RTX 5080... More Like RTX 5070? - Rumored Specs vs 10 Years of Nvidia GPUs

[@kmsmith8934]

No mention of the 4080 that was revoked to be the 4070. They tried before, they will again.

[@mr.potato9449]

It was downgraded to a 4070ti not a 4070 and it shouldn't have even been that as it was really a 4060ti, so at the start they were trying to sell a 4060ti as a 4080 with a price increase to boot :|

[@ziokalco]

​@@mr.potato9449 Yes, the 4080 12gb (specifically the 12gb variant that turned into the 4070ti) has desing characteristics (both absolute and relative to the line up) corresponding to what previously would end up being called a 60ti. TBH I think they did it intentionally absurd to distract from the fact that the final line up was entirely displaced (50 being called 60, 60 called 70, 70 called 80, 80 morphed into somewhat always being called 90)

[@BenoHourglass]

@@ziokalco They probably thought it was fine to do this because people would just buy them en masse to mine crypto. Unfortunately for them, Ethereum went to proof of stake the month prior to the release to the 4090. Suddenly the primary segment for GPUs just evaporated.

[@hyperturbotechnomike]

nVidia, what a clown company nowadays. The RTX 4070 pardon "4080 12GB" i got last year may be my last gaming graphics card from them. I say gaming card, because i still rely on their Quadro cards for my business, which are a complete ripoff. nVidia has a total monopoly in the professional industry because of CUDA. A monopoly which affects everyone, even non PC users (because engineering relies on CUDA), but no one is talking about.

[@Dexion845]

Remember they made the mistake of using AD104 instead of 103, this time they're using GB203. Don't expect an unlaunch to happen this time. But it should be priced at the same point they wanted to launch the 4080 12G at.

[@registeredblindgamer4350]

So, basically you're worried RTX 50 is going to be a repeat of 40 series? Remember Nvidia tried to sell us a 4080 12gb for $1000. It kind of looks like they're trying this again, but this time making one 80 class card which is really a 70 class. I kind of expected this. Nvidia is good at slowly boiling consumers. Weird how people already forgotten the 4080 12gb. Nvidia is just repeating itself.

[@Desalater2]

plenty of room to stick a 5080 super at 60% perf for 1200 and a 5080 super ti extreme platinum at 80% perf for 1600

[@Chicken-o5e]

​@@Desalater2extreme platinum😂 might as well add pro max plus ++

[@registeredblindgamer4350]

@@Desalater2 The 4080 Super was only 5% faster than a 4080. This won't happen and if it does it isn't good. We want gtood cards at launch. Not them walking bad poor choices again for them to try it again for the next gen.

[@kennyoffhenny]

Well they obviously can’t compete with consoles price/performance, so they have to make their money with the __80’s & __90’s so they can even try to compete with consoles with their __60’s & __70’s.

[@Uthleber]

@@Desalater2 in fact there is no room for a 5080 S/ti. They HAVE to use heaviely cut down GB202. If they do, just in a small amount.

[@seamuspink9098]

they will never release a 1080ti type of card ever again

[@MemosKaravis]

Really and this is very sad

[@urluelhurl]

Nvidia will never release the 1080ti type again... guilty feet has no rythme.

[@leetskeet4476]

Why would they? Nvidia barely makes any money on gaming GPU's compared to everything else they do. If AMD had some balls and made a decent high end GPU to compete with Nvidia's 90 series (even if they lost money initially) they'd swarm the market. Instead they are relegated to making okay-ish mid-range GPUs for snobs who brag about saving $300 every 6 years

[@gurjindersingh3843]

3080 would have been that if GPU pricing was normal.

[@MIchaelSybi]

@@urluelhurl And I never gonna play again, the way I played with 1080ti oh yeah

[@proklet4694]

They are doing it again...

[@blakeforrest9600]

And they’ll get away with it… again..

[@placeholder3853]

And people keep buying... Not me

[@gyo900]

Ronald Reagan to NVIDIA: “There you go again!”

[@geiers6013]

I mean what did we expect from Nvidia? I am considering getting a 4070 super on sale, because I think soon prices will again get much worse, but I struggle with accepting only 12gb Vram with such high performance...

[@mikelay5360]

@@proklet4694 you know AMD exists right 😅. Just kidding.

[@DragonBane299]

Are we forgetting 4080 12gb…?

[@christophermullins7163]

I'm not

[@mikelay5360]

@@DragonBane299 this time it's 16GB. And we'll buy it.

[@manhdepzai2k3]

@@mikelay5360 i'm still good with my intergrated card 👍

[@G4RYWithaFour]

ah yes, how the 4070 was the 4080, so they could make the 4060 the 4070, and the 4050 the 4060.

[@MuShinnen]

​@@mikelay5360 I won't.

[@AceStrife]

NVIDIA: "I got away with it, even with you meddling kids!"

[@luzhang2982]

yeah and they’ll continue to, because frankly actual customers don’t care and still purchase.

[@twiggsherman3641]

Jensen 100% learned last go around. He wants to keep moving the product stack downwards while the price goes upwards. He would've gotten away with the 4080 12gb fiasco if he had of just launched it as the only 4080. He took that lesson to heart, and here we are.

[@rtntm4812]

Don't you feel sympathy for poor Jensen? He's only got 100+ billion so far.... /s

[@Gofr5]

His leather jacket collection needs to be fed one way or another.

[@alvar891]

I bet this time they launch only the one 5080 (actually the 5070), but will later launch the 5080 ti (actual 5080). This will work.

[@flimermithrandir]

100% this. Agreed.

[@ishaansaral502]

This actually makes a lot of sense. We are only getting one 5080 this time, so there is nothing for people to get outraged about. Man Nvidia sucks.

[@Uthleber]

Dont forget the 4060 is based on AD107. it is a real 4050.

[@Gadtkaz]

While also costing more than 60 class and performing worse. Time for round 2.

[@mhrussell4264]

So is 4070 a 4060?

[@ianwolter6142]

but it can lay modern games with path tracing at 45 to 60fps at 1080p, it's a real beast tbh

[@Uthleber]

@@mhrussell4264 The regular 4070 is MORE cut down than a 1060 6Gb was.

[@misterijaaaa]

​@@mhrussell4264It is.

[@hinz1]

Basically they're saying, either buy a 5090 or you're not worth it.

[@korosoid]

Also Tim didn't count RTX values as Adored TV did.

[@avatarion]

Doesn't bode well for 4K gaming. You either pay up for the 5090 or you are not gaming at 4K without lowered settings. How are we supposed to move up in resolution if it's only for rich people?

[@SemperValor]

It really doesn't matter to Nvidia anymore, they could sell literally 0 gpu's and STILL hold 85% of the GPU market. They're that huge. So a few people complaining about VRAM wont touch them. You will buy the 5080 at 16Gb and love it because you have no one else to buy from and AMD is out of the picture now. With Nvidia having no competition they're free to do whatever they want. You can't go anywhere else to get a gpu as good as theirs anymore. Back with the 12Gb 4080 Nvidia had to compromise because of AMD competition. Not this time though.

[@J-Vill]

​@@avatarionWhat are you on about?? I've been 4k gaming since I got my 3080??

[@avatarion]

@@J-Vill You are not playing any of the big games that support ray tracing.

[@wavetrex]

Thanks for raising even more awareness of the extreme ripoffs that nvidia is pulling lately.

[@BlackJesus8463]

AMD tryed to pull the same bs!

[@kl0ness]

Nobody cares. They're eating good.

[@XashA12Musk]

@@BlackJesus8463 but we ain't buying amd

[@danishprince2760]

I fucking miss the 900/1000 series days. 80-model for $5-600 and 70-model for $3-400 is the perfect price point for that tier of cards and the fact that they're now over double that is insane

[@stanley3647]

I paid for my GTX 1080 - nearly 800 GBP in store (this is more than 1000USD after conversion) it was before 1080ti. So, 1080 was a more expensive than You remember. Prices of 1080, was reduced down after 1080ti release (to 500 GBP, or 700 USD after conversion). I was furious. Because my card lost a lot of value just after few weeks... (In UK prices includes 20%VAT) This was last time i paid premium price for GPU - This year my GTX 1080 was upgraded to 2nd hand RTX 3070 (280GBP), and I don't regret anything ;)

[@JasonKing-m6m]

@@stanley3647 The real question is that you were foolish once, will you be foolish again and jump high as Nvidia commands you to?....

