ixiFeels baaad maaan for your wallet.

The wallet is fine; I'm just short one kidney. But hey, that's why they come in pairs, right?

FrickThis is a nonsensical point. To quote myself, "Our entire hobby could not exist outside times of extreme abundance". The only thing a human needs is water, food and shelter, and maybe companionship as a fourth place.

Abundance is a funny word, used mostly by people who fuel and promote the AI bubble.

All other people live on a planet of scarcity and 600W GPU is not what a planet can handle for most people.

TheinsanegamerNSo we need to live in an active delusion in the blind hope that it will make 5090s reappear? Are we manifesting more DDR7 for nvidia to use to bring prices down?

We are PC enthusiasts, but above all, we are gamers—and these graphics cards are bought for gaming. Even when you want to promote your beloved Nvidia, it's gaming benchmark charts that you post. Therefore, we ought to view and discuss this card strictly as a gaming card. And not just this one: the entire RTX 50 series, the entire RTX 40 series, the entire RX 9000 series, the entire RX 7000 series, and so on. Why? Because the moment you walk up to Nvidia or AMD and tell them, "I view your product as something far beyond just a gaming GPU," you hand them the exact justification they need to price it exponentially higher. They are going to attempt that anyway since it serves their bottom line; we don't need to give them a green light.

The 5090 is not a pure gaming card. IT can do other things, as can the rest of the lineup. Denying that to live in a fantasy world only leads one to become completely unhinged from reality.

It was marketed as a gaming card and it is part of a consumer gaming lineup. Beyond that, graphics cards have capable background utility dating back decades; from the moment a GPU could decode video, it was doing more than just accelerating 3D graphics—yet gamers rarely let that alter their evaluation of whether a card was worth its asking price. Whenever gamers rushed to justify a card's price tag with "Yes, it's slower in 3D, but it supports CUDA," they ended up paying for that mindset out of their own pockets. That isn't smart. To draw a parallel: when CPUs integrated graphics processors, should consumers have been charged for a standalone GPU as part of the CPU price? AMD has released several APUs priced under that exact logic, but if someone is looking purely for CPU performance, they shouldn't touches APUs at those prices—unless they specifically require the integrated GPU as well.

That is precisely how we should approach gaming graphics cards. If my interest is gaming, I couldn't care less that a card happens to run fast in AI workloads. That must be our stance as gamers; otherwise, in a few years, we will be paying double the price simply because we'll be buying "AI cards" instead of gaming cards.

SOAREVERSORThe power connector problem is almost entirely constrained to the PC DIY gaming crowd.

We're well past blaming the connector on user error, there have been cases of it melting in SI built and OEM systems as well.

SOAREVERSORSince PC gaming is moving to the cloud or local APUs

Companies have tried and failed for years, the infrastructure to stream games for most people isn't good enough, and most would rather own their games.
Cloud gaming is also prohibitively expensive to companies that have or are trying to push it, as can be seen with Microsoft hiking prices of gamepass, even putting play time limits on lower tiers.

TheinsanegamerNGood. All gamers have done is whine and cry and scream about how evil nvidia is and how bad the connectors are and how bad the cards are and how huang is evil and how nvidia hates gamers and yadda yadda yadda. Everyone was filling their diapers wailing about "Oh NvIdIa HaS aBaNdOnEd GaMeRz WaAaAaAaAa NoBoDy CaN aFfFoRd GpUs wHeRe Is My $500 5090 WaAaAaAaAh!" for half a year, then when nvidia stops allocating the 5090 nobody can afford, we have to wail AGAIN about how evil nvidia is and how everythign is over and blah blah blah yadda yadda yadda.

Nvidia should go further. Get rid of everything above the 5050. Gamers dont deserve anything above that. Wouldn't want them melting any cables, now would we?

