News Nvidia GeForce RTX 3090 and GA102: Everything We Know

JarredWaltonGPU

Senior GPU Editor
Editor
Feb 21, 2020
553
454
760
0
Let me just post this here as well: There are some minor issues in the Micron "Categories of Ultra-Bandwidth Memory" table. RX5700XT sits under Titan RTX for GDDR6, but then the table says 12 placements and 12GB -- clearly not for Navi 10. The HBM2E AI Training Accelerator also says 6 placements and then lists 16-32GB, while the Nvidia A100 has 6 placements but only 5 active, with 40GB total memory.

Given everything that's been said about Ampere and RTX 30-series hardware, I think it's far more likely that those two items are the errors, rather than the GDDR6X reference to RTX 3090 being an error. 21, 21, 21, 21 ... too many instances of that number come up. Maybe Nvidia will change it to 20 Gbps and ship me an edible GPU hat to eat, just for fun? Mmmm, GPU hats...

We're basically two weeks away from final confirmation of many specs, though, and this is probably the last chance I'll have to publicly say anything before the launch. Or maybe not; we'll see. Micron pulled down the PDF, though, and I doubt Nvidia would have made a stink and had them remove the PDF if it didn't have a bunch of accurate details. It's my opinion and I'm sticking to it, but you know what they say about opinions.
 

daeros

Distinguished
Jan 29, 2009
30
1
18,535
0
Man, you guys really love whatever swag team green throws your way. This thing needs three 8-pin PEG connections, but Radeon VII draws too much power. You complain that it's only slightly faster than the 5700xt while ignoring that it beats the Titan RTX in compute workloads. And you're defending the idea of a $1500-2000 launch price. Tom was right, and I'll be filing this right along your 'Just buy it' article.
 
Reactions: 3ogdy

JarredWaltonGPU

Senior GPU Editor
Editor
Feb 21, 2020
553
454
760
0
Man, you guys really love whatever swag team green throws your way. This thing needs three 8-pin PEG connections, but Radeon VII draws too much power. You complain that it's only slightly faster than the 5700xt while ignoring that it beats the Titan RTX in compute workloads. And you're defending the idea of a $1500-2000 launch price. Tom was right, and I'll be filing this right along your 'Just buy it' article.
I’m not defending the price at all, I’m just saying what I think the price will be. And we don’t know how many 8-pin connectors this will use, or the price. The Radeon VII was okay at best, but at launch it basically tied the already old 1080 Ti in most workloads — except specific compute tests. It’s disingenuous to accuse us of bias while clearly exhibiting your own.

For the record, Nvidia has not provided anything for this article — no information at all, other than what is known about the A100. It’s okay for enthusiasts to be excited about a new extreme GPU, and I am a GPU enthusiast. That’s what this article is about.

If this is a $2000 card that’s twice as fast as RTX 2080 Ti, it will be impressive and yet too expensive by half. At $1500, it’s at least somewhat acceptable but still very much out of reach of even high-end gamers. I’d love to see it land at $1000, but that’s a pipe dream.
 
Reactions: Gurg
Nov 27, 2019
52
10
35
0
Predictions on AUD Pricing: I'm going to go for $4000 AUD. The xx90 might push it into a whole new pricing tier
Things are very expensive here, but current climate has pushed GPU prices 30%-40% higher than same time last year.
 
Reactions: spentshells

Gurg

Distinguished
Mar 13, 2013
416
35
18,820
1
If Nvidia prices this too high they will lose unit sales. In July 2020 Steam survey the 1080ti had a 1.57% market share vs just .91% for 2080ti. When released in March 2017 the 1080ti was priced at $700. whereas the 2080ti market price was around $1300 in Sept. 2018. The combined 2080 and super have a 1.7% market share priced at $700-750.

While the demand from the top end gamer is relatively inelastic with that segment willing to pay for the best, for everyone else the demand is elastic. If Nvidia gets too aggressive on pricing the 3000 series, many potential buyers especially in current economy will opt to just stay pat with their 1080ti and above which will run even a 60hz 4K or all lower resolution monitor at playable frame rates.

