter.

[@maynardburger]

@@bighit20100 While 12GB is definitely workable, this is clearly an x60 class card, just like the 4070 was. 12GB is basically minimum for modern gaming with good settings, and nobody should expect less when you're spending $500+ on a GPU. The problem is that demands will continue to go up. 12GB may be sufficient today, but what about in two years? It's never good to be spending a lot of money on a new graphics card knowing your VRAM quantity is near the floor of what's acceptable now, since that likely means problems in the future. It's the same shit many of us tried to warn people about who were buying 3060Ti/3070 8GB cards back in 2020. They would not age well.

[@NefeTri]

Maybe I can afford 4090 after this announcement.

[@temp50]

Since everyone else thinks alike, chances are that 4090 will be more expensive than it was its original MSRP.

[@trainerbluetv]

Interesting - a bot complaining about tech costs. Is this the equivalent of people complaining about the cost of gas/groceries?

[@buggerlugz6753]

you are funny!

[@aaronc3032]

He politely told people if they don't like it bad luck get used to it. He really meant that for the GPU I'm sure!

[@breach005]

Im getting used to none of my rigs having nVidia GPUs

[@tobykornreich2719]

@@breach005 Indeed.

[@sondoobie1987]

I am 51, been into gaming since the 70's, had so many different computers, consoles back in the day. Amstrad cpc 464 with Harry the Hedgehog haha, I always imagined how graphics and environments would look in 20 years, must admit, my build is low to medium, have way to many bills to purchase latest gear anymore. Those lucky enough to experience any of these cards. Have fun and enjoy.

[@memcmeepants2392]

I had an Amstrad cpc 464 when I was a kid too, I lived a few years in the UK in the 80s, I miss being able to put my games into a 2 deck cassette player and pressing play on the game tape and pressing record and play on the blank cassette and just copying games for friends lol. I dont think I ever saw harry the hedgehog but my favourites where the dizzy games.

[@Steve-en1ws]

I'm about a decade behind ya and have been gaming since '85. I had different PCs and consoles as well growing up. I didn't start getting into "elite gaming" until the 1080ti was a thing. I can see why people do it now. I currently have the 4090 and I love every second of it. There is definitely no need for me to move to the 5090, especially at these prices. I've no doubt once we see actual benchmarks that the 5090 will not have massive gains over the 4090. We saw that with the 3090 and 4090. For anyone coming from a 3000 series or older, this will be an amazing upgrade.

[@sondoobie1987]

@@memcmeepants2392 yeah I loved dizzy.. So simple, but very frustrating haha. That noise off the cassette loading.. 👌

[@yo_jalen]

I believe Jesus will bless you with a new setup

[@jackofall2305]

We created this super powerful ai to tell your benchmark software that its really making more frames than its actually making

[@RageBaiter1]

except it actually feels like youre playing the frames. Stop crying and accept it

[@kyle8952]

@@RageBaiter1 No it doesn't. These fake frame schemes work by buffering two real frames and then averaging them out to create the fakes. You're always facing at least three additional frames of latency. I have no idea why you would defend these techniques, they exist only to fake benchmark results to disguise how bad the performance of new cards is for the price.

[@AegisK]

"Not possible without AI" translates to "the same frame rate as a 4090 (running natively) using DLSS 4". Kind of a nothing burger unless DLSS 4 runs on everything.

[@natecartwright2319]

at this point we need full RT Ultra at 4k native 120hz i dont give a damn what game it is. STOP WITH THE DLSS

[@selectionn]

its a nothing burger anyway because DLSS and all AI upscaling is total garbage the amount of ghosting and artifacts and graphical downgrading is never worth it. Everytime I play a new game, and have some weird graphical issue, I always check if DLSS is on, and turning it off fixes it, every single game. It also doesnt matter because the game developers themselves have to update the DLSS version in their game, which, spoiler, none of them do. You have to update it manually which is problematic and will have to be done every time a game updates on steam.

[@JM_Traslo]

Best part is they decided to switch out the models they were using when showing Flux AI on their performance graphs so that it looks better, but in the small text beneath they admit it's literally just using a different version that is lower quality and faster.

[@VArlaud]

that's pure evil 😦

[@chillsoft]

Yes, i remarked this too. And FP4 gives way less details and the quality really is hit bad. How the F is this legal? They are literally lying lol.

[@DiamondGB]

5090 graph doesn’t even look like 2x the performance of a 4090 despite processing literally HALF the same weights, which should literally double both processing and bandwidth capacity…

[@xbolt90]

You gotta at least love Jensen for his great memes, lol

[@GamersNexus]

For all the things he's said over the years that we actually had a problem with, I will genuinely be sad for the loss of memes if he ever retires. Definitely has a unique presentation style.

[@RyanFennec]

@@GamersNexus man has some aura about him thats for sure

[@afos88]

Some people hate it because they find it cringe or embarrassing but I think it's actually funny People need to take stuff less serious and have some fun

[@squidikka]

@@afos88 It's cringe because he's not being ironic, which makes the memes better.