Well this is certainly...a take. Good for Nvidia shareholders, I suppose, but this is bad for the overall market especially with a company that has a near monopoly on PC gaming, and could do better with pricing and supply if they cared enough to. Nvidia is for the most part abandoning the market, and everywhere else pricing consumers out of buying a graphics card, they also played a major hand in causing memory and storage prices to skyrocket out of control.
Gamers only deserve a 5050 now, paying hard earned dollars for a GPU and expecting it to work without the cable melting was too much I guess. Then again I'm not sure why I'm even replying to what is clearly ragebait, or maybe you seriously are simping for a multi-trillion dollar mega corporation, but it's hard to tell on this forum sometimes.

BoggledBeagleAbundance is a funny word, used mostly by people who fuel and promote the AI bubble.

All other people live on a planet of scarcity and 600W GPU is not what a planet can handle for most people.

This is so far from the point, your point at that, it's actually meaningless. Maybe I will walk through it all at some point.

SOAREVERSORThe power connector problem is almost entirely constrained to the PC DIY gaming crowd. I've installed 5090s in bulk for business consumers and not had a problem. It's the DIY group that's having them with the exact same cards. OEMs shipping cards with the power connector are also not seeing problems outside of the usual rate.

The high rate is strictly with the DIY gaming crowd. Which shows you who's to blame. Let's face it building a PC takes no skill now compared to what it used to. There's no way around that. So there are a lot of people building gaming PCs now that have no business building them as they do not know what they are doing. Those who grew up with molex, IDE, floppy, SCSI, jumpers for all sorts of nonsense do not have these problems.

There's no need for nvidia to fix this because it's not happening where most of these get installed in abnormal numbers. It's not happening at the extremes people here claim if you buy a Dell gaming computer or even a custom one from say Digital Storm. It's sure as hell not happening with the Pro cards. Where it is going wild is a bunch of DIY gamers running into it and that firmly shows there are a lot gamers building computers who should not be. That's not on nvidia. This doesn't even rate in the long history of ways people have screwed up DIY builds it's a blip.

Since PC gaming is moving to the cloud or local APUs and nothing can change that there is no reason for nvidia to care about people botching something that had no business trying. The people installing this will all know what they are doing and the people who don't won't really be able to buy one. If you are worried about the power connector now simply don't build your computer yourself. Dude you're gettings a Dell! Part of the markup with premades is the lengthy testing and validation process along with skilled technichians. Most DIY gamers are not capable of doing this and wouldn't even know how. The card should have never been released to the DIY crowd at all. Hopefully nvidia learns their lesson going forward and removes the DIY market for their top options.

So its not a questionable design issue but a DIY problem? Your argument quickly hits a dead end when theres documented evidence showing the connector system is the problem. The main culprit being uneven current distribution and inadequate protections.

FrickThis is so far from the point, your point at that, it's actually meaningless. Maybe I will walk through it all at some point.

Not sure what are you talking about, but those datacenters that have been built caused at least in US major rise of electricity prices for normal people and many of these planned datacenters cannot be built at all, because there is not enough electricity available to power them. That is not a world of abundance, but a world of scarcity.

This does answer a question I've been thinking about for two years--what can Nvidia do in a 6000 series to increase performance? Most of the performance increases of the last 10 years have come at the cost of increasing TDPs, but I couldn't figure out how Nvidia could release a GPU with a TDP over 600W so that it could beat the 5090.

But if the 5090 is gone and the 5080 becomes the new flagship, then that's only a 450W TDP, and some hypothetical 6080 can be a flagship with a 500W or more TDP and increase performance relative to the 5080. In that case it looks like the 6000 series will be akin to the 600 series where the xx4 chip was the flagship.