While the 3090 should certainly max out a 60hz 4K monitor, to take full advantage of its potential you would need to upgrade to a 144hz 4K monitor which starts at $700 . For me that is simply a bridge too far. If the 3080 is priced around $750 and offered enough performance increase to max out my existing 60hz 4K monitor then that would be a buy after initial pricing spike fades and partner cooling is available.

If you spend $2,000 plus on a 3090 and a 144 hz 4K monitor you will most likely only be half way to saturating the 144 hz and won't be able to fully saturate it until the 5090 in 4-6 years.
 
Last edited:

vinay2070

Distinguished
Nov 27, 2011
168
28
18,720
2
If Nvidia prices this too high they will lose unit sales. In July 2020 Steam survey the 1080ti had a 1.57% market share vs just .91% for 2080ti. When released in March 2017 the 1080ti was priced at $700. whereas the 2080ti market price was around $1300 in Sept. 2018. The combined 2080 and super have a 1.7% market share priced at $700-750.
So they essentially made more profit from the 2080Ti compared to 1080TI???

While the 3090 should certainly max out a 60hz 4K monitor, to take full advantage of its potential you would need to upgrade to a 144hz 4K monitor which starts at $700 . For me that is simply a bridge too far. If the 3080 is priced around $750 and offered enough performance increase to max out my existing 60hz 4K monitor then that would be a buy after initial pricing spike fades and partner cooling is available.

If you spend $2,000 plus on a 3090 and a 144 hz 4K monitor you will most likely only be half way to saturating the 144 hz and won't be able to fully saturate it until the 5090 in 4-6 years.
A lot of consumers will be using the Ultra Wide QHD and QHD at 144Hz and some even 200 Hz, so its not just 4K that would require a 3090. Also if you max out RTX, that would bring performance down a lil bit if there is no DLSS implementation.
 

Jim90

Distinguished
Feb 3, 2013
408
31
18,990
54
Quite simply, you would have to be a fully certified idiot to purchases an Ampere card BEFORE RDNA2 cards appear. We've seen what RDNA2 can do when significantly scaled down in the new consoles (i.e. Unreal 5 demo on the PS5 @4K) and we've read plenty. Now it's time to wait till BOTH companies release their products and we finally see proper reviews.
 
Last edited:
Reactions: 3ogdy
If Nvidia prices this too high they will lose unit sales. In July 2020 Steam survey the 1080ti had a 1.57% market share vs just .91% for 2080ti. When released in March 2017 the 1080ti was priced at $700. whereas the 2080ti market price was around $1300 in Sept. 2018. The combined 2080 and super have a 1.7% market share priced at $700-750.

While the demand from the top end gamer is relatively inelastic with that segment willing to pay for the best, for everyone else the demand is elastic. If Nvidia gets too aggressive on pricing the 3000 series, many potential buyers especially in current economy will opt to just stay pat with their 1080ti and above which will run even a 60hz 4K or all lower resolution monitor at playable frame rates.

While the 3090 should certainly max out a 60hz 4K monitor, to take full advantage of its potential you would need to upgrade to a 144hz 4K monitor which starts at $700 . For me that is simply a bridge too far. If the 3080 is priced around $750 and offered enough performance increase to max out my existing 60hz 4K monitor then that would be a buy after initial pricing spike fades and partner cooling is available.

If you spend $2,000 plus on a 3090 and a 144 hz 4K monitor you will most likely only be half way to saturating the 144 hz and won't be able to fully saturate it until the 5090 in 4-6 years.
The top tier gpu isn’t about selling high volumes. It’s all about marketing and being able to say team green have the best gpu’s in the market. As long as it recovers development costs and returns some profit it main goal is to raise the brands profile and help sell the rest of the range.
 

Chung Leong

Upstanding
Dec 6, 2019
277
81
260
0
The top tier gpu isn’t about selling high volumes. It’s all about marketing and being able to say team green have the best gpu’s in the market. As long as it recovers development costs and returns some profit it main goal is to raise the brands profile and help sell the rest of the range.
People don't like the feeling of getting the second best though. If the RTX 3090 doesn't exist at all, more people would step up to the 3080 from the 3070. Winning the gold is worth the sacrifice. Moving from bronze to silver--not so much.
 
