Nvidia GeForce RTX 5080 Review

Two years for 11 percent is pretty devastating but hardly unexpected. It's the same TSMC process. The improvements for the same TDP always be marginal. I expect a short lifespan for these cards.

This is just the effect of Nvidia skimping on silicon to keep their profits high.
 
Oh boy, this is much worse than I expected.

I was thinking there would be at least a 10-20% uplift, but this is 5% or less from their prior card.

Terrible!
 
AMD prepared to lunch RX9070 soon but after they knowing that RTX5080/5070 wasnt faster by that much they delayed it (postpone) the launch to march
AMD expected RTX5070 to be at least equal to RTX4080/SU. Since that will not happen they will change the price to same Nvidia performance. This is why they declared 300-1000 USD price at CES.
 
As I've noted for some time now, we're basically at the end of being able to gain performance through die shrinks alone. If you don't see significant architecture enhancements, you really shouldn't be expecting significant performance improvements anymore.

Put another way: we're starting to hit a performance wall.
 
Everyone was waiting for the smoking gun, instead we got the smoking pile of horse ***t from Nvidia. Fanboys rejoice! Your "MULTI-FRAME GENERATOR" has arrived xD
 
Time tells a tale...

I paid $1,170 for my RX 7900 XTX Merc 310 on day-1 and immediately started comparing it to my ARMA3 clanmates and in other games (COD/SC) and we went full dERP!

Mind you, at the time, 4080 was going for $1,600+ & 4090 was $2,600~ $3k. One of my clanmates who stood in line and got his 4090 at retail, sold his card for nearly $3k and bought a XTX. We knew.



That was 2 years ago^
This March, AMD plans on releasing a "mid-grade" Gaming card that has the gaming performance of the 4080 (ie 7900GRE). So slightly under the 7900XTX performance...

Rough estimates, is that the new RX 9070xt is 15% slower than the RTX5080, for about $599'ish. With a faster version (RX 7080xt?) using faster memory coming this summer. (And high-end professional chiplet gaming cards at the end of the year.)


CUDA is great if you use a Workstation, or are a Content Creator, etc.. and heavily use CUDA. But if you are buying a GPU for Gaming, than looking forward Radeon is the new wave..!





 
Retail are reporting single digit numbers of cards being provided. NVidia don't give a monkeys about gaming GPU's any more, hence this awful release. Over 2 years after their last release and they can't even manage a die-shrink.
Not only are they bad value at $1000 they are awful value at $1400 for rumoured 3rd party card prices and abysmal value at $1800-$2000 for scalper prices (the only way you would be able to get hold of one if you are stupid enough to try, as NVidia are only making about 10 a year).
As for the 5070 still only have 12GB of RAM....it won't even run some games like Alan Wake 2 at it's launch as it's not powerful enough, yet it's going to be at least $600 realistically to get one as the FE's will be AWOL at their RRP.
The 50 series is awful and should be getting much worse reviews than they are. A complete waste of silicon and will cost more per frame than their 40 series counterparts in the real world. 3rd party cards will be the only one people can buy and they are all way more expensive (which NVidia totally knows will happen). The FE's are just a fake launch to sweeten the pill of a sh1t card at a sh1t price.
 
Last edited:
I thought I would be looking on in annoyance as a 7900xtx owner and thinking I would have to figure out an upgrade much faster than expected, but clearly not if the competition in Nvidia can barely get an uplift on a card I was on par with for rasterization, and with less vram, I thought the 5090 looked rough because they caned the power consumption to make sure they got the big numbers, but this 5080 is looking so rough in comparison, clearly AMD haven't left high end after all, all because Nvidia has got lazy
 
As for the 5070 still only have 12GB of RAM....it won't even run some games like Alan Wake 2 at it's launch as it's not powerful enough, yet it's going to be at least $600 realistically to get one as the FE's will be AWOL at their RRP.
 
As the owner of a normal 4080, what was the point of the 4080 super?
- Nvidia doesn't do price cuts, they never devalue their existing products.

Their strategy is to release a new card and launch it for the modified price (Supers ever since Turing).

That way they never have to take a haircut on existing stock with retailers or their AIBs. They also build brand loyalty and brand perception.

4080 wasn't hugely popular at $1200 and NV didn't want to drop the price, so they launch a 4080S with almost the exact same performance but an official MSRP of $1000.
 