Doesn't the 5080 super technically already exist in the 5090 mobile with the GB 203 chip and 24 gigs of GDDR7 vram at 256 bit bus lol? It was always here. :slap:
Also while cloud gaming has been a failure by many companies it will no stop the cronies to retry that strategy. Cloud gaming is also not a net zero cost for both the gamer/end user and cloud service.
Subscription costs are getting also significantly higher every year.
The service is becoming rationed off with monthly allowance caps.
Cloud compute is becoming more expensive to run/ upkeep for the service provider. Via hardware cost, security, networking etc.
That cloud compute power is also competing with ai token services besides gaming. If shark company a wants dedicated ai data centers for exclusivity then gamers will also have compete for cloud compute. Lastly while not necessarily cloud compute related, Steam's engineer came out with a new Pyrowave video codec that significantly improves latency.
Hopefully everyone is set for things to come and can weather out the storm. it's not like we didn't see this coming from miles away.

The irony of Ngreedia becoming who they are by the very customers that they are now abandoning. They have their market dominance now, so there is no need for those pesky human consumers that they left behind.

5090 not a gaming card, so what are 5090 owners doing with them, polishing it till the price goes up another 50%?

FoulOnWhite5090 not a gaming card, so what are 5090 owners doing with them, polishing it till the price goes up another 50%?

BigshrimpThe irony of Ngreedia becoming who they are by the very customers that they are now abandoning. They have their market dominance now, so there is no need for those pesky human consumers that they left behind.

This is stretching it to the max and then some. By now almost every enthusiast who wanted to have an RTX 5090 has one, and gamers aren't paying $8000+ on a video card. Unless they've got far more money than brains. The 5080 is already uncontested by competitors, and most gamers will do okay with a 9070 XT anyways. Just buy that and help AMD get out of the hole they dug themselves into over the past few years, especially if you have such a negative view of Nvidia.

lepudrukOk... If someone thought the 5090s were expensive, just wait and see - prices will shoot to the moon at the speed of light. Damn, I'm really glad I managed to buy mine for €3000.

I'm happy with the 5080 in my laptop that I bought for $1500, after selling the 16gb ram and 1tb SSD, and 2% cash back on the credit card.

john_We are PC enthusiasts, but above all, we are gamers—and these graphics cards are bought for gaming. Even when you want to promote your beloved Nvidia, it's gaming benchmark charts that you post. Therefore, we ought to view and discuss this card strictly as a gaming card. And not just this one: the entire RTX 50 series, the entire RTX 40 series, the entire RX 9000 series, the entire RX 7000 series, and so on. Why? Because the moment you walk up to Nvidia or AMD and tell them, "I view your product as something far beyond just a gaming GPU," you hand them the exact justification they need to price it exponentially higher. They are going to attempt that anyway since it serves their bottom line; we don't need to give them a green light.



It was marketed as a gaming card and it is part of a consumer gaming lineup. Beyond that, graphics cards have capable background utility dating back decades; from the moment a GPU could decode video, it was doing more than just accelerating 3D graphics—yet gamers rarely let that alter their evaluation of whether a card was worth its asking price. Whenever gamers rushed to justify a card's price tag with "Yes, it's slower in 3D, but it supports CUDA," they ended up paying for that mindset out of their own pockets. That isn't smart. To draw a parallel: when CPUs integrated graphics processors, should consumers have been charged for a standalone GPU as part of the CPU price? AMD has released several APUs priced under that exact logic, but if someone is looking purely for CPU performance, they shouldn't touches APUs at those prices—unless they specifically require the integrated GPU as well.

That is precisely how we should approach gaming graphics cards. If my interest is gaming, I couldn't care less that a card happens to run fast in AI workloads. That must be our stance as gamers; otherwise, in a few years, we will be paying double the price simply because we'll be buying "AI cards" instead of gaming cards.

You're not wrong, but also with the same mindset then everybody in the hobby the last 20 years has been the problem. Again, not wrong.

Dr. Dro





This is stretching it to the max and then some. By now almost every enthusiast who wanted to have an RTX 5090 has one, and gamers aren't paying $8000+ on a video card. Unless they've got far more money than brains. The 5080 is already uncontested by competitors, and most gamers will do okay with a 9070 XT anyways. Just buy that and help AMD get out of the hole they dug themselves into over the past few years, especially if you have such a negative view of Nvidia.

