Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
The 2070 Ti (at least the super) has 13.5bn transistors.Also since some people love Moore's law and base their entire argument on it.
![]()
Moore's law - Wikipedia
en.wikipedia.org
The 3070 Ti has 17.4bn transistors.
![]()
NVIDIA GeForce RTX 3070 Ti Specs
NVIDIA GA104, 1770 MHz, 6144 Cores, 192 TMUs, 96 ROPs, 8192 MB GDDR6X, 1188 MHz, 256 bitwww.techpowerup.com
The 4070 Ti has 35.8bn transistors.
![]()
NVIDIA GeForce RTX 4070 Ti Specs
NVIDIA AD104, 2610 MHz, 7680 Cores, 240 TMUs, 80 ROPs, 12288 MB GDDR6X, 1313 MHz, 192 bitwww.techpowerup.com
Not the first time, that's how the last 50 years of IC manufacturing has generally worked and what that Wikipedia article you linked to describes.Point is nvidia have more than doubled the transistors for a specific GPU class.
I suggest you read what the Wikipedia article describes because it ain't "a much better" node every two years, it's not even close.Nvidia simply used a much better 4nm node than the last generation and even what AMD is using now. They have actually satisfied Moore's law.
And it does but only in relation to reducing the reticle size so you can fit more into the same area.Some people here who think die area matters, sound like they would have been happier if nvidia used a cheaper/worse node.
They didn't intended it to be that though, so, yea...If anything this shows the 70 Ti naming was actually appropriate.
Ain't my comparison, you're the one who posted the Wikipedia article and a couple of links to the TPU GPU database.Also in your comparison, you want to focus on the fully unlocked TU104 chip which is the 2080 super from memory. That was 13.6bn transistors as well. The 2070 super wouldn't have been using a lot of the actual die.
![]()
NVIDIA GeForce RTX 2080 SUPER Specs
NVIDIA TU104, 1815 MHz, 3072 Cores, 192 TMUs, 64 ROPs, 8192 MB GDDR6, 1937 MHz, 256 bitwww.techpowerup.com
![]()
Sapphire Radeon RX 7900 XTX Nitro+ Vapor-X 24GB GDDR6 PCI-Express Graphics Card
Order Sapphire Radeon RX 7900 XTX Nitro+ Vapor-X 24GB GDDR6 PCI-Express Graphics Card now online and benefit from fast delivery.www.overclockers.co.uk
£100 off just like that, £1300 radeon not a hot seller, who knew.........
![]()
Sapphire Radeon RX 7900 XTX Nitro+ Vapor-X 24GB GDDR6 PCI-Express Graphics Card
Order Sapphire Radeon RX 7900 XTX Nitro+ Vapor-X 24GB GDDR6 PCI-Express Graphics Card now online and benefit from fast delivery.www.overclockers.co.uk
£100 off just like that, £1300 radeon not a hot seller, who knew.........
It is actually a great seller, but 600 units have dropped in today and to boot booked at 1.24 rate so doing a weekend special to blast a couple hundred out with any luck. We have sold 261 units in the last 10 days on this card, so I'd say its a pretty hot seller.
And who would know it is my job to actually try and keep the products in stock at all times, in a perfect world nothing would ever be out of stock on the website.![]()
Raven thinks he knows how well every card sells by how many show in stock though. Not accounting for an kind of automatic replenishment you have set up etc or other systems,.
Don't take my word for it, there is a recent report showing lowest GPU sales for 20 years. If nvidia can't shift £1199 4080s, you think AMD are shifting that many XTXs. Only so many radeon fans with that kind of dosh.
Price vs. performance is a regression on last gen. Take that into account with the cost to manufacture and you have yourself a joke of a product.The question is, does anyone really care or should care about the specs of the card? Lets say the 5090 is 3 times as fast as the 4090 but it uses the xx6 die. Would you actually care? I wouldn't, for sure. I buy performance, not specs.
Transistor count is a shocking metric for performance. Go work out the transistor count increase for the 4090 Vs the 3090, and the performance uplift. It's massively sublinear.I find it hilarious people focus on die area. If anything it should be transistors you count. It shows a real misunderstanding of how you measure improvements in semiconductors when people focus on the overall size than the thing that actually makes something more complex.
Otherwise you are basically saying if nvidia used a worse node, like AMD did then it would have been okay for them to charge more.
I suggest you go away and look at the transistor counts. How many in the 3090 Ti and how many in the 4070 Ti? Since performance doesn't matter apparently, the 4070 Ti blows the 3090 Ti away in transistors![]()
Transistor count is a shocking metric for performance. Go work out the transistor count increase for the 4090 Vs the 3090, and the performance uplift. It's massively sublinear.
Meanwhile, die size is a better indicator of cost and yields (because you buy wafers, not transistors), so you can more easily infer "value" from a cost to manufacture perspective.
Obviously price vs. performance is king, but when faced with objectively poor price vs. performance, you should ask "is this expensive because it is pricey to make, demand is high or are margins the primary goal of a product". The 4080 and 4070ti are obviously margin focused, which is a shame because NV make brilliant products and have historically prioritised product.
youve got to remember gibbos a salesman. hes not going to say cards arent selling well. just google nvidias sales for last three months. lol. says it all.Raven thinks he knows how well every card sells by how many show in stock though. Not accounting for an kind of automatic replenishment you have set up etc or other systems,.
Well ADA is progress its just that the progress starts at £1200.If they can't make our money perform meaningfully better, it's not meaningful progress.
It's just stagnation. The 3080 I bought for $700 CRUSHED the 2080 that launched at $800. That's what progress looks like.
I didn't buy the 2080 though because it didn't provide a meaningful uplift over my 1080Ti.
$700 Pascal was progress.
$800 Turing wasn't
$700 Ampere was progress.
$800 Ada isn't.
youve got to remember gibbos a salesman. hes not going to say cards arent selling well. just google nvidias sales for last three months. lol. says it all.
42% decline in gpu sales in last year alone. yeah they flying out !amd down to 10 percent of the gpu market as well. why you think they cancelled buying gpu chips .