[@philipje1]

@@stanley3647 No it was not.

[@nonaurbizniz7440]

A lot has changed supply chain wise since then though. People forget that the high prices caused by the supply chain crisis during the pandemic have never fully resolved themselves and its been playing catch up ever since which has kept prices high. It's predicted it will take another few years just to get them sorted again to pre pandemic levels barring any other global disruptions like war and such. Get used to the high prices they ain't going away.

[@awebuser5914]

OK, let's do the math, shall we? The 1070 released at $379 in June 2016 and that's $497 in adjusted dollars. A 4070 was introduced at $599, so $100 more for *far* more advanced technology. This blinders-whining about prices from almost a _decade_ ago is comical. iPhones have also skyrocketed in price (along with most other leading-edge tech) over the past 8+ years, so are you going to whine about that too?

[@oopsydaisy4637]

only 16GB for 5080...a disaster if this is the case

[@Koozwad]

now now, when you turn off the RT/PT gimmicks, the VRAM usage will be more than fine in 16GB, even at 8k most likely, which you can then run without using them(though if you ask me 8k is a waste of frames as the visual difference is so tiny in pretty much any screen)

[@Eepy-Rose]

@@Koozwad Generally an 80 card shouldn't have to be an "eh just turn off RT" kind of card. It is high end you should be able to make full use of its capabilities without being restricted by VRAM.

[@fleurdewin7958]

@@Koozwad People pay alot of money for 5080 not to turn off RT. They want to enjoy all the bells and whistles in the game. Maybe not play in 8K. But 4K 120hz seems acceptable.

[@Koozwad]

@@Eepy-Rose I don't believe in RT/PT. Yes it's 'shiny', but comes with various issues and heavy reliance on fake frames and/or fake resolutions. It's not the right direction, but is one NVIDIA is trying to push for(perhaps to try and justify the high prices of their cards and force people to buy more expensive cards than they otherwise would). I would highly prefer high quality hand-crafted rasterisation, like in for example RDR2. RT/PT is a cheap/lazy way to add various things to a game while rasterised graphics usually get overlooked in comparison(including to make the RT/PT look better). RT/PT is used as an excuse to save dev time, which also results in a worse product overall. With RT/PT games are crafted with less 'love/passion' than ever before - all they see is dollar symbols. My original point was that 16 GB VRAM is more than enough for years to come when you play rasterised games(which ideally you would be). I would of course welcome breakthroughs in rasterised rendering, as everything today is becoming a blurrier and blurrier mess.

[@Fiztex553]

not a disaster, we'll just get a used 4090 from those upgrading to 5090, and let them keep this "new" 5080.

[@tomtomkowski7653]

Yup... this started with the ADA series where they gimped all cards below the 4090. In terms of the relative number of cores, they are selling lower-tier cards in higher-tier boxes with a higher-tier price. 3090 - Die Size 628 mm², Transistors 28,300 million, Cores 10496, 98% of full die cores 4090 - Die Size 609 mm², Transistors 76,300 million, Cores 16384, 89% of full die cores 3090 -> 4090 - 3% smaller die, 2.7x more transistors, 56% more cores 3070 - Die Size 392 mm², Transistors 17,400 million, Cores 5888, 55% of full die cores 4080 - Die Size 379 mm², Transistors 45,900 million, Cores 9728, 53% of full die cores 3070 -> 4080 - 3% smaller die, 2.6x more transistors, 65% more cores 3080 - Die Size 628 mm², Transistors 28,300 million, Cores 8704, 81% of full die cores 4080 - Die Size 379 mm², Transistors 45,900 million, Cores 9728, 53% of full die cores 3080 -> 4080 - 40% smaller die, 1.6x more transistors, 12% more cores --------- 3060 - 276mm² die, 33% of cores - $330 4070 - 295mm² die, 32% of cores - $600 3070 - 392mm² die, 55% of cores - $500 4080 - 379mm² die, 53% of cores - $1200 3090 - 628mm², 98% of cores - $1500 4090 - 608mm², 89% of cores - $1600 ----------- 970 - 54% of full die cores (980 67%) 1070 - 50% of full die cores (1080 67%) 2070 - 50% of full die cores (2080 63%) 3070 - 55% of full die cores (3080 81%) 4080 - 53% of full die cores

[@farazsiddiqui1312]

Can u explain why die size matters?

[@zbigniew2628]

Oh, if rtx4070 was rtx4060 for 400$ or even 450$ it would sell like crazy. Especially after some prices drop. All I need is some dual fan GPU with performance of at least rtx3070 and 12gb of vram with DLSS and physx for older games, without need for RT... Used rtx4070 cost 2x more than rtx3070 in my country and rtx4060ti 16gb is a joke. I would buy the rtx3080 even with 10gb, but it is bigger and power hungry. Even if rtx4060ti had 10gb and PCIE x16 and performance on par with 3070, I would buy it for today's prices... Ofc used one 😅

[@geepsu666]

@@farazsiddiqui1312 die size is directly related to production cost at the semiconductor fab

[@tomtomkowski7653]

@@farazsiddiqui1312 The bigger the die the more transistors you can put and the bigger the transistor the more expensive silicon. So they are selling cheap hardware with higher-tier boxes and prices.

[@flimermithrandir]

@@tomtomkowski7653 + if you have a smaller Die Size, you need less Components around the PCB too, to make the Card. (In Generell, of Course there can be Things done with new Designs and more Power and what not that also need new and or more Components but in Generell... i would say smaller Die Size = less Costs all around too) (Edit. What i mean is, that a Die Size of the 5080 that is 350mm² will need less Components compared to an 5080 with 400mm². I think that makes my Point more clear. Sorry)

[@justinpowell2400]

The bad assumption IMO is that the 5090 will only be $1600. Im expecting closer to $2000.

[@FreightTrain22]

Not a bad assumption when it was assumed the 4090 was going to be 2000+ and ended up at 1600. People overestimate flagship prices every generation

[@peterscott2662]

I'm thinking you are an optimist. It's going to be $2500-$3000.😉

[@_Ekaros]

MSRP 5090

[@johnnypopstar]

@@FreightTrain22 Or, the 4090 _Founders_ was $1600, but if you wanted a "real" 4090 with proper cooling and so on, those _were_ in the 2k range. Hell, even in £, where we should see slightly lower numbers than the equiv amount expressed in $ (and ignoring the water-cooled ridiculous models), launch window 4090s were between £1,760 and £2,400, averaging £2,080. I know all this because I was recording prices every couple weeks in a spreadsheet, which I still have.

[@FreightTrain22]

@@johnnypopstar no there were plenty of cards around msrp and even lower for a long time until Nvidia started cutting back production

[@Reza1984_]

Crypto bros single-handedly ruined graphics card pricing for good.

[@Koozwad]

don't forget AI

[@deus_nsf]

@@Koozwad at least AI has some uses, crypto was just sad.

[@Koozwad]

@@deus_nsf True, though I'm not a fan of AI at all generally. It makes things soulless and even more about maximising profit instead of making great products.

[@hflx]

This is not true, it was a combination of 3 factors at least. 1 could be that for sure, 2 was the lack of competition from AMD, which almost reached 30% of market by 2008, now it's almost nothing; 3 is the AI boom and the fact that they earn more by selling chips to big companies than end-users, my opinion this will only end in about 3-4 years when all the Fabs planned around the world will be ready and the AI boom will be an AI burst... Look at CPUs market, it's pretty much stable as we have a health competition, so it was no lack of sand as well... it was purely market opportunity to increase profits.

[@a5cent]

No. The crypto boom and scalpers are almost irrelevant. Right now it's two things: a) Lack of competition b) The fact that nVidia would much rather sell AI cards than gaming cards. nVidia doesn't care about selling a huge amount of cards. They care about their ROI. nVidia reserves production capacity at TSMC. That is fixed, and nVidia pays for that whether they use the capacity or not. Which chips are produced on those production lines nVidia can decide flexibly however. nVidia wants the percentage of consumer/gaming GPUs to be as low as possible. That is what inflates their bottom line the most, compared to their fixed cost structure. Raising the prices on consumer cards is how nVidia ensures a good chunk of that capacity is available for AI GPUs, while maintaining their flexibility to sell more consumer GPUs if the AI craze starts to subside. That's it. This would be exactly the same even if crypto was never invented.