Exactly! So many are blaming NVidia, but IMO, should be blaming themselves a lot more. So much of home computing goods is "what the market will bare"; and we all taught NVidia that the market will bare multi-thousand-dollar price tags. Gamers here, saying "I'm so glad I got mine for $3000", are the reason they are $7000 today. That's not NVidia, that's our affect on the market. Singing praise above all praise like those cards were magic encased in eternal sunshine and not just willing, but eager to step up with that two - three - four, now seven thousand dollars - and kowtowing for the privilege. IMO, you all need to share that blame to a large degree - maybe even more than NVidia. And it started with the $2.5k ~ $3.5K 3090 y'all fell for; ramping up higher for the 4090 and again as much now for the 5090... And boom! Y'all have priced yourselves right out of the market by building it up like that. Who is really to blame here?

Now it seems NVidia has found a new segment to market to: "Workstation users" (or at least is hardening that segmentation). At least this market will mostly turn it over for themselves and in a lot if not most cases, the cards will pay for themselves - so NVidia is actually moving to a more ethical standing while us gamers are caught egg-faced with our pants down. This isn't even considering that it's not the position of any corporation to "care" about any market segment over another out of some philanthropic empathy. Their job as you've heard reported so many times, is to make money for their investors - and they're doing a hell of a job at it I'd say! We're just caught in the segment being left in the dust... and that's if you even consider a 5090 to be a gaming card and belonging to "our" segment.

I personally don't and never really have - the 3090 & 4090 either. IMO (most of the time anyway - when not in a dream), I considered those cards suitable for work... as in work-stations. It seems NVidia is coming around to a similar conclusion. The size and percentage of the entertainment (gamer) section of the market willing to spend that much on the last 4 generations of the 90 cards has continually shocked me at least. (And I do say four because you all know you fell for the April 1st 2090 and started counting your bank balances .) :D That so many of us were willingly manipulated by a simple name change from Titan to xx90 shouldn't be embarrassing for NVidia, but rather ourselves.

Guess the yields have gotten so good that it makes no sense for them to cut down the GB202 dies for a 5090.

YouseethisImagine the tsmc yields on 4-5nm. They must be close to 100% usable even with those large dies.

50-55% "perfect"* chips gd202.

*That's mean errates are below number and size and chip not needs of downgrade to lower class and locking of defective areas.

RedelZaVednoWth!? I thought Nvidia loves gamers!
Jensen would never dare to betray gamers for 🤑 would he?


Seriously it doesn't even matter anymore. I payed 600€ for 4070 TI and now 5070 TI costs 1300€ while my salary went up 5%. I guess I'm gonna keep 4070TI for 10 years like I did with 1080TI.

Cause we know the 5090 was the go to for gamers. Even before the market price increases it was at 2.5k$ minimum.

GPU prices have made my pinkstar pucker for more than half a decade now.

Normalizing insane prices during the C word days did not help.

But.. people flexed their credit cards anyway.. and now look at us.

Now.. imagine how many GPU's you can stuff into a building the size of 10 Walmart's..

We are no match for billion dollar budgets. Isn't capitalism grand?

Zero fan mode and overclocking are things of the past for me now lol..

Sooner than later, NV will add the ‘uber ultimate’ subscription tire to the NOW cloud gaming.
All those who want to feel the now gone 5090RTX ePin power will be entitled to pay for it monthly, forever.
You will pay 50$\month and get comfort knowing you will get early access to new DLSSx features and 1k max fps.
And it will be the end of PC. No more Personal for you. Only SC.

Good bless fxxx capitalism.

TumbleGeorge50-55% "perfect"* chips gd202.

*That's mean errates are below number and size and chip not needs of downgrade to lower class and locking of defective areas.

