NVIDIA NERFS THE RTX 5090? Memory & SM Specs See HUGE Cut
HTML-код
- Опубликовано: 29 май 2024
- NVIDIA NERFS THE RTX 5090? Memory & SM Specs See HUGE Cut
If the latest RTX 50 Blackwell rumours are true, Nvidia seems to have NERFED the RTX 5090, as the latest RTX 50 specs leaks point towards the memory and SM seeing a HUGE cut. Join us as we examine the latest leaked RTX 50 specifications, how they line up with previous reports, and how likely it is that the RTX 5090 flagship GPU has been nerfed in both memory and SMs. RTX 50 Blackwell graphics cards are highly anticipated, especially as RDNA 4 will not be entering a high end card into the race. Will they still manage to meet the expectations of specs and gameplay and ray tracing performance?
We also have some updates on the as yet unannoucned RTX 50 SUPER / TITAN cards, with some very early specs, and what kind of improvements we could expect, and a small update on AMD RDNA 4.
LINKS
www.chiphell.com/forum.php?mo...
x.com/OC3D/status/17950411063...
Subscribe - ruclips.net/user/RedGamin...
Join our Discord - / discord
Tip on PayPal - paypal.me/RedGamingTech
Whokeys 25% off code：RGT
Windows 11 Pro $23,2:biitt.ly/581eD
Windows 10 Pro $17,6: biitt.ly/9f0ie
Windows 10 Home $15:biitt.ly/XYm9o
Office 2016 $28,8:biitt.ly/cy7Vb
Office 2019 $48.6:biitt.ly/YInvw
www.whokeys.com/
AMAZON AFFILIATE LINKS
US - amzn.to/2CMrqG6
Canada - amzn.to/2roF4XO
UK - amzn.to/2Qi6LMS
GreenManGaming Affiliate link -
www.greenmangaming.com/?tap_a=...
SOCIAL MEDIA
www.redgamingtech.com for more gaming news, reviews & tech
/ redgamingtech - if you want to help us out!
/ redgamingtech - Follow us on Facebook!
/ rgtcrimsonrayne - Paul's Twitter
/ redgamingtech - Official Twitter
0:00 Start
0:01 Intro
0:15 RTX 50 - everything we know
2:40 New RTX 50 specs
4:34 RTX 50 SUPER / Titan cards?
9:43 RDNA 4 update
Royalty-Free Music - www.audiomicro.com/royalty-free-music 
Игры
They should nerf the PRICE
lol 😂
that would be nice yes
Vote with your wallet don't buy it if you think their nerfing is too much.
The price is the only thing getting "buffed".
Why would they do that?
The market will determine the price…
Knowing Nvidia, the 5090 will be cut down to 8GB.
They'll also sell a 12gb and 16gb version.
@@pf100andahalf 12GB version $5090, 16GB version $10180.
😂😂😂😂😂😂
6gb, with 8 lanes only, $2,500.00.
😂
The leaks for the 50 series cards have been terrible. Absolutely all over the place!
If you 'leak' every possible configuration, then you haven't really leaked anything - you're just throwing turds at the wall and seeing what sticks!
Many different variations are tested. The question is whether or not we get the crappy one and the good ones are reserved for workstations.
@@OllieOtterOxen the good one have always been reserved for professional market that will pay several times more.
If you throw a problem at ai like how to you maximize public hype via on a product infinite possible simulations. I bet it would look like this. Imo Nvidia is behind all the misinformation, leaks and rumors.
But you can make a lot of useless youtube videos about it for months! :)
@@cszolee7979 yep anything to keep Nvidia in the mindshare/ public eye. I recall the rumors of the 10 series were way off target and 4090 power requirements and efficiency as well.
How do you nerf a rumor
You just do.
my thoughts exactly lmao
The rumour mill invents a new streaming pile of bovine excrement to dump on top of their last wrong steaming pile. Nothing to see, move along
It just works
For maximum hype. From 512 bit bus to 448 bit bus from 4 slot throttle monster to a 2 slot 2 fan design. Nvidia is the master of market manipulation, misinformation rumors and all. 🤪
Just release the damn 5080 already so i can upgrade from my 1080 before tariffs and WW3
no more tarrifs. at least for a year
1080 LOL, they were released during WW2!
All the rumors have simply been exaggerated to generate clicks. There was never an incentive to have a fully configured 5090, for the same reason there was never a fully configured 4090. AMD doesn't compete on the very high end and the profit margin using the silicone on their professional data center and AI solutions instead is infinitely higher.