[@azazeltheforgotten]

Well, it's not much of a surprise, remember Ngreedia named the RTX 4050 as RTX 4060 to sell it at a higher price.

[@rustler08]

Correction: pretty much every card under the 4090 is mis-named and is shifted up one tier. Maybe we can argue that the 4080 is still an xx80 series, but the rest certainly do not belong in their naming and pricing tiers.

[@maynardburger]

The entire 40 series range did this.

[@themobster7284]

Worst part is that even the god damn tech press that is supposed to, not be biased and to be objective, said RTX 4090 is good value, WHAT a F joke... Nothing after GTX 1080 have good value... Hell even before that was even better high end cards today should cost 700$ max, period...

[@joee7452]

It's because people like to scream at the wrong things, like this video. If the 5080 has those specs and is released at say 999 and offers 50% more performance then the 4090, it's a good card. If it comes out with half again as much ram and cores and is 1400 dollars and offers a 10% performance jump on the 4080, it's a bad card. Like he said at the end, it doesn't matter what the specs are or what the name is, it's the value for the performance. naming a terrible value card as a 70 class or 80 class doesn't change the value of the card. The 4060 is a horrible card to me based on it's value. I don't care if had been named a 4050, it doesn't change that it's a bad value because it wouldn't change the pricing to them. They decided that to charge X for this X level of performance. The name means nothing. It about whether that level of performance is worth the price. Now do I think that the 5080 if they have those specs will be a good value card? Probably not because Nvidia will probably over price it. But until we see the performance information on them, we don't really know anything.

[@nathangamble125]

@@joee7452 I disagree. Value for money is the most important thing in determining whether a product is good, but that product having a misleading name is still a problem in its own right, just not as important.

[@bromomento3862]

4000 series looks disgusting on those percentage charts lmao

[@Anand70707]

And it did not stop idiots from buying the 40 series cards. The real takeaway from this is that Nvidia's customer base is just braindead, and would like nothing more than to give away their hard-earned money just to have the latest and greatest.

[@Seeeaaan]

@@Anand70707Who hurt you 😂

[@F0zek]

@@Anand70707 Lol, mad aren't we?

[@G0A7]

​@@Anand70707That and the fact that AMD cant compete with that Nvidia, thats the really sad part

[@Anand70707]

@@pauloa.7609 Yeah, that's me, the bot who speaks facts. Whatever helps you sleep at night.

[@Jacketz123]

They’re 100% leaving a large gap so they can slot a 5080 Super, Ti, and Ti Super in there later. Classic Nvidia..

[@HelloImCrimson]

5080*

[@pwnomega4562]

It'l probably cost as much as the 4090 as well 😂

[@Jacketz123]

@@pwnomega4562 £4090 will be the price bro haha

[@wilhelm391]

Nah they just want to rip you off and you're okay with it

[@Jacketz123]

@@wilhelm391 in what world am I okay with it?

[@mariodrv]

They're pissing in your faces again and calling it champagne.

[@AKmumu]

People willing to buy new gen cards everytime Nvidia releases one are pretty much pissing in their own face. It doesnt take a genie to know that older GPU are still capable of running every games available perfectly at ultra settings

[@LilaHikes]

And the sheep will still buy it. They complain about prices, complain about a _monopoly_ but then still buy NVIDIA. The only reason to buy an 80 or 90 series card is for the extra features like RTX and DLSS. People *still* choosing NVIDIA at the mid and lower tiers just blows my mind.

[@XellithUS]

@@LilaHikes Im in the market for a new card. (R9 380 atm) In an ideal world I'd have one that does Ultra, 60FPS, 1440p and raytracing. Im hoping that the new 5000 series cards release lwill ower the barrier to what Im aiming for. If Nvidia takes the piss I will buy something used for cheaper, but it still might be Nvidia. (I can always sacrifice raytracing.) I figure I'd be able to get a significantly cheaper AMD card that easily meets my desires sans raytracing, but I had SOOO many problems setting up my wifes 7800XT, and AMD did nothing to help with the issue. I was able to fix it on my own, but I dread changing anything on her system in case the problem pops back up again. Nvidia seems.. safer to me I guess? Though I figure they would fuck me the first chance they get too.

[@Butte_r]

@@AKmumuWith the exception of poorly optimized modern games.

[@blabla-fw9ix]

And the people will still drink it so why should they care ?

[@grimdicer152]

10% faster but 50% more expensive

[@mikelay5360]

AMD : Exists. Thanks Edit: sarcasm also exists 🙃. I will show myself out.

[@greenumbrellacorp5744]

@@mikelay5360 AMD: Fails to compete and price matches nvidia then undercuts by 5%. Thanks

[@Pedro-Santos.88]

@@mikelay5360and play always in low setting because dont got dlss 3.5 , and get low quality in light❤and refletions, sad 😂😂

[@leviathan5207]

@@mikelay5360 AMD is utterly pointless in the high end sector. Like even 7900xtx is a bad deal. Losing out on DLSS and having RT performance that is not much better than a 3080 is supposed to keep people from buying the 4080? High end AMD is DEAD and has been for nearly 10 years now...

[@TerrasClip]

The more you buy, the more you save

[@beerskee]

Not really liking this trend. 80 class price with 70 class performance.

[@XTendingYT]

„Yestrdays Tech for Tomorrows prices“

[@mikelay5360]

With near zero competition

[@Creepus_Explodus]

@@mikelay5360 It's not like competition would really make a difference here. Nvidia are making all their money from the AI bubble, these consumer cards are just for brand awareness among the general public. They were neck and neck with AMD with RTX 3000/RX 6000, and prices sure as hell didn't go down for the next generation - they all went up instead. Both AMD and Nvidia saw that they don't need to compete for your money, you'll give them whatever they ask just because of brand loyalty.

[@mikelay5360]

@@Creepus_Explodus RTX 3000 was using a garbage node from Samsung. They will never repeat that mistake again.

[@UwU-ik5ur]

Nvidia is Apple now

[@ianwolter6142]

yeah that's what amd has been doing, their architecture hasn't seen a change since 2018, at least with nvidia we had turing, ampere, ada lovelace, and now blackwell

[@AKK5I]

The 👏 More 👏 You 👏 Buy 👏 The 👏 More 👏 You 👏 Save👏

[@1992Latvian]

Four Elephants, One GPU

[@brandonleemoss]

4 girls 1 GPU 😎

[@MisterRandomEncounter]

Got it all at shop and save ! ( the people who know know )

[@Rogerkonijntje]

Geforce now: You will own nothing, and you'll be happy ;)

[@BarrettKillz]

The more you pay, the more gigarays you get

[@Philballer17]

The Geforce 8800 Ultra often gets forgotten about, for just how beast it was for its generation. It was easily the 1080ti of its generation. You could skip the entire GTX 200 series if you had one.

[@imnotusingmyrealname4566]

It was before my time but I remember it was pretty revered.

[@jootai]

I salute you, sir.

[@merlin8208]

Nvidia trying to upsell us again? A tale as old as time.

[@highlanderknight]

Maybe, but you have a choice of what to buy (or not).

[@Arccanos]

I love when fanboys of any company have to pretend like the competition is evil. Gives me hope in the quality of comedy.

[@termitreter6545]

@@highlanderknight I wonder when the prices/performance is bad enough that people buying Nvidia habitually start being more critical about their choice? Those people have enabled their absurd pricing schemes till now.

[@AKmumu]

They are intelligent, they know people are stupid enough to buy it anyway. Nobody is forcing anyone to buy those cards but people still do and act like manchilds on the internet

[@Arccanos]

@AKmumu I love this take. AMD fanboys thinking consumers are stupid because they're not buying AMD. LOL

[@ShutterManAce]

I upgraded from a 1080Ti to a 3080 and was excited about the 5080. However, if the rumors are true, I'm holding off for now. It seems like Nvidia has left a significant performance gap for future models to fill.

[@HybOj]

lol u can only stack so much cards between 5080 and 5090 .... like... one?

