NVIDIA Ampere GPUs to feature 2x performance than the RTX 20 series, PCIe 4.0, up to 826 mm² dies and 20 GB of VRAM

Im just talking about the ti version, most likely wont be the full GA100 most likely a GA103 or 102. Whichever it is if it meets my 50% higher average performance criteria and if its below 1.5k usd then ill be alright with it. Im really hoping AMD will at least give nvidia a run for its money so that prices wont inflate that high.
30-40% higher than the 2080 perhaps, not the 2080TI, They are going to be adding more features and working on making things like RTX features work faster. A 50% increase over the 2080TI would be a super welcome surprise don't get me wrong but I don't think the dye shrink will be capable of doing that, I'm not holding my breath for any great increases until we start seeing some MCM packages hitting the consumer level and we are going to need another generation for that before its going to work.
 
dont matter what it cost if its more than 50% of a 2080ti then im in, lol.
Does anyone miss the days when the top gpu uesed the full xx0 dies and really where the top end GPU of a gen? Now we are paying top $$$$ for what in the past would have been an mid-range card. I don't see this getting better as gpus are now built mainly for other stuff with gaming and after thought.
 
Just give me a single card that will let me run near max settings in 3440x1440 glory. Cost be damned! :p
Seriously, I feel like my 2080Ti is a bit lacking at 3440 x 1440 at max setting, especially if I want it to be 120 fps.
 
Does anyone miss the days when the top gpu uesed the full xx0 dies and really where the top end GPU of a gen? Now we are paying top $$$$ for what in the past would have been an mid-range card. I don't see this getting better as gpus are now built mainly for other stuff with gaming and after thought.
its seriously the lack of competition. I bet you my left tit that if amd had a competitor for nvidias top end silicon we wouldnt be seeing these prices.
 
its seriously the lack of competition. I bet you my left tit that if amd had a competitor for nvidias top end silicon we wouldnt be seeing these prices.

When you are literally the only company in the world supplying a product with those specifications, you can charge whatever the hell they want, you aren’t their friend lol. They have 0 reason to actually put the R&D in and could even supply a substantially slower product at the same price if they wanted to really be assholes.

Jees anyone would think companies exist for some other reason other than to make money :ROFLMAO:
 
When you are literally the only company in the world supplying a product with those specifications, you can charge whatever the hell they want, you aren’t their friend lol. They have 0 reason to actually put the R&D in and could even supply a substantially slower product at the same price if they wanted to really be assholes.

Jees anyone would think companies exist for some other reason other than to make money :ROFLMAO:
Halt! Amd is benevolent and kind according to amdrones. They don't care about money.
 
Halt! Amd is benevolent and kind according to amdrones. They don't care about money.

I do love how everyone thinks AMD is some sort of friendly underdog that wouldn’t rape you and leave you in a ditch for 20 bucks :ROFLMAO::ROFLMAO::ROFLMAO:

Make no mistake people, if AMD has a competing product, they would charge just as much. If they were a market leader, they would charge more.
 
>826 mm^2 sized die on 7nm+
Do shitposters not know what defect rates are?
I love how they said they'd 'wait for navi' in order to make final tweaks, as if GPUs aren't designed years in advance. What a load of bullshit.
schiff bigger hoax lmao.jpg
 
I do love how everyone thinks AMD is some sort of friendly underdog that wouldn’t rape you and leave you in a ditch for 20 bucks :ROFLMAO::ROFLMAO::ROFLMAO:

Make no mistake people, if AMD has a competing product, they would charge just as much. If they were a market leader, they would charge more.

Given the black screen issues for RDNA cards, maybe AMD should charge more so that they can test better to resolve such issues !?
 
Given the black screen issues for RDNA cards, maybe AMD should charge more so that they can test better to resolve such issues !?

Funny how it has become an issue now :ROFLMAO: I remember this way back in the 4870x2 days and apparently it was my fault for buying their flagship “crossfire” card and expecting a mature, stable experience.

Had the same issues with the next generation then finally turned my back on AMD forever and never had a bad experience since with NVIDIA even using the out of box drivers 3yrs down the track.

It’s funny that everyone considered the Titan XP overpriced, still running it 4yrs down the track, still runs everything on max and still worth like $1000 in aud. Value for money!
 
Funny how it has become an issue now :ROFLMAO: I remember this way back in the 4870x2 days and apparently it was my fault for buying their flagship “crossfire” card and expecting a mature, stable experience.

Had the same issues with the next generation then finally turned my back on AMD forever and never had a bad experience since with NVIDIA even using the out of box drivers 3yrs down the track.

It’s funny that everyone considered the Titan XP overpriced, still running it 4yrs down the track, still runs everything on max and still worth like $1000 in aud. Value for money!
I would only consider multi GPU cards and multi GPU in general for benches, not gaming. I thought most everyone knew that. Multi GPU never thrived. If i bought a multi GPU card, it woudn't be for stability
 
That was a glorious time to be a computer enthusiast/hobbyist.

I had two 7900 GTOs SLI. It was a glorious time indeed. Watercooled them both, overclocked as hard as they would take. Then I soldered variable resistors to them to boost the voltage.

And then Left4dead came out and I discovered that those cards couldn't handle the new shadow features worth a shit.
 
NVIDIA will be launching its new GeForce RTX 3080 and GeForce RTX 3070 graphics cards later in the year, with my sources telling me the current coronavirus issues are going to halt the launch. Manufacturing, shipping, and everything in between is affected which means gamers will be waiting longer for new GPUs unfortunately.

via Tweak Town

https://www.tweaktown.com/news/7084...gb-hbm2e-at-2tb-sec-memory-bandwidth/amp.html
TweakTown has as good a track record as WCCF Tech. If you're looking for rumors then keep an eye on VideoCardz.
 