I agree. This generation I think nvidia will just be releasing poor quality silicon to gamers. The only reason to “cut down” these things is to chop off the defective bits. They are going to target a performance level just about 20-30% better than a 4090 to start with. Every 202 chip that mostly works is being shuffled up to AI workstations.
What happens if they don’t have enough defective chips to make 5090s? Thats easy, raise prices.
At this point, it feels as if these leakers are trying to leak all possible configurations
Psychic approach. Scattergun predictions for everything, then claim you got it right, forget the 9999999x that were wrong
Psychic approach. Scattergun predictions for everything, then claim you got it right, forget the 9999999x that were wrong
Figured they would cut it down. Couldn't imagine they put out a 512 buss on the 5090 maybe the 5090 ti or super what ever they decide to call it
They would be given to much performance at a base price.
Yeah, not to mention a 512 bit bus would mean something like 32gb of ram at minimum... just no way to see that on a non-professional card
Unless they decided to replace the 5080ti potential card and label it the 5090 because lack of competition. Imagine if the release the fuller fat as a 5090ti 512 bit bus gpu later on WHAT THE 5090 was rumored to be.
more interested in 5070 and 5080 and how much VRAM they will have. I hope rumors of the 5080 is false and that it will either have a different buswidth and/or more than 16GB of VRAM.
These rumors man. From day to day this tech youtubers change it. One day one specs other day completely other specs. This need to stop.
Not even day to day. There’s conflicting rumors just in this single video. I’ve never heard such rambling all over the place.
People speculate and feed him lies to see if he'll report it.
It's all over the media. Things did not use to be this bad.
Content pays the bill.
These are not tech channels they are just nonsense stop following and they will die off.
Hard to care about performance when price is going out of reach for most if not all gamers
welp. then fuck the 50 series.
60 series onwards will probably be worse, plus force you to combine only with nGreeDia CPUs that are coming out soon
No games no buy
@@razor19906 Ive been doing more with AI lately than gaming anyway, most of newer games kind of blow.
STEP 1 - design MULTIPLE variants of the same architecture
STEP 2 - when production gets CLOSER re-evaluate the engineering feasibility, the market and the competition
STEP 3 - use the more cut down variant IF the highest one is not needed. Why make a super high-end RTX5090 and make it harder to make an RTX6090?
STEP 4 - watch the INTERNET talk about your "nerfed" product.
Every day, a contradicting rumor.
Next news: Nvidia cancels rtx 50 series
No, Nvidia will release the RTX 4080 in China bc China wants to buy it. US gamers are a secondary consideration.
Hahahaha! 😂
I guess that means they confirmed AMD isn’t competing in the high end still, so why over-improve when they can hold back improvements for the 6090?
If it has
OMG did you REALLY believe they would release a none cut down card? Where have you been? This hasn't happened for several generations, only an idiot would believe that they would not do that.
Jensen really hate his customers
@@lolmao500 He needs more money for his leather jacket!
Classic nGreeDia
Oh no 448 bit bus, like a 512 bit bus is insanity nowadays given how how it is to transmit data at the frequency we want 448 bit is still massive.
For each 2GB memory chip you need 32 bits of bus width to address. So the vram on it will not be 32GB.
@@lanelesic Na I'm not even talking from a memory capacity prospective given that with GDDR7 you can get 3GB and 4GB modules, I'm talking from a bandwidth and engineering prospective. For gamers there is 0 need for more than 16GB of VRAM so it really doesn't matter.
for that insane price it should have 1024 bit bus lol
@@Etheoma There is no 3GB ddr7 modules yet and doing clamshell double sided vram is not an option because 5090 will have 3 pcbs and a huge cooler.
@@ProtectusCZ Eh I don't think you understand the engineering challenge of getting a bus that wide with GDDR.
given the spec rumor i can see them resurecting the titan name for the 700w name, and offer a just 12-15% increase of the 4090 which can be contributed to memory and power increase only
700w would be an odd amount when one of the new power connections is rated for 600w. Having another connection just for 100w seems weird.
@@Safetytrousers PCIe slot gives 75W
@@Felale That's even worse, having a connection for just 25w.
it wont need more than 300 watts. not even the 4090 does. it just came lil.bit too overclocked by factory. When it shouldnt be.
No competition is always bad for us consumers -.-
Yeah this monopoly is attrocious
Sure, but as a RTX 4090 owner, I don't give a shite! 😍 the card is a beast!
@@dtrjones Typical Biden voter. Get lost.