I thought I would be looking on in annoyance as a 7900xtx owner and thinking I would have to figure out an upgrade much faster than expected, but clearly not if the competition in Nvidia can barely get an uplift on a card I was on par with for rasterization, and with less vram, I thought the 5090 looked rough because they caned the power consumption to make sure they got the big numbers, but this 5080 is looking so rough in comparison, clearly AMD haven't left high end after all, all because Nvidia has got lazy
I don't think they got lazy, I think they saw that AMD couldn't do better and figured they could basically make money on what they already had. No push to progress really because there isn't much competition. I guess that could be lazy in a way but it's really just a way to make money with less effort.

You know the sad part is that people are lining up for this crap. Gamers just feed right into it...."take my money I have to have the one with different numbers on it even though it's basically been available for two years".
 
What a slap in the face. Now that NV is the current leader for A.I. they have just dumped their gaming customers.

Almost seems a "reluctant," release like they can't be bothered. But that would look bad so release similar to the 4080 super, call it a new gen, and make, I imagine huge profit margins.

As the review allude one reason for this pathetic "new," gen card could be positioning to release a
RTX 5080-Super, or RTX 5080 TI (Is super & TI basically the same thing?) Just before the next holiday season - 9 to 11 months later.

If that happens they will have plenty of components by then, time to tweak and possibly release a very decent card, somewhat slower than the RTX 5090.

Production costs should be lower by then, BUT: if this happens how will they price it?

TBH though, this "first," 5080 is so appalling I doubt they have such inclination.

I can't help but wonder if NV want to get out of the gaming market, except for the very highest end.
But then the RTX 5090 was not as good as people hoped, and the pricing is insane.

I don't know really. Just a huge let down.
 
Has anyone directly compared the 5080 vs the 4080 using the transformer model with no frame generation? Supposedly the TM is more computationally expensive and the 5080's TOPS performance should be more than double that of the 4080 (according to Nvidia). If that is able to help it reduce the overhead of DLSS4 TM, which is greater than that the CCN model, then perhaps that is something since most new titles will support this new version of DLSS? Trying to help the 5080 out here a bit, even though we're talking Nvidia proprietary tech. Or are the increased TOPS only useful for MFG?
 
The 5080 situation is best summarized by Arstechnica’s final conclusion.

“ The best argument for grabbing a 5080 right now, if you can find one at $999 in the first place, is that the only cards that come close to its performance are either way more expensive than they're supposed to be (the 4080 Super) or not cheap enough to justify the Nvidia-specific features you give up (AMD's Radeon 7900 XTX).”

For gamers it is a terrible place to be.
 
I intended to take the plunge by parting with my 3080 for a 5080. I want to play comfortably in 4K but after reading the Tech files on this new generation, there is reason to be disappointed with the lack of additional raw performance.

Having skipped the 4XXX generation, I was eager to discover the next one. I guess I'll wait another year for a 5080 Super.

Does the next generation of GPUs (6XXX) seem so far away to me with a DLSS 5 or 6? More interspersed images, unreasonable latencies that only 360 Hz screens will be able to display cleanly without artifacts?
 
I'll wait for the 5070 vs 9080 fight, might be interesting.

This review result is so disappointing, I had planned to build a new PC going from 2070 Super to 5080, but now I really am reconsidering. I know it'll be a big leap from where I'm currently from, but the concept behind my new build is to last 7-10 years (ideally 10!), I'm not the sort of person who regularly has disposable income for this....
 
Has anyone directly compared the 5080 vs the 4080 using the transformer model with no frame generation? Supposedly the TM is more computationally expensive and the 5080's TOPS performance should be more than double that of the 4080 (according to Nvidia). If that is able to help it reduce the overhead of DLSS4 TM, which is greater than that the CCN model, then perhaps that is something since most new titles will support this new version of DLSS? Trying to help the 5080 out here a bit, even though we're talking Nvidia proprietary tech. Or are the increased TOPS only useful for MFG?
Not that I've seen. I've even asked a few reviewers about the lack of improvement in RT given the supposed double RT compute with the 5000 series but none have responded to my comments. Not sure if that is because they are working on for an upcoming review or don't know.
 
This generation has got to be the most spoiled group of pampered brats this Country has ever seen. So the new GPU's are more of a refresh with some added features. OMG! thats only happened like, dozens of times in the past 25-30 years. Its all about the pricing, there is no bad GPU.
 
Back