Reactions: EricLane

Moleyman69

Honorable
Jul 29, 2014
21
2
10,515
0
I'd like to know if the 3090 performance is genuinely two levels above the 2080ti though or just one?

As I see it, the 3080ti should be one true level of performance above the 2080ti and the 3090 another level or performance above the 3080ti.

I'm worried though that the 3090 will only be one level above the 2080ti (ie basically really just 3080ti performance) but with nVidia charging at a two level price point.

I hope that makes sense.

Moley
 

nofanneeded

Notable
Sep 29, 2019
731
106
1,070
3
My 2080ti will handle more than anything I can throw at it at this point. My next purchase will be a future generation processor.
Not really ,in 4K gaming it is still lacking ... and when you pay more than $1000 for a GPU it should work well with 4K ... All of us have 4k TVs now and My LG OLED TV even has GSync in it.
 
Reactions: Gurg

Contiusa

Honorable
Oct 8, 2013
33
1
10,530
0
Are you guys working for Nvidia? 2k for a GPU? Is that some sort of joke / campaign to soften people's mind so than when then release a 1.5 / 1.7k GPU people will feel relieved?

Why some media people always do that? Every release we see someone pushing the prices to the roof.

These cards prices are inflated. The least people can write is that the price should go down, as expected. Ray Tracing is no novelty anymore. And if Nvidia comes with a 2k GPU, they will have to face the heat because customers and media did not expect that. And rightfully so, we are not stupid.

So now in the next generation you guys will tell us that you think it will come costing 3K, right?

Yeah.
 
Reactions: nightsc

Gurg

Distinguished
Mar 13, 2013
416
35
18,820
1
So they essentially made more profit from the 2080Ti compared to 1080TI???



A lot of consumers will be using the Ultra Wide QHD and QHD at 144Hz and some even 200 Hz, so its not just 4K that would require a 3090. Also if you max out RTX, that would bring performance down a lil bit if there is no DLSS implementation.
Can't tell if they made a higher profit on the 2080ti vs 1080ti as there was at least the indication of a 42% market share drop vs about a 60% pricing increase and no data on cost of production. More profit is a silly question anyway as the goal shouldn't be simply making higher profits but rather maximizing profits.
 

spongiemaster

Prominent
Dec 12, 2019
505
233
760
0
I'd like to know if the 3090 performance is genuinely two levels above the 2080ti though or just one?

As I see it, the 3080ti should be one true level of performance above the 2080ti and the 3090 another level or performance above the 3080ti.
No one who knows this information is talking, however, I believe this article got it mostly wrong, because it uses that Micron chart for as a basis of information it was never intended to be used for. If the 3090 is the prelaunch name for the next Titan, then it isn't going to have 12GB of RAM, it will have 24GB and it's going to cost north of $2000, while only being slightly faster than the tier below. Most people will ignore this card, like they ignored the Titan RTX. If the 3090 is the prelaunch name of the new top Ti, which wouldn't make much sense, then who knows.
 
Reactions: Shadowclash10

PCWarrior

Distinguished
May 20, 2013
99
26
18,560
0
With Pascal and Volta, you had the following hierarchy.
Titan V - $3000
GTX Titan Xp - $1200 (launched after the launch of the 1080ti to restore the pecking order)
GTX Titan X (Pascal) - $1200
GTX 1080Ti - $700
GTX 1080 - $600 (base models, dropped to $500 for base models after the launch of the 1080TI)

With the Touring cards you had this hierarchy:
Titan RTX - $2500
RTX 2080Ti - $1200
RTX 2080 Super - $800 (launched a year later)
RTX 2080 - $700-$800 (originally)

So basically in terms of pricing they moved the models a tier higher with the --80 model commanding the same price as the previous --80ti model, the new --80Ti model commanding as much as the previous Titan X/Xp and the new Titan being almost as expensive (83%) as the ‘professional’ Titan.

