Nvidia GeForce RTX 4080 Reviews

[pjbliverpool]

The takeaway from all this (which we frankly already knew but nice to see it spelt out now in the reviews) is that this card is simply terrible value compared to the previous generation.

You are getting less price/performance than a 3080 despite this being a whole new generation that's launched two years later where we would traditionally (as in always without fail for the last 20-30 years) see much better price/performance ratio's. This is absolutely a backwards step.

I'm afraid I don't agree with the 3080Ti comparisons (let alone higher). The x080Ti cards have always been in a price and performance tier above the the x080 cards. They have also used the top end chip and been much closer (within a hand full of percent) to the top end halo product. The 4080's direct predecessor in every measure aside from price is the 3080, and here it offers terrible value. Add to that that the 3080Ti didn't even make sense as an Ampere generation card outside of the mining craze prices, and at least here in the UK you can pick it up new for at least £200 less than what the 4080 will be launched at. Unfortunately, I see that Ampere prices have started to rise again now in the UK as well, likely due to the total lack of competition from the new generation.

 

[Scott_Arm]

Remember that the msrp of the 30 series was basically pure fantasy until recently. I was lucky and I got my 3080 shortly after launch for close-ish to the msrp, but the prices kept going up and up after that. I think the 40 series is priced below what a lot of people payed for the 30 series. What I'm saying is, from Nvidia's perspective, the pricing on the 30 series was way too low.

 

[DegustatoR]

Remember that the msrp of the 30 series was basically pure fantasy until recently. I was lucky and I got my 3080 shortly after launch for close-ish to the msrp, but the prices kept going up and up after that. I think the 40 series is priced below what a lot of people payed for the 30 series. What I'm saying is, from Nvidia's perspective, the pricing on the 30 series was way too low.

Prices are also heavily region dependent.
My 3080 has cost me ~$1400 back in Dec'20 (never was cheaper than that up until recently really) and if I would be willing to get a 4090 now it would cost me ~$2000.
I'd expect 4080 when this will appear to cost somewhat close in USD to what 3080 retail price was back at its launch.

 

[pcchen]

One thing interesting is that it's starting to get difficult to find anything higher than a 3070 Ti (3080 etc.) at retail here. Even 4090 are difficult to find now.
It looks to me that NVIDIA waits till the inventories are almost out to release 4080 16GB.

 

[dobwal]

Remember that the msrp of the 30 series was basically pure fantasy until recently. I was lucky and I got my 3080 shortly after launch for close-ish to the msrp, but the prices kept going up and up after that. I think the 40 series is priced below what a lot of people payed for the 30 series. What I'm saying is, from Nvidia's perspective, the pricing on the 30 series was way too low.

Agreed. I don't think Nvidia is unaware of the actual prices your average PC gamer was paying for their 2000/3000 series wares. With the crypto craze gone, it allows Nvidia to push up the MSRP to mitigate the lack of cards being consumed by miners (and the revenue produced by those buyers) while not varying far from actual street prices of their previous products.

 

[trinibwoy]

The 4080 is a clear acknowledgement from Nvidia that the $700 MSRP of the 3080 was fantasy.

 

[Scott_Arm]

Nvidia could end up being wrong about pricing if the competition is able to undercut them and that devalues the proposition of the Nvidia product. But that assumes people want to switch brands and the competitions gpus are readily available.

 

[Silent_Buddha]

Using Techpowerup's review of the FE and their relative performance charts for GPUs. It's interesting.

VS. 3080

• 500 USD (71%) more compared to 3080 launch price and even worse for current pricing.
• 42% faster.

VS. 3090

• 300 USD (20%) cheaper than 3090 launch price but 300 USD (33%) more than current 3090 pricing.
• 25% faster.

So, it's horrible value compared to the 3080 and even current 3090 pricing, but at least compares favorably to 3090 launch pricing. It obviously compares more favorably to the Ti variants but then the Ti variants were already incredibly bad value products anyway.

If you're suddenly having to find value in a lower tier performance product compared to the highest tier previous product in order to somehow justify the product's pricing. Yuk.

Regards,
SB

 

[DegustatoR]

Using Techpowerup's review of the FE and their relative performance charts for GPUs.

These are highly misleading lately as they don't include any RT benchmarks in them.
Not that it would change much in this case but still.

