• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

NVIDIA has revealed the prices for the RTX 5090, 5080 and 5070

To me the RTX 5070 Ti with 16GB memory and priced at $750 is a good deal if it performs the same as the RTX 4080.
 
The main boost seems to be in TOPS, NV really going all out on neural stuff, so games that make use of these things, smarter companions, rendering tech, texture/DLSS 4.0 stuff etc.

Double the RT perf of 4090 is also pretty huge considering how much of a beast that card is already. Gonna start to be feasible playing maxed settings RT native at 120 FPS in 1440p. 4K if you use DLSS.
 
In 2022 the 4070 was claimed to be as good as the 3090. In 2025 the 5070 is claimed to be as good as the 4090. If all of that is theoretically true, then there's literally no reason for anyone to be on the 3000 series cards anymore...because by that logic the theoretical 5060 should be about one pricing bracket above the 3090... and based on current pricing that's wallet accurate.

You then see a litany of asterixes. This includes DLSS, frame generation, and assumes that you don't get anywhere near the memory bandwidth...because of course not. If you don't really try to drive it fast, the Pinto has acceleration like the space shuttle...for about a fraction of a second. Consider me thoroughly unimpressed, and thoroughly hoping that the AMD and Battlemage offerings provide a solid 1440p offering at a price that kicks Nvidia in the shins. I like the 5700xt, the 3080, and the 3070. I don't buy a brand, I buy value for performance at a price I can afford...and right now that isn't Nvidia. Lord knows, their pricing is gouging for the AI crowd, not supporting a core market that they seem to think they don't have to care about anymore.
 
>The high-end model will be the NVIDIA RTX 5090. This GPU will have a $1999 MSRP. Then we have the RTX 5080 which will be priced at $999. Finally, we have the RTX 5070Ti and the RTX 5070 which will have a $749 and $549 MSRP, respectively.

>NVIDIA stated that the RTX 5070 will be as powerful as the RTX 4090.

hahaha. they always do this. like they said the 3070 beats a 2080ti. but 2080ti beat it mostly and in 4k 2080ti leads no issues and 11gb.

nvidia with their slective lies.

this is great. cmon low priced 3090!!
 
>The high-end model will be the NVIDIA RTX 5090. This GPU will have a $1999 MSRP. Then we have the RTX 5080 which will be priced at $999. Finally, we have the RTX 5070Ti and the RTX 5070 which will have a $749 and $549 MSRP, respectively.

>NVIDIA stated that the RTX 5070 will be as powerful as the RTX 4090.

hahaha. they always do this. like they said the 3070 beats a 2080ti. but 2080ti beat it mostly and in 4k 2080ti leads no issues and 11gb.

nvidia with their slective lies.

this is great. cmon low priced 3090!!
RTX 5070 will be as powerful as the RTX 4090 - with a catch , when using the new Frame Gen ( DLSS 4 with Multi-Frame Generation ).
 
the 5080 is a good price, the 4080 sholud have been that price. 5080 has gddr7, so if the memory bus width is the same or wider the 5080 will be a better card than the 4080… and cheaper at release. yet, i do not need another GPU, i have the 7900xtx (good enough, in light of games targeting the ps5pro, nvidia’s DLSS making games run on a 4060, look like they are on a 4080… or they say)
 
RTX 5070 will be as powerful as the RTX 4090 - with a catch , when using the new Frame Gen ( DLSS 4 with Multi-Frame Generation ).
You could argue that technically that doesn't make it as 'powerful'....

If a 2 seater small car can go as fast as a crew-cab pick-up truck, it doesn't make one as 'powerful' as the other.
Using DLSS is like stripping out the seats, un-needed interior bits, etc. - at which point it's a 'fiddle', so if that's the only way to get the 2 seater as fast then ultimately the statement is false.

Of course, its up to the marketing team to make sure people don't see it that way.
 
You could argue that technically that doesn't make it as 'powerful'....
That was my point :) who knows maybe AMD will have a Multi-Frame Generation that works cross brands or some hack to enable it on 40 series .. then for sure it wont be as powerful , and you cant really say that is as powerful when you are using trickery like Frame Gen, they are fake frames.. whatever.
 

So ngreedia still has not learned, ok hard pass... The stack should have been 1200/750/500/350, but they will learn the hard way when Im hoping user will en-mass see sense and not pay these prices. I also sincerly hope these cards will flop from a perf pov, as in will only marginaly faster than previous gen as a result of higher power draw. Unless I can finagle a deal like like I did for my 4080 where I actually only paid £300 for it, then I will pause on this cycle.
IT's actually not as bad as you make it sound. With the current inflation rate, these prices are lower than the last generation. I'm very happy that the 5080 is only $1000, it's not a bad deal at all. If only it had 24Gb of RAM :(
 
The main boost seems to be in TOPS, NV really going all out on neural stuff, so games that make use of these things, smarter companions, rendering tech, texture/DLSS 4.0 stuff etc.

