NVIDIA Reportedly Readies RTX 2060 12 GB SKUs for Early 2022 Launch

[TheinsanegamerN]

I have so much in the backlog to keep the old GTX 1660 busy, I can easily wait 3-5 years for a new GPU and not notice.


Outside of the "excessive sparkly bits vs MOAR GPU" rat-race, in many cases recent PC games have been "advancing" in all the wrong areas:-

1990's vs 1980's innovation = Birth of 3D, entire new genres (FPS, RTS, etc) and the modding scene, simple dungeon crawlers turned into Baldur's Gate epics, hundreds of entirely new AAA franchises being created (Hitman, Tomb Raider, Far Cry, Doom, Wolfenstein, Civ, etc), going from "PC Speaker beeps" to orchestral soundtracks and Aureal 3D positioning, 16-colour CGA to 16.7m colour SVGA+, etc.

2010's vs 2000's innovation = How to shoehorn trashy mobile "Freemium" monetization mechanics into full priced PC games and get away with it, "surprise mechanics", explosion in cheating for online multi-player, how to pull off a Call of Duty with almost every mainstream AAA franchise and keep recycling the same 15-30 year old IP over & over because 'new ideas are risky', etc...

Honestly, part of me is glad PC gaming hardware is being screwed up now and not 20 years ago.

Modern AAA gaming has been trash, the entire PS4/XBONE generation was a huge dissapointment to me, especially after the 2008-2012 period where we were getting one banger after another.

My worry is that even if you are interested in older games or less demanding titles, the price of a 1650 super or 5500xt is absolutely bonkers. If you dont have a dGPU already you're just plain screwed until prices finally fix themselves. If my vega 64 were to die on me, I'd have no recourse for a replacement even if I wanted to downgrade.

 

[BSim500]

At the same time, graphics technology was already at a point, in 2016-17, that for almost everyone 'it was enough to look good'. Thát is why these older cards last.

I keep looking back to Indie's like The Vanishing of Ethan Carter and thinking, wait a minute, they managed to do that on 2GB VRAM cards in 2014... Where's the 'progress'?... It seems the more you given them (hardware) the less they now seem capable of doing with it. And that gorgeous water-colour palette / art style of Dishonored 1 still looks better to me that technically superior "4k brown / grey" palettes of several years newer games with 80% slower frame-rates.

Modern AAA gaming has been trash, the entire PS4/XBONE generation was a huge dissapointment to me, especially after the 2008-2012 period where we were getting one banger after another. My worry is that even if you are interested in older games or less demanding titles, the price of a 1650 super or 5500xt is absolutely bonkers. If you dont have a dGPU already you're just plain screwed until prices finally fix themselves. If my vega 64 were to die on me, I'd have no recourse for a replacement even if I wanted to downgrade.

Yeah I know what you mean. Luckily I do have a spare GTX 1650 Super I was thinking of selling (they were £320 on Ebay at one point). Shortages are so extreme here in the UK (even the GT1030 DDR5 is sold out) I'm keeping that as a backup and simply "bunkering down" until things return to normal.

 

[Fouquin]

Seen this rumor a few days ago and I strongly suspect that these actually are their infamous turing-based CMP mining cards. "Cheapified" 1660 PCB is a dead giveaway.

I'm really confused as to how that's a giveaway. The 1660/Ti, 2060 and 2070 shared various PCB designs. So you can't really get a good sense of what it is based on the fact that it uses the same PCB as a hundred other cards. If they're making a 12GB 2060 they'll obviously be reusing the board designs that were already being reused in the first place.

 

[Sisyphus]

2010's vs 2000's innovation = How to shoehorn trashy mobile "Freemium" monetization mechanics into full priced PC games and get away with it, "surprise mechanics", explosion in cheating for online multi-player, how to pull off a Call of Duty with almost every mainstream AAA franchise and keep recycling the same 15-30 year old IP over & over because 'new ideas are risky', etc...

Honestly, part of me is glad PC gaming hardware is being screwed up now and not 20 years ago.

I only play now and then, like modding games and content creating. RT, cuda and tensor cores are a great technological advancement, accelerating many calculations by 5-10. Yes, game engine development needs time to implement the new potential. And yes, the game industry is mainly for gamer of age 10-20. The older you are, the more infantil are the games.
And your conclusion is absolutely wrong. The high price is for high demand. In a screwed up/crashing market, GPUs are cheap because nobody wants to buy them. The oposite is the reality. There are so many buyers . . . the producers can dictate the price.