@@dtrjones Was thinking of getting a 5090. Now thinking of getting a 4090 after 5090 comes out.
Whatever happened to the 70 ti/super series cards matching or sometimes beating the outgoing top Nvidia cards?
So now it's going to take an 5080 series card do this against the current gen 4090?
I am intersted in only price to performane.......
A) they’ve probably run into issues melting 12-pin plugs when running full-tilt and would rather avoid the bad press
B) and he says just as I was going to type, leaves room to have more room for a 5090ti
C) smaller chip => less wasteful yields when you know your fanboys will buy them all regardless
Bet they won't nerf the price!
28 GB Vram is weird to be honest going from 24 to 28 would make a major difference but 32 would have been much better when it comes to local ai models. The new memory management diminishes the importance of an extra 4 GB significantly especially if you have more then one gpu. Memory band with is the real selling point of Blackwell for local ai inference and ffine tuning in my opinion.
I literally called this in the comments section of your last video on the 5090 lmao
Actually you say that it is typical for the new generation like the 5080 to exceed the previous generation top GPU, but it used to be the case not so long ago that the 70 series would match or exceed the previous gen top GPU.
No longer.
Hmm. Not sure how I feel about this. If they launch a 5090 and we "know" there will be a 5090 Ti or Titan coming later then I'm not sure who will buy the 5090? For the price of these things you would expect to have the top end car for quite some time like we did with the 4090. Certainly not sure I would upgrade mine.
when 4090 was launched some people are dead sure 4090 TI will come out because of the spec. it did not deter the sales for 4090.
its nvidia, they will squeeze every penny they can so likely they will use the full die for a titan card and use some cut down dies for the 5090 and 5090 ti
Will they at least show the RTX 50 at computex?
Of course now it's coming late Q4 still hell of a way
no. 50 series will be brand new architecture. nvidia will not going to hijack event like computex to launch brand new architecture. heck they did not even do it at their own GTC. nvidia will only launch 50 series at it's own event.
The 5090 IMV was always going to be 448-bit bus, leaving a 32GB, (or 48GB if 3GB modules) 512-bit later for a 5090 Ti for gamers, 512-bit on "launch" for professional.
i just want RTX racer.
Prices are set by the market.
5090Ti will have full die potantial to cover top rdna5 i suspect
Wont exist top AMD, they even said it
Hard to care about performance uplift from NVidia when it is prety much 1:1 with price uplift.
everyone the rumor has been nerfed. apparently now thats a thing lol. im assuming its going to be the same specs as 3090 4090 etc.
I really hope they at least then make a 5090ti/super etc. I don't want to upgrade to a 5090, just for them to release a new flagship I could have waited 2 more quarters for.
Tbh, Nvidia cutting down the memory bus from 512-bit to 448-bit isn't a big problem, in my opinion, as a 448-bit bus would give the 5090 a massive 28GB VRAM capacity along with a considerable boost to bandwidth as 448-bit is still a big increase over a 384 bit bus. 448-bit bus alongside GDDR7 should provide plenty of bandwidth and would be a big bandwidth increase over the 4090. What I am concerned about, though, is that Nvidia will cut the number of SMs by a large margin. I really hope they don't disable something like 30+ SMs, as that would be a massively cut down chip for the 5090.
At the same bus speed GDDR7 is 50% more performant than GDDR6.
Looks like they might be steering towards 5090ti this year
@@Safetytrousers What about GDDR6X because that's what the RTX 4090 is using?
I think they're going to put 172SM on the 5090 vs the 192SM of the GB202
We can't use more vram than the gpu and we only use vram to get game on card it should be done in real time not rendered of die in vram that makes lag n can't do real-time live native rendering is not pathtracing the gyx 80tis did it all in real-time native, rt cards are all using rendered which is the worse way to do it but they can't do it on rt the bus is only used to get the data on the gfx card and pcie 16x 8 n all at once 8bits to a byte twice that's asci code binary limit 16 bit programs sent but need board to do it n can't send if using cpu n vram separately because it needs to send it back hence latency can't do pathtraced or hq sound, no rt card today can do high end live gfx n snd at lossless they have to render and use codes because the hardware is using more than 1 core , loads of cores is not good you want 1 at fastest n we can't do faster than 16 so quad core we use each qtr 16 for os gfx n memory n control no drives to slow down its sent and 2 16 memory is all not needed its all turned down to be faster 32 lanes is slower cl32 is slower than cl16 xmp because it wasted like more keys on a keyboard doesn't make the typer write faster it just a waste . Games are now downgraded to sound better but are upscaled and can't do real-time raytracing or native it would use too much power so they wouldn't make money ao rhwy lied to make fools buy them thinking they won't notice and want more bigger numbers you all got sold a scam to keep shares up the factories shut down 6 years ago and Taiwan fixed the rejects to make a massive socket 80tis are not made by nvide they made the boards, the fab high end companies for vram memory n cpus are closed down, Samsung n China used old stock n soldered workarounds n win 11 is free for a reason it does the upscaling and fakery fps cpus don't matter the gfx cards are the fastest part always cpu n Windows hold it back and emulate the games so it's always running the os no fullscrwen native and nobody wants this downgrade rt and Intel don't make they have 20 years of stock
They would have to nerf it. Blackwell uses 1000 watts in its full form.