Coming soon, RTX 3080Ti. 100% faster. 100% more expensive. Pre-order your founders editions for $2,499.99.
 
Coming soon, RTX 3080Ti. 100% faster. 100% more expensive. Pre-order your founders editions for $2,499.99.

I am expecting an RTX 3070 at $500 with performance equivalent of RTX 2080 super

Because Xbox X is 12 teraflops & expected to debut at $500
 
  • Like
Reactions: Auer
like this
I am expecting an RTX 3070 at $500 with performance equivalent of RTX 2080 super

Because Xbox X is 12 teraflops & expected to debut at $500

that's not how that works. That's not how pricing works at all.

M$ is getting their GPU's for next to nothing since they're committed to purchasing 10's of millions of them.

Where an AIB will have 10-20k commitments depending on sales.

So, strictly fictitious numbers here. If M$ bought 10 million 2080 GPU dies lets say they paid $50 per. An AIB order of 10k GPU dies is going to be in the $200-300 range. The more you commit to buying the less per unit your cost is.

There isn't a $500 GPU sitting in an Xbox. There's a die mated to a proprietary board of M$'s design.
 
A bit off topic, but how many discrete GPUs per year does Nvidia & AIBs tend to sell these days?
 
Coming soon, RTX 3080Ti. 100% faster. 100% more expensive. Pre-order your founders editions for $2,499.99.

Careful. WiFi marketing people might hear you. 100 plus 100.....

RTX 3080Ti - 200% gaming!
 
that's not how that works. That's not how pricing works at all.

M$ is getting their GPU's for next to nothing since they're committed to purchasing 10's of millions of them.

Where an AIB will have 10-20k commitments depending on sales.

So, strictly fictitious numbers here. If M$ bought 10 million 2080 GPU dies lets say they paid $50 per. An AIB order of 10k GPU dies is going to be in the $200-300 range. The more you commit to buying the less per unit your cost is.

There isn't a $500 GPU sitting in an Xbox. There's a die mated to a proprietary board of M$'s design.

nVidia sells millions of gpus a quarter, 10k a negotiation is a little bit low lol.

IIRC the GTX 970 by itself sold about ~3 mill units/quarter.
 
nVidia sells millions of gpus a quarter, 10k a negotiation is a little bit low lol.

IIRC the GTX 970 by itself sold about ~3 mill units/quarter.

You're not looking at the big picture. Gigabyte, Asus, XFX, whoever, is not making million unit purchases of a given die every quarter. These AIB's don't sell a million 2080's let alone a million mid to high range cards combined, every quarter, each. Their bread and butter is low end OEM cards.

You think these AIB's are committing to purchase orders on the scale that M$ is?

Consoles out sell mid to high end GPU's 10 to 1. And last I checked M$ only sold 39.1 million One, One S and One X's combined.
 
So twice as fast, is it going to be twice as much? or the same price as current gen and we can expect the 20 series to take a massive dump in pricing?
 
People need to consider this is Volta vs Ampere.
Aka Tesla V100: https://www.nvidia.com/en-gb/data-center/tesla-v100/

So these are the relevant numbers (V100):
TensorFLOPS 130 TFLOPS
Integer Operations (INT8) 65 TOPS
Half Precision (FP16) 32.8 TFLOPS
Single Precision (FP32) 16.4 TFLOPS
Double Precision (FP64) 8.2 TFLOPS
On-die HBM2 Memory 32GB
Memory Bandwidth 1,134 GB/s

These are NOT gaming SKU's...
 
So, strictly fictitious numbers here. If M$ bought 10 million 2080 GPU dies lets say they paid $50 per. An AIB order of 10k GPU dies is going to be in the $200-300 range. The more you commit to buying the less per unit your cost is.

So uh what you're saying is that the more you buy the more you save!

 
I would only consider multi GPU cards and multi GPU in general for benches, not gaming. I thought most everyone knew that. Multi GPU never thrived. If i bought a multi GPU card, it woudn't be for stability

Sorry but no, you make a high end retail product, you don’t get to claim a consumer should not expect a perfect experience. Thats BS. See that is the difference between Intel / NVIDIA and AMD, AMD are happy to use their customers as beta testers regardless of how much they spend.
 
You're not looking at the big picture. Gigabyte, Asus, XFX, whoever, is not making million unit purchases of a given die every quarter. These AIB's don't sell a million 2080's let alone a million mid to high range cards combined, every quarter, each. Their bread and butter is low end OEM cards.

You think these AIB's are committing to purchase orders on the scale that M$ is?

Consoles out sell mid to high end GPU's 10 to 1. And last I checked M$ only sold 39.1 million One, One S and One X's combined.

You wildly understimate how many high end GPUs are sold. The 1080 was nVidia’s best selling Pascal according to their financial reports which they cannot lie on.

But nVidia has a 60% gross margin. Consoles are break even at best? I’d say that’s where the differences are...
 
So uh what you're saying is that the more you buy the more you save!


It's technically true. Bulk discounts are a thing. And in the context of the source for that video the machine NVIDIA was touting was a $150k device that would replace a $3m rack that is exponentially more power hungry.
 
It's technically true. Bulk discounts are a thing. And in the context of the source for that video the machine NVIDIA was touting was a $150k device that would replace a $3m rack that is exponentially more power hungry.

Haha I know I was just having some fun with that quote.
 
Back
Top