[@rodrigoferreiramaciel4815]

​@HybOj I mean... 5080 ti 5080 super 5080 ti super (But let's be honest it's not like they would even care to make these slot in the line up in such a way that we could actually have an honestly good gpu for each price range)

[@HybOj]

@@rodrigoferreiramaciel4815 yea, but I dont believe this neither. 4080 and 4080 super are same cards for example. And there was no 4080 ti super. I think this is deliberate. It will allow em to make more money, its just another shady move from nvidia. If you are lucky, you get one card in between 5080/5090 and it will be like 10% faster than 5080. 5080 will cost a arm and leg, and jacket man will tell you its 10% faster than 1600USD 4090 thus a 1299 USD for it is a great deal. LOL :X bad times ahead. Hope my 3080 will hold on.

[@rustler08]

@@HybOj They say, apparently ignoring that our current generation of GPUs includes the 4070, 4070 Super, 4070 Ti, and 4070 Ti Super.

[@Geno1isme]

@@HybOj If you ignore existing Ti and Super prefixes, and the fact that NVidia could easily add more naming variations. There is already precedent for model numbers not ending with 0 or models with no number at all (never understood why they didn't do that in favor of "Super" variants). And lets not forget that NVidia likes reusing existing model names for different configurations.

[@StaelTek]

I remember back in late, 2010 you got Nvidias top tier chip, fully unlocked, for $500, GTX 580. Good times.

[@mitsuhh]

and it only had 2gb vram

[@FenrirAlter]

​@@mitsuhh1,5Gb* which was normal back then even the HD 6950 had only 2gb of VRAM

[@Darkswordz]

It's sad when the $1,600.00 flagship is the only value card in the lineup.

[@Ernismeister]

upselling 101

[@melexxa]

if you talk about value then the 4060 is the best price to performance ratio card in 40 series lmao, for source check gamer nexus review of 4080 super and price to perfm chart

[@themarketgardener]

@@melexxa It still has shit price compared to performance relative to it's competitor cards from AMD. On low end NVIDIA pricing is really bad

[@AceStrife]

@@melexxa why get a 4060 when you could get a used 1080 Ti, or 3080, or 3080 Ti, for less money and higher performance (sans 1080)? The entire generation outside of the 4090 was not worth buying; exactly as NVIDIA intended.

[@Koozwad]

@@AceStrife 4090 is a stupid card -way too much power for the cable to handle in the long-term(and short-term in some cases) -way too much power usage in general which not only increases power bill and is harder on the components/electronics -large size, which is impractical in many cases and can be restrictive such as blocking slots or the airflow itself to other components especially -bad value compared to for example 4080 Super 4090 will just get replaced again in no time, it's quite pointless to waste money on the fastest cards while sacrificing the above

[@QuentinStephens]

Ah yes, the 4060 - a -50 class card at -70 class prices. Sad to see they seem to be continuing that with the 50 series. When you were listing prices I would really liked to have seen an additional line showing inflation-adjusted prices.

[@arenzricodexd4409]

it is the logical progress. consumer need to accept the reality that it is getting expensive to build this stuff. 5nm? $17k in 2022. next year TSMC already said they will raise the price to $20k.

[@MuShinnen]

​@@arenzricodexd4409This is a disingenuous argument to say the least. Nvidia is boasting insane profit margins on their way to becoming a trillion dollar company. The whole "it's getting expensive" argument doesn't work when the company you're trying to shield is basically a monopoly.

[@QuentinStephens]

@@arenzricodexd4409 Don't give them ideas!

[@kazefw3834]

​@@arenzricodexd4409do you have any idea how much the margin is for Nvidia?

[@runayswarts970]

​@@arenzricodexd4409 but Nvidia profits are increasing so who gives a damn about them having to pay more to build

[@tim.fletcher]

This is a really solid video. Well done collating those data and statistics.

[@Desalater2]

5070 with 5080 prices

[@tonydavis8696]

Maybe the 5070 can get a 128 bit bus too!

[@TheIndulgers]

256bit bus 🤦🏻‍♂ Anything more than $600 would be an insult. So nvidia will probably charge $1000. Edit: Why I say this, the 1070 has a 256bit bus and launched at $380 msrp in 2016. With inflation it is still under $500. Nvidia is giving you less for more YEARS later. Same with vram.

[@DeadPhoenix86DP]

The 4080 is 1300 where I live, and the 4090 close to 2000. So expect the same prices if not more. AMD Had stopped competing with Nvidia, when it comes to high end class GPU's. So Nvidia can charge whatever they want.

[@attilalendvai7167]

if they sell it at 800$ and its basically on par with the 4090 im like ok whatever.. but anything above that is just stupid again

[@arenzricodexd4409]

1070 is based on TSMC 16nm. back then wafer price is $4k per wafer. in 2022 5nm starts at $17k. if GPU price going up in more linear fashion to wafer price increase then cards like 4060Ti will start at $1000 instead of $400.

[@semyonzhigunov671]

gtx 680 had 256-bit bus, so what?

[@mojojojo6292]

That's still plenty of bandwidth for how fast gddr7 is. The card should really have at least 20gb of vram though

[@justins1917]

The gap between the 5080 and 5090 is INSANE

[@deus_nsf]

literally twice lol, at least in electronics.

[@Pabula]

Nvidia simply wants to move their customer up the ladder, having more 5080 users say its not worth it compare the 5090, expecting the jump.

[@THU31]

But did people miss the fact that the rumored 5080 is 400 W and supposed to be 10% faster than the 4090? In no universe it could be a 5070. And personally I don't want a 400 W card anyway, no matter the price. I already had a 320 W 3080, which was pretty much unusable during the summer, even undervolted.

[@imnotusingmyrealname4566]

It's so all the wallet warriors can look down upon the peasants who couldn't afford the one true graphics card. It's building prestige that will sell the card far better than value ever will.

[@imnotusingmyrealname4566]

​@@THU31Had? So you upgraded to a 4090? Also you are wrong, you are living in the world NVIDIA wants you to. The 70 card used to give you previous 80 Ti performance. The 1070 had 980 To performance and was cheaper.

[@Game_LevelUp_]

No price to performance ratio is going to save how the Black Well cake was cut, people should vote with their wallets and sound their concerns just like they did with the RTX 4080 12GB. This shit has to end, since it will effect all of us down the line when Nvidia starts giving the 80 tier 30% of the total shader count of the flagship for 1199$.

[@ishaansaral502]

I don't understand why Nvidia would repeat the same mistake as the 40 series. Almost no one bought the 4080 for its 1200 bucks price, and people were also very unhappy with the 4080 12GB. Sales for 40 series other than the 4090 only became much better with the super refresh. I think people can connect the dots when they see how cut down the 5080 will be relative to the 5090, which will force Nvidia to make it the 5070 Ti again. They just love to fuck around to see if the consumer will notice, and once they are caught they backpedal with the real plan.

[@haukionkannel]

When NVIDIA put 5090 to $2500… more people are willing to pay $1500 for 5080! = profit! 😂

[@albertcamus6611]

someone I know manages at a retailer. He said the amount of kids using loan services is crazy. not sure they have that money at all, but yea, some were buying the 4080 out of desperation it seems.

[@xxNATHANUKxx]

Nvidia don’t care about gaming. They earn a tonne from ai so there isn’t the same incentive to create value for gamers

[@lolol50735]

the 30 series people who skipped 40 series for a better deal will get fucked

[@inflatable2]

They can because Geforce is now only a very small insignificant part of their market.. Their focus is on selling AI chips with way more margins..

[@gmt1]

Any of the performance gain from next gen cards is negated by Nvidia/AMD rebranding mid-tier cards as high-tier. The 7800 XT should be a 7900 XT. A 7700 XT should be a 7800 XT. Etc.

[@mirceastan94]

Great analysis - the sad reality is, that if 5080 will only cost 999$ and offer 4090 performance, people will be extremely happy, although it's literally a 70 class card sold for a thousand dollars. Yet, it might cost 1200$.