 

[Tardian]

current $500 average... For a $349 MSRP, 2019 graphics card.

800-900 AUD for an ultra 1440p@60fps card in some games? Old school?

 

[Sisyphus]

Yeah! A sub-performance card of the 3060 that has just as much VRAM and won't really be able to use it all. Sounds like a waste of RAM.

Nvidia scared that Intel will make bank off the sub-RTX 3060 gaming segment or something?

The 2060 placement has a very easy explanation: 12 nm TSMC process runs smoothly, capacity is available, the main product lines of AMD/nVidia are 7/8 nm. More VRAM is always welcome for content creators and miners.
Nobody is scared of Intel, but there is a high demand and 7/8nm lines, TSMC and Samsung are booked out, so you keep 12 nm products alive. That's a huge profit margin for a 12 nm product. I see absolutely no reason, why nVidia should stop the rtx 2060 line.

 

[RedBear]

The 2060 placement has a very easy explanation: 12 nm TSMC process runs smoothly, capacity is available, the main product lines of AMD/nVidia are 7/8 nm.

If it was just because of available capacity they could have just ramped up production of the RTX 2060 Super, overall it remains the superior card because of the larger memory bus and higher CUDA/RT/Tensor counts; I might be completely wrong, but personally I suspect that they're just getting rid of a surplus stock of TU106-300s.

 

[silentbogo]

I'm really confused as to how that's a giveaway. The 1660/Ti, 2060 and 2070 shared various PCB designs.

I understand your confusion, so need to clarify a bit if you aren't familiar with details.
It's not necessarily about shared designs, but about PCB model numbers. CMP30HX and 40HX are built on PG161(cheaper 6-layer PCB), the same reference PCB that's used for this new card, while regular 2060 and above usually uses PG160(10-layer). PG161 is also common for 1660[ti]. AFAIK, only EVGA RTX 2060 KO the absolute cheapest 2060s use PG161.
Speaking of which, @Raevenlord , you have a typo.

 

[defaultluser]

The 2060 placement has a very easy explanation: 12 nm TSMC process runs smoothly, capacity is available, the main product lines of AMD/nVidia are 7/8 nm. More VRAM is always welcome for content creators and miners.
Nobody is scared of Intel, but there is a high demand and 7/8nm lines, TSMC and Samsung are booked out, so you keep 12 nm products alive. That's a huge profit margin for a 12 nm product. I see absolutely no reason, why nVidia should stop the rtx 2060 line.

So, where are you going to Magic-away that 2GB x 6 GDDR6 chips from?

This thing will end-up being held-back by one or two seemingly superfluous (but absolutely necessary) components, making impossible to buy reliably at msrp!

GDDR6 memory prices to increase 8-13% by Q3 2021: hurts GPUs, consoles

GDDR memory prices will hurt Xbox Series X/S and PlayStation 5 consoles, as well as NVIDIA GeForce and AMD Radeon graphics cards.

www.tweaktown.com

 

[Sisyphus]

If it was just because of available capacity they could have just ramped up production of the RTX 2060 Super, overall it remains the superior card because of the larger memory bus and higher CUDA/RT/Tensor counts; I might be completely wrong, but personally I suspect that they're just getting rid of a surplus stock of TU106-300s.

You can only sell, what you have in stock. In the current market situation, they can sell everything with high margin. 12 GB VRam is marketing for gamers and real value for content creators/students. For CAD, RT, encoding etc. every VRAM counts, because VRAM is used as fast cache by many applications.

 

[Fouquin]

I understand your confusion, so need to clarify a bit if you aren't familiar with details.
It's not necessarily about shared designs, but about PCB model numbers. CMP30HX and 40HX are built on PG161(cheaper 6-layer PCB), the same reference PCB that's used for this new card, while regular 2060 and above usually uses PG160(10-layer). PG161 is also common for 1660[ti]. AFAIK, only EVGA RTX 2060 KO the absolute cheapest 2060s use PG161.
Speaking of which, @Raevenlord , you have a typo.

And the 1660 Ti reference was PG160. Hence the confusion to the original statement that being PG161 was a dead giveaway for these new cards being remarked CMPs. 160/161 overlapped almost the entire range, so criticizing that these new cards must be remarked CMPs seems weird. The conclusion I'd come to is that they never stopped producing PG161 and so it's what had tooling ready for this kind of product. Not that they're cannibalizing another product to do it, and in turn shoving lower quality cards to the consumer nefariously.