[@SL4PSH0CK]

Jensen keeps on givings us memes

[@royconestoga7326]

I’m so glad I no longer list “computer building” in my hobbies list.

[@mechanicalmonk2020]

Why?

[@WybremGaming]

@@mechanicalmonk2020 because it's embarrassing today.

[@triv4555]

For real... thank God I finally did my first build last year. Got my 7900 XTX and I'm never looking back (or forward, at this point- not for the next decade or so at least)

[@Ninjatomte]

I've never in my life been less enthusiastic about a hardware generation

[@maynardburger]

As Jensen says, get used to it. Things are never gonna get better. Nvidia played a game of chicken with customers using the 40 series and they won. This permanently ruined the market by normalizing an insane, exorbitant price hike from the previous generation(more than 100% price increase like for like, ignoring naming). In the end, consumers are to blame for this. We can expect corporations to be greedy and see what they can get away with. The whole point of a free market is that consumers are supposed to be principled and push back against such greed. If we fail to do that, then we've ok'd and normalized the corporate greed.

[@AndrewB23]

You couldn't afford it anyway bud

[@palisad5279]

I see gamers say stuff like this every time NVIDIA releases a new consumer-grade GPU series. I'm looking forward to doing some development with a 5090.

[@aeromotive2]

you say that every year.. try taking up knitting

[@lolzforlunch]

When 4000 came out I was pretty much waiting for this generation of GPUs. I'll just switch to AMD at this point

[@PyroBBQSauce]

Very excited for "6070 = 5090" next generation when FSR pushes out double the generated frames at ~1.5x the latency and twice the power draw with DLSS5

[@anothercritic]

Over 9000 generated frames for each frame

[@PashaSlavaUkraine]

😂😂😂

[@henrymago2710]

Nah skip for 7090

[@SpacecowboyCarl]

FSR is really cool until I get more latency than cloud gaming.

[@Joe-wk9ow]

Yet I think Nvidia hit a brick wall on power. Whereas apple’s new m4 processor do things at 40w. I think we need to look into metal vs open gl in terms of power per watt standpoint. I’m no fan of apples way of doing things but at least the are going more power friendly.

[@nathanl4417]

<a href="#" class="seekto" data-time="522">8:42</a> Your hair stylist needs a raise

[@Robbie-mw5uu]

men are so dumb his hair isnt styled at all in fact his curls arent brought out properly because he doesnt use curl cream and scrunch his hair thats why his hair starts straight then wavy unto curls. Steve needs to adopt curly hair method and fix his rizz.

[@Vintagesupermarketstore]

I’m getting old

[@powerpower-rg7bk]

It feels like the RTX 5090 is the only thing really moving things forward. The rest of the line up is like a price cut for the previous performance levels. The kinds of stuff like 4090 performance but at the 4080 super price makes it a 5080. Not the worst thing not that big either, especially at these high prices and power consumption. I have a feeling these will be refreshed relatively quickly with faster and higher capacity GDDR7 memory. 3 GB dies are a thing with GDDR7 so 48 GB, 24 GB and 18 GB capacities are potentially possible for future 5090, 5080 and 5070 respectively. GDDR7 is of course new with effective clocks starting at 28 GT/s with room to scale up to 40 GT/s. The RTX 5080 gets 32 GT/s from the start which I'd expect a refresh to all move to this speed or faster with higher memory. Doesn't see like as many functional units were disable this time around on the chips so it'll be a clock speed improvements for the refresh alongside memory capacities. The exception would be the RTX 5090 which did get a memory capacity increase and a price increase over the RTX 4090. There is room to refresh that beast of a chip but I doubt it. (The nVidia/Quadro variant will likely see an update as moving from 64 GB to 96 GB of on-board memory would see further benefits in the professional market. nVidia did something similar during Turing where there was an RTX 6000 initially and then an RTX 8000 later.) Two wild cards remain: AMD which was shockingly quiet about RNDA4 during their earlier keynote and potential tariffs on the way. I have a feeling that that is why nVidia is launching at the prices they are currently and will refresh later on after tariffs have been imposed to do a 'pricing correction'.

[@owliom]

Based

[@Nitrox22]

As soon as I heard the "5070 is a 4090" thing I knew the GN video about it was gonna be good. I look forward to that real world testing.

[@jjninja850]

Agreed, I don't get all the hate yet, let me see it running and I'll make my own call

[@HiznardoDiaz]

<a href="#" class="seekto" data-time="56">0:56</a> I saw the event live and I though Jensen was going to pass some gas.

[@xnitropunkx]

The most efficient machine is Steve pumping out content like a beast

[@PutYourQuarterUpGaming]

It would be impossible without AI

[@Mohynoy]

<a href="#" class="seekto" data-time="217">3:37</a> massive ?

[@Henryisdjfunkyho]

Loooooʻoo taper fade

[@nickbelitz2117]

you know what else is massive?

[@1vaultdweller]

1) 10% hardware increase 2) 10-20% power draw increase 3) 0% VRAM increase 4) 100% BS increase

[@breach005]

AI == Bullshit anywhere I see it at this point. These jokers can piss all the way off with this nonsense

[@mithrandir491]

Every time there a new release there is a promise of 8k gaming, I am not falling for that one again.