[@goalieman206]

If you go with the leaks the card isn’t even a 4090. It won’t even come close to the 4090 and still be $999

[@BonusCrook]

Used Ada GPUs 👍

[@anastasy2]

​@@goalieman206Do you guys understand why Nvidia stopped 4090 production that early? Because their 5080 will underperform while still being overpriced and they don't want any competition from previous gen. 😏

[@MrKardovalk]

@@anastasy2that smart af

[@pt-yt8322]

@@anastasy2i can’t believe more people don’t realize this, everyone is so stuck up on the idea that you’re stupid if you have a 4090 rn because the 5080 will surpass it for way cheaper

[@rahoolagin5841]

i loved this analysis but a 5080 performing at like a 70 series but priced at a 80 series is horrifying!

[@HybOj]

true. But jacket man will spin it in a way that 5080 is more powerful than 4090 and costs 200 dolars less. So the fanbois will jump on a 1400 USD 4080 to "save money"

[@Gofr5]

I don't understand why so many people are stuck on Nvidia. The only reason many want AMD to be competitive is so that they can buy Nvidia cheaper (which means AMD will never be competitive because they are just never buying AMD anyway). Honestly, how does Nvidia have such a death grip on so many people when we see them pulling anti-consumer shit like this on the regular? It baffles me so much.

[@erttttt]

@@Gofr5 Current-gen features offered by Nvidia are impossible to ignore. I've been on AMD, but I'm definitely switching to Nvidia on my next upgrade.

[@Gofr5]

@@erttttt What features? RT is overhyped. The differences in comparison videos I've seen comparing the two are.....well it's not enough to make want the performance penalty of RT. RT to my eyes does not look better. Upscaling? I hate upscaling. Everything should be played native. Native is best and always will be. Frame gen? I hate fake frames. Get that garbage out of here. What else is there that Nvidia offers?

[@rodrigoferreiramaciel4815]

​​@@Gofr5hey man... I get what you are saying, but if we can actually have an upscale that is just as good as native (or 95% the way there) for 60% of the price, I'm all in. And if I could get double the smoothness without the drawbacks (specially in scenarios where the game is bottlenecked by the cpu), all I'm for that one as well. Let's embrace new amazing tech, as long as it actually does what it is supposed to. That being said, you are completely right abt ray tracing, the one place where I find it absolutely amazing is in the possibility of playing old games remastered with the nvidia tool thingy they used to remake quake and doom II using rtx. That shit is awesome. One of the techs that have lately making me very excited to see where the future leads us.

[@beemrmem3]

I believe the second scenario is correct. The 5090 will be $1999 and the 5080 $800. Of course it's actually a 70 class card, but hey... Nvidia will act like they are re-introducing affordability

[@band0lero]

That’s a really good point. Ainda it could leave room for a 5080Ti with a cutdown 5090 chip. Also, 4080 to 4090 performance gap, although larger than other generations, was ~30% at most, even with a more cut down chip relative to past generations. If 5080 is 40-50% of the pricing (under $1000) of the 5090 and delivers ~70% of its performance, NVIDIA will definetly spin that on their favor.

[@nossy232323]

@@band0lero 70% of the performance of a RTX 5090 would mean that the RTX50 scales really badly. I don't see that happening. The 90 series literally has double the specs of the 80 series this time. Of course the CPU will be the bottleneck in most games even at 4K with the RTX 5090, just like the RTX 4090 is bottlenecked by the CPU at 1440P.

[@ekinteko]

​@@band0lero This. I think Nvidia tried this and failed with the Fake-4080 and they will repeat it this time but wont back down. 5080 will be treated as a value card (for USD $799) by them and the (eg MKBHD) fake reviewers. There is a good chance we will get a $1200 RTX-5080Ti as a launch, but we won't get any RTX-4070 upon launch. The lower tier cards are in-competition with the previous gen cards, and Nvidia would want to wean the market dry first for their top-tier old cards. Later I am sure Nvidia will release the Super variants, to cover the gaps. So I think an RTX-5080-Super and RTX-5090Ti will come.... but due to the large gap, I think we might see the first RTX-5080-Ti-Super as well. We already saw this "double naming" recently with the RTX-4070-Ti-Super so I won't be too shocked. How can Nvidia do all this? If there is no competition, they will get away with it. There was no effective competition during the 10-series from AMD (Vega 7, Vega 64, Vega 56, RX590, very meh) which what started this with the RTX-20 Series being a huge cash grab from Nvidia. They barely served the market with the GTX-16 series and RTX-Super series once AMD released their RDNA-1 cards. And Nvidia were forced to compete again. Which is why the RTX-30 series looked so good, despite it being actually pretty average. AMD knocked it out of the park with the RX-6000 series, only to drop the ball the subsequent generation. So will Nvidia get away with it? Probably. The rumours are showing the AMD RX-8000 series will focus on price and efficiency, meaning Nvidia will have no competition above the RTX-5080. However, I would be pretty disappointed to spend any money above an RTX-4070-Super to be honest. The graphics stop improving beyond that point and we are at the point of diminishing returns. Just crank the settings to 1800p, upscale to 2160p, aim for 120fps, and choose the Settings carefully to a hybrid Medium-Ultra like some of the guides show. Games have become unoptimised in the past decade, so you have to be extra vigilant. I would rather throw money at upgrading the monitor than the dGPU, as you might be able to go from a HDR/IPS screen to an OLED/Dolby and improve the visual in that manner.

[@fleurdewin7958]

Well I don't think 5090 will be $1999 . It will be $2499 just like Turing's Titan RTX. 90 series are actually Titan class.

[@band0lero]

@@nossy232323 yeah, I’m under the assumption that it might not scale that well for gaming performance, not because of CPU performance necessarily, but rather because of architectural reasons, it might be way more AI focused than what we used to get.

[@Anon01X]

It's an insane thought that any single other creator on KZbin could've put this data together and made the same conclusion, not even just KZbin but other platforms of media. Even then, Hardware Unboxed is the only media outlet to do this, amazing. Great work lads.

[@BlackJesus8463]

KZbinrs have been saying this for years they usually just do it one card comparison at a time when a new model launches.

[@SamOnKBD]

Wow, this was very good analysis. The only thing I would add, is that around 700 series, yes, the Titan would have been the flagship, but enthusiasts were running 2-3 780 Tis in SLI instead of buying a Titan. So in some ways, enthusiasts were paying roughly the same as a 4090 back then and the top tier looked a bit different than just "the highest priced Titan card" being the top tier. That isn't to say that modern enthusiast is not expensive, it's just that if you had 2 780 Tis 10 years ago then you probably have no issue buying a 4090 today.

[@mark.082]

Nvidia is done neutering all their gaming consumer cards except the 90 class. The 60 class with series 3000, 70 class with series 4000 and now the 80 class with the upcoming series 5000.

[@Unhuman-0111A]

As a Pc Gamer I don’t like the prices but I can’t blame 100% Nvidia Bc Consumers are the ones That keep on Buying the blatant upselling Sad to say but until there’s a stop and restraint of our part Nvidia will keep on going

[@pdmerritt]

Oh, there's no doubt this will work out in Nvidia's favor and they know it because this is, exactly how the greater majority of consumers spends their money.

[@Gofr5]

Yeah I have no idea why so many people are stuck on Nvidia. It genuinely baffles me.

[@LuizFernando-li1du]

I blame AMD, their cards just aren't as feature complete as the NVIDIA ones. If budget is not a concern, people will just buy NVIDIA for the most part. The CPU market used to be just as bad, but when AMD came up with Ryzen CPUs things became better, even Intel started to offer better deals than before. They need their "Ryzen" moment in the GPU space.

[@MrKardovalk]

All started with apple with iPhones they call them useful idiots now other companies getting into it also

[@raunakmishra9645]

No they make small profit from consumers they make much more from selling ai chips so no impact 😢 means even if we avoid buying em they won't stop they are tooo strong to be controlled

[@kasmidjan]

Vote with your wallet gentleman... Walk the talk

[@jepulis6674]

Vote with your vote. Either they push out cards for gamers and compete, or we open the patents and let other companies compete.

[@sega_kid4288]

@@jepulis6674 it’s not that bleak my guy, AMD cards compete in pure raster and exceed when it comes to vram. If I could have found one at the time I think I would have been more than happy with 7900xtx.

[@pgamez4077]

ill buy 12

[@05DonnieB]

They won't, everyone will buy up these gpu's. Even reviewers who blast nvidia go out and buy their flagship.