They will use the chips that are not Upto the Pro uses or Al, the Performance will be Carefully Picked to beat last gen but not give to big a Jump so they Can respond to compition if needed but also allow next gen to be better with out to Much effort whilst Putting Prices up Gen on Gen for Gamers.
Give me back 512 bits
No!
You just can't safely move that much power through that small gage wires. That's just the hard truth.
With the bandwidth of ddr7 and the amount of cache being "leaked".. id bet the 202 die is 384 bit anyway. That bus, ddr7 and amount of cache would easily get 50% + vs 4090 so why would nvidia wast so much high end dies by cutting them down so much? I do not belive they would.
Atleast the price will stay the same
nerf the dam price!
Ijust assume we were never going to get a 512 bit gpu, it was always 384 bit
Panzerlied is probably spoken Panzerleed, as ie is a long spoken e in german.
the 448-bit bus is just 2 less VRAM chips. Now with that said they also can go from 2GB to 3GB or 4GB GDDR Chips which can make up for the less amount of chips but get more VRAM. So up to 56GB on that card with a 448-bit bus. As the highest end GPU business line has 48GB having 56GB on it will make sense for a decent generational bump. Unsure GDDR7 work internally but if the controller in the GPU chip doesn't have to change from 2GB to 3GB to 4GB having the same GPU chip with 3 different VRAM Chip sizes would make sense for this generation.
The 2GB size for consumers the 3GB size for the business line and the 4GB size for the data center line all with the same GPU. Meaning alot more VRAM with no physical change needed by nVidia. Now with that said that is assuming the memory controller can be the same for all 3 sizes which is true for a CPU with DDR but unsure with a GPU with GDDR.
5080 performing at 4090 level isnt impressive at all. Even at $1000, its garbage value gen over gen. With ampere, their $500 card was equal to the previous gen flagship. They can do whatever they want, consumers don't have to buy them and the AI bubble will burst.
i just want a 5060 with about 16G and a 256Gb bus...
This video is counter productive!
Im still fine with my 3070 3800xt ryzen 1440p
Cut down to 1 slot, 1 fan, and no external power required.... and still beats RDNA4! Oh I forgot, AT 3X the price!
Bruh, if the 5090 is 1 slot width, it will use water cooling! Blackwell for AI are water cooled exclusively!
No thanks I prefer a quiet card with beefy cooler vs an airplane acoustics simulator.
@@ewitte12my 6950 XT ref is offended.. more like helicopter sim. 😂 It's not so bad I usually have it under 250watt total board power.
The nerfed one will be the 5090 and the original spec will be the 5090ti
Sorry to be that guy but its pronounced Panzer leet/leat.
Also 168 SM with that 448 bus?
panzer lead
@nipa5961 in German, it sounds closer to a T at least to my ears.
@@Death_Saved In German it's spelt "Lied" with a rather soft "d" in my opinion.
Why would nvidia raise the price of the 5080? The 4080 sold poorly and was basically laughed out the door, the super being $1,000 probably didn't well well either since it's been readily available and can be found for hundreds below msrp (at least Is here in australia) I can't see nvidia being dumb enough to price it wrong twice in a row, especially if its only 16gb lol. $999 or bust. As for the 5099 that's a different story .if it actually stays the same price and is ober 50% ahead of the 4090 I might consider it. But I have a feeling that's the card they'll try to push further. The 5080 though...only an idiot would pay over $1,000 for an 80 class card imo. No offence to the crazy people that bought a 4080 (terrible choice 😂)
Nvidia is utilizing the "all of it" cash configuration.
That's a massive nerf to just accommodate space for a Super and/or Ti version... I can forsee them also re-introducing the Titan moniker, with a titanic price too of course....