 

[Silent_Buddha]

I haven't caught up on all of the reviews, but I've watched the Digital Foundry one. Good content, but Rich keeps saying (roughly) things like "value should increase the further you go down the gpu stack." Should it? If I were a company and I wanted to upsell, I'd have value increase in the opposite direction. I think the crypto market basically broke gpu pricing. It's the first time we really saw what people were willing to pay. Gamers were buying second hand gpus at scalper prices, because they wanted an upgrade and it's all they could find. Competitor pricing might put pressure on nvidia to reduce the price of the 4080, but that assumes the competitor products remain in stock and highly available, and people are willing to switch brands. Very curious to see how this plays out, but I wouldn't be shocked if Nvidia can sell every single 4080 they can produce. It does have a very large performance upgrade for people using 20 series, or even 3090s. I can't pay that much for a gpu, but there are clearly a lot of people that will, and a lot of Nvidia customers that upgrade every one to two gpu generations.

Consumer buying habits don't generally work that way. Usually a very small percentage of people are willing to pay unreasonably high prices for any perceived advantage. IE - paying double the price for a 1% better product. So, to maximize profits, you generally want to move as much product as possible at the lower tiers up to the point where consumers balk at the price. Then you sell something at a higher price tier until consumers balk at the price. repeat until you get to the top tier and you basically charge as much as you can that will move some product.

All of that, of course, governed by how much of X product you can produce. If you can only produce A quantiy of X product then you try to price it at P price that will, hopefully result in A amount of people buying all A quantity of that product. If you charge too much then you don't sell the product, if you charge too little then you've just left money on the table.

At each tier you will have some number of people that are willing to pay up to some price for something. The tricky part is trying to determine the price to sell each tier of product such that it sells just enough products that you move most of it while still allowing some product to remain in the retail channel (shelves or warehouses). You do that because it's better to lose a little bit of opportunity cost to ensure that a consumer can still buy the product if they want it versus buying a competitors product if there are none available when the consumer wants to buy it.

Regards,
SB

 

[Malo]

The 4080 is a clear acknowledgement from Nvidia that the $700 MSRP of the 3080 was fantasy.

That's your take on it. Or that the $700 would have been fine given normal product consumption instead of the mining craze combined with COVID spending/supply issues, and that Nvidia firmly believes it can get away with charging consumers whatever they damn want to now whilst also using the prices to clear excess Ampere inventory.

 

[techuse]

They would sell every GPU they could produce even if they priced this at 2000.

 

[DegustatoR]

They would sell every GPU they could produce even if they priced this at 2000.

I wouldn't be so sure about that. They clearly can't sell all the 30 series cards they've produced right now.

 

[techuse]

I wouldn't be so sure about that. They clearly can't sell all the 30 series cards they've produced right now.

They already sold through a ton of them and the whales have been waiting for Ada.

 

[DegustatoR]

They already sold through a ton of them and the whales have been waiting for Ada.

Sure but it just shows that your notion that they'll just sell anything they'll produce is wrong.
It was wrong back in Turing days and it's wrong now, doubly so as we're not only in "crypto hangover" but also in a worldwide recession.
I don't think that 4080 will be selling that well, and I don't even think that 4090 does that well right now from what I'm seeing in retail.

 

[techuse]

Sure but it just shows that your notion that they'll just sell anything they'll produce is wrong.
It was wrong back in Turing days and it's wrong now, doubly so as we're not only in "crypto hangover" but also in a worldwide recession.
I don't think that 4080 will be selling that well, and I don't even think that 4090 does that well right now from what I'm seeing in retail.

Well here in America 4090s are mostly out of stock and being sold for 2500+.

 

[DegustatoR]

Well here in America 4090s are mostly out of stock and being sold for 2500+.

America is rich ¯\_(ツ)_/¯

 

[davis.anthony]

I can't help but think it's quite bandwidth starved as even the 3090ti has a good chunk more bandwidth.

I will be interested to see performance scaling with only memory overclocking.

 

[pcchen]

I can't help but think it's quite bandwidth starved as even the 3090ti has a good chunk more bandwidth.

I will be interested to see performance scaling with only memory overclocking.

Actually it's one of the reasons why I decided not to wait for 4080, when I heard about its 256 bits memory interface.
4080 did use faster memory to mitigate the problem a bit, but that also potentially making it harder to overclock the memory.
On the other hand, 4090 actually have the same memory bandwidth as a 3090 Ti, but it performed quite a bit better. So it's entirely possible that either Ada is more bandwidth efficient or Ampere does not have enough computation power to use all these memory bandwidth.

 

[DegustatoR]

I can't help but think it's quite bandwidth starved as even the 3090ti has a good chunk more bandwidth.

It most definitely is as the scaling in comparison to 3090/Ti gets considerably worse in 4K against 1440p.
It would look a lot better as a $700-1000 product with a AD102 salvage part with 320 bus being at its price point.
But alas no such luck for now. Next year maybe?