[@Arccanos]

​@@jepulis6674 lolol

[@centurion1443]

not surprising at all. It looks like 4080 12GB all over again at 4080 16GB price!

[@arc00ta]

The 4090 has a higher percentage on the steam hardware survey than the 4080, so this isn't shocking. The 4070ti Super makes WAY more sense and if you're paying a grand you might as well go big because the performance delta between the 80 and 90 is so wide.

[@awebuser5914]

The 4070Ti Super is kind of meh... If you want performance, the 4080 Super is far superior for $200 more (or less!), conversely, the 4070 Super is a massive jump on the 4070 and is perfect for 1440p gaming for $200 _less_ .

[@arc00ta]

@@awebuser5914 I've had probably 20 different 4080 Supers come through my hands. The problem is, its not faster-er enough than the 4070ti Super to make it a real 4K card and at 1440p there isn't much difference. The 7900XT almost matches the 4080 Super except in RT and you can get those for $679.

[@awebuser5914]

@@arc00ta _"The 7900XT almost matches the 4080 Super except in RT..."_ Bullshit... The 7900XT falls behind the 4080 Super by between 15%-25% (or worse!) in raster and just forget about RT.

[@jayb2705]

I'm glad Kimi Raikkonen spends his post F1 career spying on Nvidia

[@kazefw3834]

Hello there fellow F1 enjoyer

[@skythundersky1544]

Bwoah

[@syncmonism]

It's basically just a hobby for him

[@kevinamery5922]

Leave him alone, he knows what he's doing....

[@novocainDaimon]

That comment is Keke, Rosberg approved!

[@rustler08]

I am honestly curious what NVIDIA will do. It's very clear that they were willing to backtrack on a price by going down in price with the 4080 Super, but that makes me hypothesize that they're either going to go crazy with the price of the 5080 and backtrack later if it sells poorly or they're going to simply price it well to begin with. I do believe it's going to be the former, simply because there is no way they're going to price the 5090 anywhere near the "reasonable" price of $1,600. After all, the 4090 was effectively sold out for the entirety of its existence, so they understand there is clearly more money to be made. As such, they're hoping that by having the 5090 priced higher more people will defer to the 5080. Now, if the rumors of a 5080 Super/5080 Ti are true that are based on the 5090 die, then they might possibly leave the 5080 at a reasonable price and come out with the 24GB 5080 Ti and have that essentially be price like a 4090.

[@TheZoenGaming]

Is Nvidia planning to have a distinct gap between a 5080, 5080 Super, 5080 Ti, and 5080 Ti Super so that AIBs can offer overclock models without coming near the next tiers?

You may be onto something. Finer stratification of the product tiers. Good for buyers who are informed and spend the time to analyse; bad the majority of buyers who go "look it's a xx70 card so it must be great right". Either way great for business...

[@T3m0xx]

ah yes, the upcoming 5080 generation. Most likely gonna also shove a 5080D somewhere in there

[@BlackJesus8463]

Not until 40 sells out in never years.

[@TheZoenGaming]

@@BlackJesus8463 40? 40 what? 4080? Can you be a little more clear?

[@rbh2na]

AMD: exits making HIGH end GPUs... NVIDIA: It's time to scam people buying 70 Series Card for 80 Class money 🤑🤑

[@rbh2na]

It' Just like the 3000 to 4000 Series...Selling 50 Class card as 60 Class (RTX 3050 / RTX 4060 [8GB, PCIE x8]), 60 Class card as 70 Class (RTX 3060 / RTX 4070 [192Bit Bus,12GB]) 🤦🏻‍♂

[@BlackJesus8463]

It's the last two generations all over again.

[@PinguinoPocho]

I changed my 1070 ti from his long service to the XTX. For how i see the market and what Nvidia is doing i had no choice.

[@Anonymous-sb9rr]

@@rbh2na The 3060 12 GB was just weird, it had more VRAM than the 3080. The regular 3060 only had 8 GB.

[@Lovecars1218]

@@Anonymous-sb9rrapart from the 3090, the 3060 12 gig might be the long lasting ampere card. I had one and honestly regretted selling it for a 3070

[@1ch1r1n]

Don't worry, they're leaving room between the 5080 and 5090 for a 5080 Super, 5080 TI, and 5080 TI Super!

[@noticing33]

😂 yep, which is weird cuse amd is no where this time so tye don't need to do that just release a 5080ti

[@justins1917]

Ya 5080 now is garbage

[@Anonymous-sb9rr]

Probably not, this time is different. There's a huge gap in die size between the 5090 and 5080, so they won't really have any options to make something halfway in between.

[@BenWillock]

It's worth mentioning the 30-series and after use a doubled shader core count due to the FP cores being able to also do INT calcs, which previously had thier own seperate unit on the core (it's a bit like SMT); the actual shader core count as comparable to previous gens should be halved.

[@BlackJesus8463]

fake cores!

[@angrynimbus270]

fake tflops also!

[@36cores]

This is what happens when there is no competition. Unfortunately, the 5090 will sell well for $2500, they are not too concerned about anything under $1000 anymore.

[@TobZeN666]

The 4090D should've been the 4080 specs-wise. Thanks Tim. The video I've been waiting for.

[@CaptainCave-t3n]

5080 should have 24gb minium

[@haukionkannel]

Just wait the super version next year… 2024 5080 256 bit memory bus aka 8*2Gb = 16Gb 2025 5080 super same 256 bit memory bus, with 8*3gb = 24gb of vram. And ofcourse higher price for that memory!

[@BlackJesus8463]

They should have 16 and 24 models.

[@haukionkannel]

@@BlackJesus8463 as i said they will do it! Not this year because 3gb modeules are not in the market yet! The main point is that 24gb version will cost $1700! While you can get 16gb version at $1300…

[@Neopulse00]

If it was 20GB it'd be fine.

[@123Suffering456]

I have heard rumours of the 5080 getting 24gb. We'll have to see what happens, I guess. It should be 24gb, really, but I wouldn't be surprised at all to see them going for 16gb again.

[@MrMuggsly]

This is the kind of content that I LOVE from HUB

[@Mark-s146]

room for the 5080ti to slot in with 24gb and increased bus

[@KeinNiemand]

they probably won't make a 5080ti so they can upsell people to buy a 5090

[@TRHardware]

AdoredTV said this years ago, that this is the way Nvidia goes in near future. xx80 cards will use xx60 class dies.

[@Desalater2]

numbers are made up, only the price matters

[@Eric-ct2ri]

@@Desalater2 the irony is prices are also made up and are numbers themselves...

[@ninthburn1199]

Was about to comment how he would point out all of this every time a new generation came out Shame he is not making videos anymore

[@Arccanos]

Lol Adored. Where the numbers are made up, the conclusions are biased, and the fans are dogmatic white knights.

[@Arccanos]

@ninthburn1199 Shame? Adored was pushed out of the industry because he literally was just making shit up, and there aren't enough AMD fanboys uneducated enough to sustain someone just lying constantly. He also got utterly blacklisted by the industry after he was exposed following the 1060/480 fiasco. Dude was a quack.

[@viniciusms6636]

Now we are talking. Thats the kind of discussion we need.

[@JusticeGamingChannel]

I think we put too much emphasis on a 'name' - What primarily matters is the price to performance, and relative position versus the previous generation. You can call it RTX Pink Poney Ti, it doesn't matter, what matters is how much does it cost, how does it perform, and how does that relate to the previous generation at that price point.

[@davidepannone6021]

This is what happens when you really have no competition.

[@Frigobar_Ranamelonico]

Thanks AMD for that; and paradoxically I'll be happy. Because, being an enthusiast, the 5090 seems as insane as the 4090 was. Unfortunately that doesn't translate to lower models

[@federicocatelli8785]

Thx consumers that keep buying Nvidia no matter the price

[@arenzricodexd4409]

and that's because of consumer.

[@Eleganttf2]

Careful, the amd fanboys will not like your comment now

[@KeithZim]

Competing for what exactly? Name a game that can't be played at max setting on AMD? Given that RTX is mostly a scam and you none of you want to admit your were ripped off, what other reason do we have to care about Nvidia?