Hamstringing memory is a typical Nvidia thing to do, so theyll quite possibly release the 5090 with 448bit bus AND the slowest gddr7 ram possible, with the super having a few more cores enabled and slightly faster gddr to increase the memory bandwidth, then the Ti will have more cores again and fastest gddr7, so that finally the titan can have the full cores on die enabled as well as the full memory bus and fast gddr7... Titan at $3500 anyone?
Intel's way out of business
How so? Can’t judge the gpus considering they’re just getting started and their cpus still sell good
We love a lack of competition. Nvidia can cut down the 5090 and price it how they like because there's no competition keeping them in check.
It appears that this will be another generation of NVIDIA GPUs that I will be skipping.
So the titan will be th big one
A 5090 Titan/Ti for a small fortune makes sense if they are trying to get the most money possible out of the cards. Without competition at the top end, they can do whatever they want, that's true. Thumbs up for great videos.
I think rdna 5 comes out late 2026 or early 2027
Also don't forget that AD102 had a maximum potential of 800 Watts.
That’s only if Nvidia used the garbage Samsung node processor for a 4090ti.
Just because you can shove power through a chip doesnt mean it does more work faster.
It's like bragging about how much fuel your car consumes
I want to buy a 4090 but i am so torn if i should wait for the 5090
RTX 5090 should be launched around October 2024.
What kind of price would you pay for a used 4090 right now given that 5090 will be released in a few months? Just curious as I’m considering doing this
If you want 4090 go get it. RTX 5090 will be sold out till 2025. Scalpers may be coming soon.
@@Gonedeath my question is will the 4090 price change once 50 series releases?
nGreedia too cheap to give people 32GB of VRAM. Yet another reason to avoid anything nvidia like the plague.
Why do you need 32GB of Vram for gaming? 24GB hasn't been stretched yet.
And do you know what the cost of GDDR7 is?
@@Safetytrousers No, what is the cost?
@@randysalsman6992 I don't know. But asking for more than 24GB without knowing is very foolish.
But 32GB for the 5090 has been a long standing rumour.
@@Safetytrousersnvidia doesn’t care for gamers they care about ai in that case they need to get better on vram
@@An-intricate-man Nvidia is a very big company. They can easily have a team who are totally dedicated to making gaming GPUs.
Will amd cpu be at computex
Yes.
@@bgtubber ok thanks
lol they'll probably pull the old surname of ULTRA... like the 8800 Ultra and 6800 Ultra
so, small thing. NOT Panzer LIED as in English. a German lesson: Lied (pronounced Leed or Liit or Leet) means "song." So does 15th C English word 'leed." Panther song, or Tank song, or Armor song ---- whatever.
5080 will probably be a little cheaper than 4090 with comparable performance but less VRAM like 16-20GB at 1200 to 1400 and 5090 can cost anywhere from 1600 to 2500 since there's no competition from within either.
They said this is the biggest leap (though they say that every year) I’m expecting 5080 to be much better than the 4090
If I'm averaging 30fps with medium settings in most recent games like Ghost of Tsushima with a fucking gtx 1060 6gb and a Ryzen5 1500x, then surely I don't need to go to rtx 4xxx series, and every new amd cpu is overkill for 1444p ultra, Ray Tracing aside.
I can't wait for the rtx 6090 only for 3999.99$ ;)
The 4080 launch price was $1200, not $1000. If Nvidia goes over $1200, which made the 4080 the worst selling 80 class they had ever released, somehow managing to undercut Turing, then the card simply won't sell again. People with a bunch of disposable income will buy the 5090. People who maybe saved a little bit and couldn't quite justify $2200 for a 5090 will look at $1200 for an 80 class card and either stick with what they have, or buy a 4080 Super used for $650 or so, then in a year Nvidia will have to cut the price $200 bucks and release a slightly card while the old 5080's sit on shelves with retailers and AIB's screaming at Nvidia to offer some rebates on cards that are literally impossible to sell. I mena come on, they are $200 more than the Super models while performing 2-5% worse.
Jensen is too intelligent to price them at $1200 again, never mind over that unless something insane happens, they beat the 4090 buy 30% and have more than 16GB of VRAM as again, that is going to start turning into an issue at 1440P native/4K upscaled in single player games within a year of the damn card's release. Frankly I was astonished that they were going with 16GB to begin with, then I remembered the 3080FE I have in my PC and shook my head in utter contempt.