 

[matar]

$349 MSRP

 

[Sisyphus]

Not that they're cannibalizing another product to do it, and in turn shoving lower quality cards to the consumer nefariously.[...]

There is no other low budged nVidia card with 12 GB VRam. Nothing to cannibalize here. The product fills a long open void. Unfortunately, if it comes a little late, I would have liked the card in 2019.

 

[Arkz]

Waste of silicone.

It's a graphics card, not a pair of tits.

 

[TheoneandonlyMrK]

It'll play games well but a bit drab sounding as a prospective buy.

 

[DeathtoGnomes]

Likely a limited run, I'm guessing the chips were leftover from lost sales and refurbished RMA cards.

 

[Deleted member 185088]

Because it's the only Turing RTX card that they can afford to sell at around $200 (aand it should be faster than Intel's 1060 performance level)

And right now Intel is threatening with RTX. (might be slower at it than the 1660 Ti with the same effects turned on, but the PR is strong with Intel)

Of course, unless we have Etherium go Proof of stake before this launches, this will all be pointless.

The 2070 has the same die size, it shouldn't cost more especially with 8gb rather than 12gb of GDDR6.

 

[N3utro]

This won't change anything to the current situation. As long as gpus will be easy money making machines with mining the miners will still fight to buy any available gpus.

What will change it is when major states will realize like China that cryptocurrency is a threat to their economy and ban it. Then the bubble will burst, and the market will be saturated with gpus at all time low prices. Just wait and see this happen.

 

[isvelte]

I can't. Neither can millions of others. Low latency high speed internet service is still a pipe dream in large swathes of the US. I'm not even in a rural area and I can't get ANY wired high speed internet service at all. Extremely unreliable and slow 4G, and equally slow and outrageously overpriced fixed wireless is all I have to choose from. They can barely handle a 720p youtube video without buffering, I certainly won't be able to use a game streaming service.

I live in a rural area (literally pineapple fields) in a 3rd world country (philippines) and I got lined fiber connection here since 2016. Although I dont think theres any game streaming service/servers available anywhere in southeast asia lol

 

[Tardian]

I live in a rural area (literally pineapple fields) in a 3rd world country (philippines) and I got lined fiber connection here since 2016. Although I dont think theres any game streaming service/servers available anywhere in southeast asia lol

In that case, I live in Australia, which therefore must be a 4th world country because we have a Hybrid Fibre Coaxial (HFC) connection.

 

[ZoneDymo]

ya know, im actually suprised Nvidia did not put in a LITTLE bit of extra effort on the card and then just called it a RTX3050.

 

[Vayra86]

It's a graphics card, not a pair of tits.

You won the internet today.

 

[Richards]

ya know, im actually suprised Nvidia did not put in a LITTLE bit of extra effort on the card and then just called it a RTX3050.

They did its in laptops which is a bigger market

 

[AusWolf]

Considering that it's still a capable 1080p gaming card, the comeback would make sense - unlike the 12 GB VRAM, which is totally pointless. I'm just surprised that nvidia managed to squeeze in some allocation for Turing GPUs at TSMC.

 

[AnarchoPrimitiv]

I can't. Neither can millions of others. Low latency high speed internet service is still a pipe dream in large swathes of the US. I'm not even in a rural area and I can't get ANY wired high speed internet service at all. Extremely unreliable and slow 4G, and equally slow and outrageously overpriced fixed wireless is all I have to choose from. They can barely handle a 720p youtube video without buffering, I certainly won't be able to use a game streaming service.

I know how you feel, I moved to New Hampshire a few years ago, and the first place I moved to ONLY had DSL with 1.5Mbps service..... That's right, 1500Kbps....then I moved to a slightly more populated area and now have 1.2Gbps (1200Mbps) for $33.99/month is you can believe that.... I've had a 10GBase-T home network for years and now my internet actually makes real use of it.... But yeah, that first year of 1.5Mbps was horrible, had just built a new PC and couldn't even claim the free game downloads that came with my hardware (I remember I was downloading Borderlands 3, it had taken me days to get to 60GBs, and then Epic's stupid program just restarted the download for no apparent reason) couldn't even watch full HD YouTube.