[@leoni7649]

Good old times from GTX 1080

[@BonusCrook]

1070 was the goat :p

[@tim3172]

@@BonusCrook I wonder if "goat" will also included with "based" in the history of the downfall of human intelligence.

[@BonusCrook]

@@tim3172thats a pretty based take, you're the goat.

[@fairyfire6481]

Didnt by 2000 series cause they were not worth it Didnt buy 3000 since they were scalped for a long time Didnt get 4000 cause they sucked Now 5000 is following 4000

[@BonusCrook]

@@fairyfire6481 planning to grab a used 4070Ti S myself. Used Ada or RDNA3 is probably going to be where the value is for a long time

[@justin_hyde]

The glory days of PC gaming are behind us.

[@angrynimbus270]

Why though, not necessary "PC = high end desktop" Handhelds are thriving for example, those are also PCs

[@chinchilla5162]

As a data guy those tables are a great visual comparison of hardware and price. Great job

[@John-Wolfe]

The US has a trade embargo on AI tech being sold to China. Remember that the 4090 and cut down versions of it were seen as violations. Nvidia rumors indicate that the 5080 is the global launch card, including China. For it to be released there without getting sanctioned, they need to keep its power as close to 4090 as they can. If nvidia intends to sell the 5080 to China gap makes sense for a launch. I know tech channels don't like to mention geopolitics but in this case I think it matters. Nvidia, and specifically Jensen, can't afford to be found in non compliance that could have bad consequences for the company and him personally.

[@BlackJesus8463]

It doesnt have to be a global launch though its another one of Nvidias artificial limitations.

[@John-Wolfe]

@@BlackJesus8463 your right in that but that doesn't change what Nvidia intends to do. If the 5080 is what the rumors indicate then it lines up with a global launch including China. It also explains the gap. Plus it's bad for business if you can't sell the flagship card everywhere. Granted if this is the case I probably would expect to see a 5080 TI super or something if they think they can get away with it that fills the gap. It may be an artificial limitation, but they can explain said limitation to investors as being embargo compliant and if it sells well in China they will further justify the action. I could be wrong, and am prepared to be. But the logic for me checks out right now.

[@theofficialpollo]

Nvidia has really became a villain

[@parm2-x7h]

lol, what ?? where have you been for past 10 years ??? you just notice that about them . wow

[@soiltek2015]

16:34 most important takeaway.

[@eliadbu]

True, this what I always said. It's not the name that really matters but it's the price relative to the performance. People tend to care more about the x70 or x80 moniker than what it really provides - which big reason why nvidia tried to sneak in cards like "RTX 4080 12GB" which is 70 class card in disguise at 80 class price.

[@jshico3417]

good to see confirmation that the 4060 was indeed the 4050 and the 4070 was the 4060 hope the push back will be enough but i doubt so considering how far the AI brainrot martketing has propelled nvidia the 5080 needs to have 20-24 gb vram if they want to advertise it as a 4k capable card leave it to nvidia to disappoint before even launching, no wonder that upper managment is fuming they got leaked

[@avatarion]

"as a 4k capable card" Haha, they way things are going only the 90 class is 4K capable. They are saying: pay up or get lost.

[@canyongoat2096]

Yes exactly here is how it needs to be: The RTX 5090 needs to have 32gb vram, the 5080 24gb, 5070 ti 20gb, 5070 16gb, 5060 ti 16gb, 5060 12gb, 5050 8gb or else there is no point to ever upgrade from my 4060 Ti 16gb. I sure won't spend and get 80 and 90 tier cards with insane prices just to get 20-24gb vram in the future.

[@uss-dh7909]

AI Bot here, don't forget to thank your local CryptoBro who wanted to make money out of thin air. I remember getting lucky twice with both my 2015 and 2020 build that they were on a downhill trajectory before prices shot up again months later.

[@konstantinlozev2272]

GTX 1070 was faster than previous gen GTX 980 with double the VRAM and mucu, much cheaper!!!

[@MitremTheMighty]

Totally doing it to leave space for a $1200 - $1600 5080Ti Super, which they will advertise as "amazing value" compared to the 5080 and 5080Ti.

[@Tom_Quixote]

Even if I could afford it, I don't think I would buy a 600W GPU.

[@lucazani2730]

Cap man. The 5090 is going to be amazing. Let's be honest with ourselves everyone here hates Nvidia (I'm the first to do so) but the 90 tier class card will NEVER be underwhelming (performance speaking) , because it's what they're trying to sell you, because that's the card with the most margin. The 5080 will be trash on purpose to make you purchase the 5090. If the 5090 was trash as well, that means Nvidia would be instantly dead as a company. 600W? I don't care as long as thermals are OK

[@miha1999grobar]

​@@lucazani2730with 600w power consumption heat and room temperature is gonna be a real issue, if power is expensive where you live that's another issue and power supplies? You need at least a kilowatt PSU to run it so most people would have to upgrade that too. Another potential issue is the power connector, if they don't split the 600w on at least 2 connectors it's gonna be a hazard

[@suppasandwich3310]

I have 450 watt card and its too much. Sticking with 300 or less next time i upgrade.

[@BlackJesus8463]

@@suppasandwich3310 200w is a lot of heat.

[@rustler08]

There it is "even if I could afford it." No offense, but you have a perspective shift when you can afford things.

[@lostandabandoned7090]

I don't know if it's because I'm hitting middle age and I'm becoming more aware of it, or if it is really the pricing, but I'm put off. My last couple purchases on upgrading my computer were not happy or satisfying. I like computers, been my hobby since God knows when, yet all things considered, I'm done buying new parts for my computers', the main parts anyway. I might buy a better sound system or mouse and keyboard, maybe even put it in a new case eventually. The upgrading game?? Nah, I'll just learn to accept my system as it is. Then switching to gaming....I don't know, I rarely game anymore, just tired of it all I guess. I could have taken a neat little vacation with what I got into my computers. I need to live more in the real world I guess, less online, less in make believe fantasy lands. I think it's just a bit of everything that's led me here. It's never good news anymore is it?

[@Chris-yc3mm]

5080 is actually a 5070, with a high tdp and most likely a super high price

[@jamescampbell8482]

Nvidia are selling consumers the dregs now because we only make up about $2 billion of their bottom line. 10 years ago we made up substantially more than that as a market segment. I think game developers have to readjust and just really learn how to leverage the lower end cards, because when Nvidia has no competition this is what we get

[@greggmacdonald9644]

A factor that wasn't discussed but is relevant, is the export restrictions to the mainland chinese market, where a 4090 is barred, but a 4080 isn't. It could very well be that the 5080 is as cut down as much as it is, so it fits just under the 4090 limit, offering a (very) modest uplift over the 4080 in performance, but also allowing the 5090 to have the generational uplift it's capable of. This explains the big discrepancy between the 5080 and 5090 specs, I believe. Also: we have rumored names, which can "change", and we have no idea of the price yet.

[@walter1824]

16GB 4090 4k max settings RT is gonna eat that memory, leading to blurry textures and stuttering

[@UltimateLF]

For 1080p clowns

[@QuintinHowell]

Who remembers the cancelled RTX 4080 12GB? This is Nvidia back to their tricks.

[@meusana3681]

This is missing 1 very crucial point. Not all buyers are new buyers, most people in fact who buy mid to high end GPUs are repeat consumers. So if you launch a new gen card with a similar name, the assumption of the customer is that it is an upgrade...How many uninformed people do you think upgraded from 3070 to 4070 just to realize (or never realize) they could have kept their current card instead.

[@AshtonCoolman]

We need 900 or 1000 series pricing again. It's getting hard to justify $1200 to ~$2000 for a GPU.

[@scarletspidernz]

it wont ever happen again, GPU's are gonna cost the equivalent of the rest of the PC is what NVIDIA gunning for

[@hhectorlector]

I'm only a bit into the video and i already love it. Great work, Tim! I'm a big fan of charts/infographics done for stuff like this.. I've not seen video card specs broken down like this

[@somethinlike23]

Really gotta add in the RT cores for this analysis. The story of the 10 series and later doesn’t make sense without them factored in. Whether you like it or not RT cores will be an essential if not the main part of any rig in the future. Any hardware based feature that can assist software developers in making feature rich products will win out over the long term.