5090 Ti in two years xddddddddd
Well, i mean if AMD does not compete in the high-end, why would they release a monster of GPU, they better off actually sandbagging instead and ask for a ridiculous of money anyway lol
Reality will be that Nvidia says to customers that they wont release 5090 instead it will be 5080 AND 256bit bus card with 2000$ price. 5090 series will go to AI only and they cost 10k. I hope Nvidia buyers will be happy this is what you wanted still buying nvidia cards. Dont be sad now
5080 is still 16G?
Yep
@@sumins357 thank you！
This is bad why would they do this
the more they nerf it the less likely i am to buy it at this point im waiting until post blackwell fml
5090 needs to be 50-70%+ faster than the 4090. There are too many games where the 4090 can't hit 4k120+ without using DLSS. I want the 5090 to be a proper UE5 native 4k gpu.
As long as it's good for FS2024 VR 4k120 I will buy it. The rest of new games just suck after 2016
All I know is nvidia priced me out of pc gaming, I now only buy 1 or 2 generations old gps used.
If you expect 512bit, they will give you 448bit. If you now expect 448bit, they will give you 384bit :p. Nvidia is listening and they will downgrade your expectations.
Amd letting nvidia mess with people in the highend to hurt nvidias image, 7D chess by AMD
It's not a chess move people who are paying for 90 Class cards aren't crying in the RUclips comment section.
@@ISeeQuality-ox8ku They are, do a poll
AMD needs to step it up with the performance while maintaining a good price so nvidia will be forced to drop their numbers…no competition no reason to lower prices…..🤷🏾♂️
Of course they do, there's a reason they're called Ngreedia. And they will still take the full price as planned (Maybe 2500$ because of "inflation")
Perfect GPU for 2000€
Only question I have is power consumption, connector(s) and heat.
Everyone knows RTX 5090 will be better than RTX 4090. That's not even a question. The question is how much exactly and at what price.
It will be 384bit, wider card will probably exist but it will not be 5090. There's just no point for them to go with wider bus, GDDR7 will give them enough bandwithe to make it faster enough for average sheep to buy and there will be another model for professionals with "professional" price.
Only idiots can think that nvidia will in situation without pressure from competition give them best they could do. Just by going with narrower bus and thus avoiding additional heating (memory controllers generate significant part of die heat) will enable them to spare on coolers more money than people here can imagine digits.
everything getting more expensive while the majority of people remains so so, lmfao.
Nvidia are fuckboys for this. Thats why AMD RDNA5 is probably going to dethrone them entirely. God forbid Intel does it first.
NVIDIA is now holding performance on the upcoming 50 series cards to wait and see what AMD has planned, and if it's big they will release the full die spec.
I hate to support nvidia by buying their cards but I love raytracing😭
Oh, look! The bullshit from before is being replaced by more recent bullshit.
Didn't they stop the 4090s from going to China because they were too powerful? How is nVidia going to sell even more powerful GPUs in China going forward?
That is why the 5080 will not be as powerfull as the 4090, more like the 4090D, or they wont be able to sell it to China. Well, thats what I think at least. And the 5090 will be complete insanity, performance and price wise, and will be hard to even get one (Paperlaunch)
Nerf the 5080 so it can be sold there.
@@paleion0 The 4090D is about 5% slower than the full 4090, so if the 5080 hits that it would be a large uplift for the 80 class.
@@Safetytrousers Dude it's only a 25% uplift that's not really worth upgrading if you already have a 4080, wait for 60 series that should be a 60-80 percent uplift.
@@ISeeQuality-ox8ku It depends on the price. I don't intend to move to the 50 series from my 3090.
This youtuber seems to forget the Christmas market as a justification for release, Nvidia makes so much money off new gaming cards and AIB systems as holiday gifts.
It makes sense for Nvidia to cutdown since they don't have competition on the high end. The more they nerf more often you will buy an upgrade.
AMD should exit the discrete GPU business. They are a joke.
If they cut the 90 class so much... it has to be 1000/1100$ card, while the 80 class a 700/800$.... then the 90Ti is a 1500$
If the 4090 in Europe had starting price 2000 euros. I imagine the 5090 will only be worse
@@notaras1985 but the 4090 was not cut...... and stayed til today the highest end card.
@@greyraven9164 i will buy AMD. Screw nGreeDia products
Keep dreaming
@@smoothbraindetainer i know its Nvidia right.....
Or maybe its like the 4090, nerfed high end, cheaped out and you wont even see the ti for the biggest card😂
so just rumors lol