That would have been the case in 2020 with defect rates at 0.1 defects per cm^2. I bet that the defect rate now is below 0.05 defects per cm^2 even going as low as 0.01. That means 93% perfect dies if the yield calculator is accurate.

Also the rtx pto 6000 is cut down already, that’s why I said close to 100% usable dies.

The 3nm yields are rumored to be almost as good as 5nm but 2nm according to tsmc, better that 3nm.

I wish they disclosed the real numbers, perhaps they will not resist the bragging at some point.

MTS-CYou're not wrong, but also with the same mindset then everybody in the hobby the last 20 years has been the problem. Again, not wrong.

Not with CPUs. With processors, we looked strictly at performance. Even when the Athlon64 launched, we evaluated its raw performance against the Pentium 4. We didn't say, "Oh! It's 64-bit, so it's more future-proof, let's buy it". It was just as fast or faster than the Pentium 4, consumed less power, and—if I remember correctly—wasn't more expensive, so we chose it. When Intel introduced new SSE instructions, we didn't sit there saying, "Oh, it has SSE4 while the other model has SSE3, so $50 extra is justified for SSE4". Intel's QuickSync was treated as a very nice bonus, but only those doing video transcoding actually weighed it in their decision. Everyone else kept looking at FPS to decide whether Intel's asking price was worth it.
Even today with X3D processors, we look at FPS; we don't say, "Oh, they only have 6–8 cores, so they won't cut it, let's go for non-X3D options". We have always bought based on FPS. Unless someone needed a very specific feature, they didn't go out of their way to pay extra for a CPU that did something beyond delivering higher frame rates in games. And again, that was our exact mindset for CPUs—devices that are supposed to be general-purpose. Yet precisely because we always viewed them as general-purpose, and because the CPU market operates under a cross-licensing agreement between AMD and Intel that prevents structural imbalances, we never caught ourselves saying, "Oh, this one has feature X, so even if it's 10% slower and 20% more expensive, it's worth every penny".

With gaming GPUs—again, strictly for gaming—a completely different logic has dominated over the past 15–20 years:

• "Sure, it's more expensive, but it has this non-gaming feature. So it's worth paying extra".
• "Yeah, but I only want it for gaming; I don't care if it's good at transcoding or streaming. I have Netflix and I don't stream".
• "Well, you never know, you might start streaming tomorrow. So it's worth the extra cash!".

We have walked right into our own trap, which is why we end up reading arguments like, "Yes, GPUs are more expensive today compared to 15 years ago because TSMC raised its prices". Seriously? Why doesn't that same logic apply to CPUs? Today, you can buy a CPU with at least 6 cores using pocket change left over from delivery, yet we somehow justify a low-end GPU costing more than a mid-to-high-end GPU from a decade ago? Why? "Oh, right, because it supports AI". So what? What does that even mean? Does AI boost raw frame rates? Don't start with "DLSS 5". It's way too early. As proven with Ray Tracing, we'll be waiting 2 or 3 generations before DLSS 5 becomes a practical toggle in game settings.
In general, we've shown immunity against the marketing of CPU manufacturers, but we swallow GPU marketing whole without chewing. That is why today €200 buys you a perfectly capable processor, whereas €200 won't even buy you the cheapest current-gen graphics card.

P.S. This text is general in nature and isn't directed at you specifically; I just used your point as a springboard to lay out my broader thoughts on the matter.