[@Ignisan_66]

16 GB VRAM for 5080 when RX 7800 XT {which sits between 4070 and 4070 Super} has 16 GB is criminal. 5080 should have something like 20 GB of VRAM and 5070 16 GB.

[@EHKvlogs]

lets see what happens after ai bubble bursts.

[@Arccanos]

Uh. If you think AI is a bubble...I got some bad news for you...

[@Dr.WhetFarts]

You are clueless, AI is here to stay and all companies without 60+ year men on board are looking into AI. AI has not peaked yet.

[@BlackJesus8463]

@@Dr.WhetFarts Hasn't made money for anyone but nvidia either.

[@Arccanos]

@@BlackJesus8463 lmfao AI has transformed just about every industry already. If you're not using it to help write long emails, help with projects, etc., you're years behind the curve.

[@blizyon30fps86]

Yeah the bad news is, that’s not a bubble. Ai is here to stay unfortunately

[@amixofgeekcontent]

Thank you for informing us. I didn't know the 4070 had a lower % of flagship gpu's shader cores than when compared to previous xx70 gpus.

[@PaulSpades]

invidia : latin Noun. - envy, grudge, grudging, jealousy, prejudice, spite -odium, unpopularity, dislike, hatred, infamy, malice, resentment, ill-will

[@fohhee]

Introducing......RTX 5080, RTX 5090, and also......RTX 5090 TI Nvidia learned their lesson this time

[@Horendus123]

If only you could get your hands on all of these cards and see how accurate the performance scaling is to these on paper specs. That would make one hell of that article and give extremely in-depth insight into price performance overtime.

[@Uthleber]

Cant wait for the RTX 5070 with max potential 50 SM, while the 4070 has 46 and the Super has 56 SM's. But 90 % keep buying nVidia, great work guys.

[@arenzricodexd4409]

90% of nvidia consumer is semi pro. why did you think AMD ditch their RDNA/CDNA strategy and combine them into a single architecture UDNA? because gaming market is a lost cause. just look how gamer cannot accept the reality when sony priced their PS5 pro at $700.

[@Uthleber]

@@arenzricodexd4409 and what has this to do with a completly mess up with the 5070 ? this will gonna be like 3060 ti vs 4060 ti.

[@arenzricodexd4409]

@@Uthleber the point is there is no need to give gamer a good card. the only one that really make noise was those niche enthusiast since they can no longer get their powerful GPU at cheap price.

[@Dr.WhetFarts]

You look at specs, actual buyers look at performance. AMD is below 10% marketshare for a reason. They offer jackshit.

[@Adreno23421]

Nah... The 5080 is at least 1199, at least. And the 5090 ill be something like 2199-2499.

[@ertai222]

F that. It won't sell at that price. The 5080 has to be under 1000

[@KarlTheExpert]

​@@ertai222It will be. And it will sell. There's no reason to low-ball a nextgen card even if it's mid-range.

[@Daryth84]

Thank you, looking forward to the cards in your fingers and your evaluation.

[@spartan8705]

With 32gb of VRAM, the 5090 sounds more like a workstation card with a bumped power budget than a typical gaming card I can't help but wonder if we'll get a 5080ti to fill the flagship slot with the 5090 being the extreme flagship

[@JakeySurani]

There could be a titan Blackwell or a 5090ti. 5090 only uses 170sms of the full 102 which is 192sms. Just like the 4090 only uses 128sms of 144 total. Which translates to 16k shaders for 4090 of 18.4k full die and 21.7k for 5090 where full die is 24.4k. The 3090 was 82sms of 84 sms for full die and the 3090ti was 84 of 84 the full die. Another fun fact if the 5080 has 84sms it has same cuda core amount as the 3090ti. It won't beat the 3090ti.

[@tourmaline07]

@@JakeySurani "Another fun fact if the 5080 has 84sms it has same cuda core amount as the 3090ti. It won't beat the 3090ti." Not sure about that... my 4080 Super beats a 3080ti by 35% or so despite having the same number of shaders... ;) Do you mean the 4090? I think it will beat a 4090 , but only barely.

[@JakeySurani]

@@tourmaline07 I was talking about beating it in terms of cuda cores not performance. A 4080 is 20% faster than a 3090ti but if u oc the 3090ti it's very close to 4080 perf ofc u can overclock a 4080 to. Slightly better node for 5080 and what 800 more cuda cores around 4090 performance is a safe bet I guess the 400 watts is nvidia crutch to get it just over 4090 perf but that's not a good thing.

[@Geno1isme]

@@JakeySurani The full chips are used for the professional lineup, e.g. RTX 6000 or L40. Much higher profit margin and basically infinite demand these days thanks to the "AI" hype. Same reason why the performance gap between the 80 and 90 cards keeps getting larger: The AD102 is designed for professional customers that don't care about price, the AD103 is for normal customers that don't have infinite wallets (but NVidia still wants "reasonable" profit margins).

[@tourmaline07]

@@JakeySurani Fair enough - I don't think Ampere cores are massively comparable to Ada ones but it is an interesting comparison tbh. Shows how much nVidia gained going from Samsung to TSMC.

[@falmatrix2r]

Selling a “5080” for 1k? Yup I’m not surprised

[@Rodjeurs]

Exactly the kind of video I expect from your channel. Very insightful content, great video, thank you!

[@roasthunter]

Nvidia doesn't want to sell you a good product as they'll be no incentive for you to upgrade again so soon.

[@user-nl7hw7uh1m]

Being scummy and anti consumer is the default behavior of most companies these days. There's no internal moral compass anymore, it's just about how much antisocial behavior they can get away with.

[@Koozwad]

yes, the so-called 'elites' are screwing regular people over in every way imaginable

[@PaulRKeeble]

The other analysis that often places cards not based on their name but relative to their historical position is basing on the die size itself. While the achievable die size has gone up to an extent it used to be the case that the x80 was a ~500mm^2 die whereas since the 600 series its been more in the 250-320mm^2 range. One thing Nvidia has been doing is the die size has shrunk in many generations and we see that in the pricing gap from top to bottom becoming relatively large. In effect what they are doing is charging a lot more for the large die sizes creating the new performance tiers above the x80 and that actually compared to history they would just be an x80 in the 500 series range. Die size is an important way of tracking what class a card genuinely is as the industry for the most part has been able to make the same sized dies throughout.

[@Ale-ch7xx]

Always keep your mind open about buying competitors my fellow gigachads.

[@MarginalSC]

What competitors. AMD isn’t bothering. Intel?

[@Ale-ch7xx]

@@MarginalSC Be bullheaded then. I am sure the 7900 XTX would be faster than the 5070. I am on an Arc A580 but I know most people don't have the time to deal with driver issues. Money drives a company not people crying but still buying.

[@MarginalSC]

@@Ale-ch7xx Depends on what you want. 4k with ray tracing? There's nothing comparable.

[@Ale-ch7xx]

@@MarginalSC Then be the 0.91% percent of gamers that has an NVIDIA GeForce RTX 4090 or any of Nvidia's high-end stuff.

[@MarginalSC]

@@Ale-ch7xx I have a 4080, so no problem there. My point is, the competition's rather thin in that space so there isn't much of an option. It'd be great if Intel or someone else stepped up.

[@Trickzter1337]

Again they’re trying to sell 70 card as 80.

[@tonyjohansson8395]

This list should go back further. These Nvidia cutdowns started earlier. For the GTX 600 series there were rumors about the GTX 680 (GK100) and GTX 660 (GK104). Rumors where that the GK104 was AMD's flagship HD7970. The GTX 660 (GK104) was then renamed to GTX 680. There was also rumors that Nvidia had problems with power, thermals and yields on the GK100. And that the previous generation of Fermi GF100/GF110 being know for have bad thermals + high power requirements played a role in the rebranding of the GK104. Below information might be a little off as I'm pulling from memory. The top GK100 (Real GTX 680) supposedly had 2560 shaders, The cut down GK100 (Real GTX 670) supposedly had 2304 shaders. The most cut down GK100 (Real GTX 660ti) supposedly had 2048 shaders.

[@Lebon19]

Nvidia: but... but... they will allow you to do AI suuuuuper fast! that's why we price them so high! Everybody else: we don't give a single fuck about AI; we want to game.