So with Ampere we will probably have the following:
RTX Titan (Ampere) – (to be launched later when the 3080ti will launch or a bit before) - $3000
RTX 3090 - $1600
RTX 3080Ti - $1200 (to be launched in 6-9 months right before/after Intel releases their gaming gpu)
RTX 3080 - $800

The RTX 3090 will essentially play the role of the Titan X (Pascal) but in an era where the lower tiers have gone up in pricing. I predict it will cost $1600 (or $1599). The RTX 3080 will be $799 for the founders. Later in the face of competition (or as a pre-emptive attack) Nvidia will release the RTX 3080Ti and offer the same (or almost the same) performance as the RTX 3090 at the current 2080Ti pricing ($1200) in the same way the 1080ti fared against the Titan X (Pascal). At the same time or a bit before they will also introduce the full RTX Titan (Ampere) for $3000 (or $2999)
 

JarredWaltonGPU

Senior GPU Editor
Editor
Feb 21, 2020
553
454
760
0
No one who knows this information is talking, however, I believe this article got it mostly wrong, because it uses that Micron chart for as a basis of information it was never intended to be used for. If the 3090 is the prelaunch name for the next Titan, then it isn't going to have 12GB of RAM, it will have 24GB and it's going to cost north of $2000, while only being slightly faster than the tier below. Most people will ignore this card, like they ignored the Titan RTX. If the 3090 is the prelaunch name of the new top Ti, which wouldn't make much sense, then who knows.
I laid out very concrete reasons for the pricing. To recap:

1) 3090 brand. We didn't have 2090, 1090, 990, or even 790. Bringing back a 90 model suggest Nvidia will increase pricing relative to the 2080 and 2080 Ti.

2) I think Micron absolutely did unintentionally leak the specs of GDDR6X and RTX 3090. Possibly Nvidia was supposed to have launched the card months ago, before COVID delays. Regardless:

a) 3090 will not have less VRAM than 2080 Ti.
b) 12GB in a 3090 still leaves space for a 24GB Titan RTX 3000
c) Could RTX 3090 just be a 'gaming' Titan with 24GB? Yes. Even more reason for it to cost $2,000+.

3) When was the last time Nvidia walked back pricing on a particular model? This one deserves some extra work, via a table. (Prices in parentheses are either the reduced price after higher end models came out, and/or the non-FE price.)

Kepler 1
(600-series)
Kepler 2
(700-series)
Maxwell
(900-series)
Pascal
(10-series)
Turing
(20-series)
*Ampere (est.)
(30-series)
*60$230 ($180)$250 ($220)$200$300 ($250)$350 ($300)$400
*70$400$400 ($330)$330$450 ($380)$600 ($500)$600-$700
*80$500$650 ($500)$550$700 ($550)$800 ($700)$1,000
*80 TiN/A$700$650$700$1,200 ($1,000)N/A (at launch)
*90$1,000N/AN/AN/AN/A$1,500-$2,000
TitanN/A$1,000$1,000$1,200$2,500$3,000?

Basically, Maxwell was the exception, and that was six years ago. Pascal was a generational increase in pricing at every level, and Turing took that to a whole new level. My best guess right now is that Nvidia will not release Ampere at lower generational pricing -- it's why the 3090 exists -- but it will stay somewhat close to the launch Turing prices.

A lot will also depend on how competitive AMD is with Navi 2x. If Navi 21 is 80 CUs as rumored, and clocks at 2GHz, that would end up being 20.5 TFLOPS of compute. That's ambitious but not out of the question. However, I'd also expect such a chip to be priced closer to $1,000 as well. Double the RX 5700 XT, double the price basically.

Traditionally, AMD gets less real-world performance per TFLOP relative to Nvidia, but Navi 2x may also improve the architecture to remove that deficit. Plenty of unknowns and assumptions are still in play, but that's where I sit in terms of what I expect right now. RX 5700 XT at launch was basically an RTX 2070 competitor, and AMD followed right along with Nvidia's new higher pricing tiers -- even without ray tracing support.