[@proggz39]

1440p is more than enough on a 27in monitor

[@chillnspace777]

​@@proggz39fr

[@bloodgodalucard4769]

@@proggz39 Yeah but not all of us like it that small. If we're out here gaming on 65" screens, we want the higher fidelity.

[@Skobeloff...]

@@bloodgodalucard4769 If you are out here gaming on 65" screens, you have other things you should be thinking about, like thinking.

[@bloodgodalucard4769]

@@Skobeloff... That makes no sense whatsoever. Is that supposed to be an insult? Try harder, kid.

[@hoangdungmai8781]

Your videos are always so helpful, thank you!

[@RFC3514]

"Insane power levels and blurry, shimmery, upscaled and interpolated rendering worked for us last year, since most reviewers swallowed the whole thing, so we'll be doing more of that."

[@brando3342]

Sadly.

[@scottmatthews-td7sn]

For real…admittedly I play at 1440p but there are numerous issues with up scaling and FG and it’s only going to be compounded with newer iterations. For example, FG does not increase the frame rate of RT reflections from what I can tell or at least does not do it to the extent of the rest of the frame. DLSS upscaling in general is as you said, just blurry in motion to my eye

[@ricequackers]

And that's just Jensen's jacket.

[@RFC3514]

@@ricequackers - His jacket isn't even possible without AI.

[@hughwattmate8467]

Are you going to cry about it?

[@Aveok]

I have the feeling a lot of people who waited for this will be disappointed when it finally launches.

[@apollo209]

You're right, i just hope the prices of current GPUs will lower. It's obvious the consumer side of the business is not a priority anymore for Nvidia.

[@GregObpar]

in what way? For someone who refuses to ever use DLSS maybe

[@gheorghilassilviu]

<a href="#" class="seekto" data-time="595">9:55</a> I really hate the wy Nvidia makes those benchmarks, with framegen and DLSS enabled. I want to see raw power, RTX 4090 vs RTX 5090 frame for frame, real performance. And one thing, back in the days of 1000 series they showed the lower tier card vs the top tier card of the last generation (e.g. GTX 1070 vs GTX 980).....honestly I'm skeptical of these cards, I want to see reviews, especially from you guys

[@SirDeeznuts]

so true

[@andrelaus9904]

I think it makes sense really, Most games now utilise DLSS and more and more use FG, they want to show you what performance the cards will get when ran at full power❤

[@definitelynotthecia4998]

Good to finally know what my opinion is on this.

[@SiXiam]

Make more money is all you need to know.

[@chillnspace777]

​@@SiXiamthe more you spend, the more you save

[@RCmaniac667]

No opinion yet

[@garydiamondguitarist]

Offer the most minimal improvements for the most maximal profits. This is a popular business model for tech companies in the 2020s, didn't you know? 😂

[@BehindthesafetyofLinux]

If frame generation gets any larger, you will soon be playing a rough approximation of a frame rate and every developer will crank up their visuals to require it. We will be nostalgic about the days when a frame was a framed rendered. Very similarly how were were nostalgic for high refresh rate when we first swapped out our CRT's for LCD's. Lets hope it doesn't get to a point where it looks like high latency lag in a multiplayer game, but locally on your own machine.

[@maynardburger]

Why do gamers keep saying this nonsense about developers somehow being in league with Nvidia to sell more expensive GPU's? Why on earth would developers want that? It's not just not their business, it would literally hurt them by reducing the market of people who can buy and play their game. It's such a stupid claim.

[@BehindthesafetyofLinux]

@@maynardburger It's not about being in "league" with any company, it is making use of the perceived technologies available to the point where they become requirements. I think they do it because it is new tech and it allows them to push the boundaries a bit more. The problem comes in when you start seeing fame generation being used ubiquitously across the recommended specs. Which is happening in newer titles I might add. It is a general lack of foresight on the part of the developers for perceived visual fidelity headroom. This is exactly what happened with early LCD tech. Everyone raced to LCD because of the sharper image quality and flat screens, not realizing that they were giving up high refresh rates for at least a decade plus into the future. God help us if frame generation does that to frame rates in general.

[@maxdamage4919]

​​@@maynardburger90 % of Game Developers are paid from nvidia, You forgot the famous Nvidia logo in all games ???

[@Voyajer.]

​@@maynardburgerguy forgot about Nvidia Gameworks™

[@haukikannel]

It actually is soon better to just show generated frames, because the real ones in between would look different than the generated ones! You get better imago quality by leaving real frames out totally! 😂😂😂

[@lordbloby]

My real concern is the percieved latency that comes with these Frame gen techs. It typically makes the inputs feel slow and floaty as you only have 20-30 FPS worth for input instead of the 140 your "supposedly" getting out. Would you guys be able to do any testing around this at some point?

[@Alex50969]

Yea latency testing for MFG would be something i like to see as well

[@YouOnlyLiveOnce...]

@@Alex50969 As per their demo shown at CES the latency should be the same as regular FG. The real question is the image quality.

[@AlannahFranki]

Awesome video, looking forward to the next one!