Double the RT perf of 4090 is also pretty huge considering how much of a beast that card is already. Gonna start to be feasible playing maxed settings RT native at 120 FPS in 1440p. 4K if you use DLSS.
Their own charts don't support the claim of double the RT performance. In any case, we only have to wait three weeks for reviews.

1736266127056.png
 
It was... in 2016.

Accounting for inflation that'd be $1577/985/657/460. We'll see where actual retail prices stabilize at. If the 5080 holds there it'll be good.
Between TAA and RT i feel like graphics have regressed a bit. Games from 2016, to me look as good if not better than some games today, and it's a little weird.

I kinda blame it the Unreal Engine, and art direction. All games look the same to me and you can only get so "realistic" before everything looks even more the same.
 
But it looks like raster performance is only up 20-30% up
Not necessarily bad but I want to see how that stacks up with regards to energy consumption. Probably needs to throw in consideration the number of CUDA cores and such to see what's the actual improvement. Waiting for reviews as usual.

Consider me thoroughly unimpressed, and thoroughly hoping that the AMD and Battlemage offerings provide a solid 1440p offering at a price that kicks Nvidia in the shins.
Not sure about Intel, they still got a ton of work to do, so even if the hardware is fairly decent, the driver kinda sucks currently, and seems like it will keep sucking for some time still.

And AMD is up in the air until pricing is revealed but I doubt the 9070 will be priced anything that's actually good enough to stand on its own. The price difference would have to be such that it really makes people go for an AMD card and... I don't see that happening, even if the price is actually fairly low, because mindshare and all that.

Their own charts don't support the claim of double the RT performance.
To be honest, those charts are shit design to begin with :laugh:
 
Think ill skip the 50 series this time since the tax gonna be very steep.
 
Last edited:
Seems like good pricing on that 5070...
 

So ngreedia still has not learned, ok hard pass... The stack should have been 1200/750/500/350, but they will learn the hard way when Im hoping user will en-mass see sense and not pay these prices. I also sincerly hope these cards will flop from a perf pov, as in will only marginaly faster than previous gen as a result of higher power draw. Unless I can finagle a deal like like I did for my 4080 where I actually only paid £300 for it, then I will pause on this cycle.
Outside of the 5090, the prices are the same or better than last gen.

$1000 5080 is the 4080 Super price. $200 cheaper than 4080 at launch.
$749 5070 Ti is $50 cheaper than the 4070 Ti Super.
$549 5070 is same price as RTX 4070 and $50 cheaper than the 4070 at launch.

And performance seems to be ~20-33% improvement (depending on tier) gen/gen.

Just a few days ago someone here was going around saying the 5090 would be $2500, 5080 would be $1500 and 5070 would be $700 :roll:

5090 had a price increase, but its also 2x the specs of the 5080 and for me that's an unobtanium GPU anyway with the 575W TDP, which is almost as much as my PSU alone.

50-series should do at least as good as the 40-series (which is the series that got Nvidia to 90% market share).
 
Last edited:
NGL, considering theyll have no competition on high-end market, these are better prices than I expected. But it looks like raster performance is only up 20-30% up, so it's very 'meh' generation.

However, seeing Nvidia comparing fake frames to even more fake frames and based on that claiming 5070 = 4090 left a really bad taste in my mouth (again, but even worse than what they have done with 4000 series where they at least put raster results in the slides too). This is plain deception.
yea i think about 30% faster on all cards compare to last gen of cards but still thinking of getting a rtx 5070ti which is a upgrade from my rx7900xt in rt atleast
 
Scalpers going to sell 5090 for around $4-5K and fools are going to buy it in a heart beat :kookoo:

People can just hire GPU time on H100, H200, A100, L40S for LLM / ML development for not much. Not going to make sense to use an inferior 5090 for most.
 
Their own charts don't support the claim of double the RT performance. In any case, we only have to wait three weeks for reviews.

View attachment 378800
Yeah this looks like a 25% delta performance gain for the 5090 vs 4090 outside the smoke and mirrors dlss. Unless this is with the small factor fe design that is throttling it's potential performance. I am personally looking forward to the 360 mm all in one hybrid design. So far their are 3 8n the pipeline.
 
Could be better, could be worse.

Halo got halo-ier while everything else got a slight price/performance/MSRP improvement.

Wish NV would just give us the raw performance numbers, but marketing going to market. I can only hope there are guys out there unloading their 4090's for $600 because "The 5070 is gonna be just as fast for $550".
 
looking at 5070/5070ti only problem is new/old administration and pricing.
 
People can just hire GPU time on H100, H200, A100, L40S for LLM / ML development for not much. Not going to make sense to use an inferior 5090 for most.
Yeah, hire time just once per month and you might save money. but if I'm going to run it frequently, it'll be cheaper to own the card.
 
Back
Top