It would be surprising (and refreshing!) to see a true price war that dropped things back to 2015 levels. But it's 2020, not 2015, so I really don't expect that to happen. AMD will sell lots of PS5 and XBSX chips to Sony and MS at very low margins, but for PC we'll continue to see very high prices.
 
Reactions: Chung Leong

spongiemaster

Prominent
Dec 12, 2019
505
233
760
0
I laid out very concrete reasons for the pricing. To recap:

1) 3090 brand. We didn't have 2090, 1090, 990, or even 790. Bringing back a 90 model suggest Nvidia will increase pricing relative to the 2080 and 2080 Ti.

2) I think Micron absolutely did unintentionally leak the specs of GDDR6X and RTX 3090. Possibly Nvidia was supposed to have launched the card months ago, before COVID delays. Regardless:

a) 3090 will not have less VRAM than 2080 Ti.
b) 12GB in a 3090 still leaves space for a 24GB Titan RTX 3000
c) Could RTX 3090 just be a 'gaming' Titan with 24GB? Yes. Even more reason for it to cost $2,000+.

3) When was the last time Nvidia walked back pricing on a particular model? This one deserves some extra work, via a table. (Prices in parentheses are either the reduced price after higher end models came out, and/or the non-FE price.)

Kepler 1
(600-series)
Kepler 2
(700-series)
Maxwell
(900-series)
Pascal
(10-series)
Turing
(20-series)
*Ampere (est.)
(30-series)
*60$230 ($180)$250 ($220)$200$300 ($250)$350 ($300)$400
*70$400$400 ($330)$330$450 ($380)$600 ($500)$600-$700
*80$500$650 ($500)$550$700 ($550)$800 ($700)$1,000
*80 TiN/A$700$650$700$1,200 ($1,000)N/A (at launch)
*90$1,000N/AN/AN/AN/A$1,500-$2,000
TitanN/A$1,000$1,000$1,200$2,500$3,000?

Basically, Maxwell was the exception, and that was six years ago. Pascal was a generational increase in pricing at every level, and Turing took that to a whole new level. My best guess right now is that Nvidia will not release Ampere at lower generational pricing -- it's why the 3090 exists -- but it will stay somewhat close to the launch Turing prices.

A lot will also depend on how competitive AMD is with Navi 2x. If Navi 21 is 80 CUs as rumored, and clocks at 2GHz, that would end up being 20.5 TFLOPS of compute. That's ambitious but not out of the question. However, I'd also expect such a chip to be priced closer to $1,000 as well. Double the RX 5700 XT, double the price basically.

Traditionally, AMD gets less real-world performance per TFLOP relative to Nvidia, but Navi 2x may also improve the architecture to remove that deficit. Plenty of unknowns and assumptions are still in play, but that's where I sit in terms of what I expect right now. RX 5700 XT at launch was basically an RTX 2070 competitor, and AMD followed right along with Nvidia's new higher pricing tiers -- even without ray tracing support.

It would be surprising (and refreshing!) to see a true price war that dropped things back to 2015 levels. But it's 2020, not 2015, so I really don't expect that to happen. AMD will sell lots of PS5 and XBSX chips to Sony and MS at very low margins, but for PC we'll continue to see very high prices.
The names of the cards are of no interest to me. Price/performance/features are all that matter to me. So basically, you're predicting there will be 3 card announced on the 1st.

A $600-700 card.
A $1000 card.
A $1500-2000 card.

The rumor mill currently believes that 2 cards will be available in September. Presumably the top 2 cards which means, according to your prediction, nothing under $1000 MSRP will be available until sometime in October when a $600-$700 card is launched. Rumors also seem to think an x60 card could be launched as soon as November. If that doesn't happen, we're looking at no 30 series cards under $700 this year. That doesn't seem plausible unless NVidia knows that Big Navi is going to be another huge disappointment from AMD and Nvidia can just continue on naming whatever price they want.
 