john_Not with CPUs. With processors, we looked strictly at performance. Even when the Athlon64 launched, we evaluated its raw performance against the Pentium 4. We didn't say, "Oh! It's 64-bit, so it's more future-proof, let's buy it". It was just as fast or faster than the Pentium 4, consumed less power, and—if I remember correctly—wasn't more expensive, so we chose it. When Intel introduced new SSE instructions, we didn't sit there saying, "Oh, it has SSE4 while the other model has SSE3, so $50 extra is justified for SSE4". Intel's QuickSync was treated as a very nice bonus, but only those doing video transcoding actually weighed it in their decision. Everyone else kept looking at FPS to decide whether Intel's asking price was worth it.
Even today with X3D processors, we look at FPS; we don't say, "Oh, they only have 6–8 cores, so they won't cut it, let's go for non-X3D options". We have always bought based on FPS. Unless someone needed a very specific feature, they didn't go out of their way to pay extra for a CPU that did something beyond delivering higher frame rates in games. And again, that was our exact mindset for CPUs—devices that are supposed to be general-purpose. Yet precisely because we always viewed them as general-purpose, and because the CPU market operates under a cross-licensing agreement between AMD and Intel that prevents structural imbalances, we never caught ourselves saying, "Oh, this one has feature X, so even if it's 10% slower and 20% more expensive, it's worth every penny".

With gaming GPUs—again, strictly for gaming—a completely different logic has dominated over the past 15–20 years:

• "Sure, it's more expensive, but it has this non-gaming feature. So it's worth paying extra".
• "Yeah, but I only want it for gaming; I don't care if it's good at transcoding or streaming. I have Netflix and I don't stream".
• "Well, you never know, you might start streaming tomorrow. So it's worth the extra cash!".

We have walked right into our own trap, which is why we end up reading arguments like, "Yes, GPUs are more expensive today compared to 15 years ago because TSMC raised its prices". Seriously? Why doesn't that same logic apply to CPUs? Today, you can buy a CPU with at least 6 cores using pocket change left over from delivery, yet we somehow justify a low-end GPU costing more than a mid-to-high-end GPU from a decade ago? Why? "Oh, right, because it supports AI". So what? What does that even mean? Does AI boost raw frame rates? Don't start with "DLSS 5". It's way too early. As proven with Ray Tracing, we'll be waiting 2 or 3 generations before DLSS 5 becomes a practical toggle in game settings.
In general, we've shown immunity against the marketing of CPU manufacturers, but we swallow GPU marketing whole without chewing. That is why today €200 buys you a perfectly capable processor, whereas €200 won't even buy you the cheapest current-gen graphics card.

P.S. This text is general in nature and isn't directed at you specifically; I just used your point as a springboard to lay out my broader thoughts on the matter.

Yea I get it, but I meant entirely Nvidia in nature due to CUDA. We've used other metrics over the years for both CPUs and GPUs, but either way I wasn't disagreeing with what you were saying. Just that we've all had a part to play for the last 20 years.

TheinsanegamerNSee, this is the kind of thing I was talking about. It's demonstrably wrong, every tech company in the space spends exponentially more on R+D now then they did a decade ago, and those numbers are not going down. Node costs continue to rise above the rate of inflation, hundreds of billions are being spent on new production facilities, and transportation costs have skyrocketed due to fuel and tariff and wage prices.

And frankly, AI is a very useful tool. It's not a human replacement, but its capabilities are very impressive for what it can do, which can be readily seen in content creation and the coding space. Just look at game recompilations that are getting churned out now VS the rare release of a few years ago.

It's not "insanity", it is economics 101. That money is not being collected like Smaug's gold, it is being burnt, at insane speeds, to push what is seen as the new tech revolution.

Whether you love or hate AI (or anything in between), you've got to admit that this is how things work. If there wasn't an insane amount of money spent on AI, we would have plenty of cheap RAM and cheap GPUs available to us. The industry spending its money is exactly what's driving prices up, not inertia (which would do the opposite if it was the case).

john_.... That is why today €200 buys you a perfectly capable processor, whereas €200 won't even buy you the cheapest current-gen graphics card.

A CPU is a chip, substrate and a metal plate - heatspreader.

A GPU is a chip, substrate, small motherboard with all the power delivery circuitry, memory, cooler, fans, shroud and other structural hardware.

I am not sure how can you directly compare these two products. One is a single component and the second one is almost a whole finished computer.