Chung Leong

Upstanding
Dec 6, 2019
277
81
260
0
Kepler 1
(600-series)
Kepler 2
(700-series)
Maxwell
(900-series)
Pascal
(10-series)
Turing
(20-series)
*Ampere (est.)
(30-series)
*60$230 ($180)$250 ($220)$200$300 ($250)$350 ($300)$400
*70$400$400 ($330)$330$450 ($380)$600 ($500)$600-$700
*80$500$650 ($500)$550$700 ($550)$800 ($700)$1,000
*80 TiN/A$700$650$700$1,200 ($1,000)N/A (at launch)
*90$1,000N/AN/AN/AN/A$1,500-$2,000
TitanN/A$1,000$1,000$1,200$2,500$3,000?
One thousand is a huge psychological barrier. It's far too high for the RTX 3080 if it's not going to be the top-tier product anymore. My guess is that the 3080 will stay at $800 while the 3070 will jump to $700. The 3090 will be priced at $1600 so that the 3080 would seem like a great bargain. By spending an extra hundred dollars, you gain 10 meaningless units, while by foregoing 10 meaningless units, you save 50%. What a deal!

EDIT: Maybe not exactly $800. Something like $829--just to grab an extra 30 bucks.
 
Last edited:

JarredWaltonGPU

Senior GPU Editor
Editor
Feb 21, 2020
553
454
760
0
The names of the cards are of no interest to me. Price/performance/features are all that matter to me. So basically, you're predicting there will be 3 card announced on the 1st.

A $600-700 card.
A $1000 card.
A $1500-2000 card.

The rumor mill currently believes that 2 cards will be available in September. Presumably the top 2 cards which means, according to your prediction, nothing under $1000 MSRP will be available until sometime in October when a $600-$700 card is launched. Rumors also seem to think an x60 card could be launched as soon as November. If that doesn't happen, we're looking at no 30 series cards under $700 this year. That doesn't seem plausible unless Nvidia knows that Big Navi is going to be another huge disappointment from AMD and Nvidia can just continue on naming whatever price they want.
Rumors are just that: rumors. Guesses. This applies to prices as well as specs and release dates.

I do think Nvidia will start with two cards (3090 and 3080), but a third (3080 Ultimate, or maybe it's the 3090 Ultimate) isn't out of the question. Everything else will almost certainly be staggered. When was the last time Nvidia launched an x60 series part concurrently with the x70 or x80?

2060 came six months after 2080.
1060 6GB was four months after 2080 Ti, and 1060 3GB was one month later.
960 was four months after 970/980.
760 was one month after 770/780.
660 was five months after 680.
560 Ti was two months after 580.

So if the 3090 launches next month, it could be 3090 + 3080. 3070 would come one month later in October, 3060 might come as early as November. That's my best-case scenario for a 3060 showing up this year, and it's definitely possible. Problem is, AMD is building Zen 2 and Zen 3 on TSMC 7nm, AMD is also doing Navi 2x on 7nm, Nvidia is doing Ampere A100 and probably at least GA102 on TSMC as well. And potentially Intel is also tapping TSMC for some Xe HPG parts at some point.

So maybe Nvidia uses Samsung for the 3060 GPU (probably 3050 as well) and gets it out this year, but just as plausible is to save 3060 for early 2021. Nvidia doesn't want to flood the market with extreme GPUs and the provide a sensible alternative at the same time. That would potentially cannibalize sales of the extreme(ly expensive) models. So the earliest adopters only get 3090 and 3080 as options, then a month or two later the 'sensible' 3070 shows up, and then a month or two after that, the 'affordable' 3060 arrives for those who weren't willing to spend $500+. Maybe.

Nvidia could absolutely surprise everyone by doing 3060 earlier. But no leaks of any 3070 or 3060 images have surfaced anywhere I'm aware of, which means they probably aren't in production yet. Which means they're at least three months out from launching. Basically, either Nvidia hits Black Friday 2020 release, or it's going to hold off until January 2021 or later.

My hopes:
  1. No Founders Edition 'tax' -- just launch at the base price and let partner cards do whatever (like the 20 Super cards)
  2. No more than 1-2 months between releases
  3. Pricing similar to 20-series, so 3090 would take over for 2080 Ti
Of those three, only number two seems likely to happen. One and three are long shots at best.
 
Reactions: Chung Leong

ASK THE COMMUNITY

TRENDING THREADS