5070 6400 Cuda ~~ $640
5080 10752 ~~ $1200
5090 21760 ~~ $2200

5080 Ti 28 GB in Q3 2025.

Dang! looks like i need to wait a little bit longer for 24gb variant if its true.:banghead:

I'm good. I'll stick to what I have. I have a huge backlog still and the occasional new game should be no problem as I have a lot of spare performance room still at 1440p.

Legacy-ZAGimped right out the gate

"Gimpy is as gimpy does" hahahaha

Space Lynxthey have to milk each release as much as possible

Ah yes, fresh, ice cold & creamy milk, yea, the nGreediya way ! Can we get some cookies with that ?

(fyi, I prefer chocolate chip, or even peanut butter !)

Space Lynxassuming AI wasn't a thing

Well, the thing is AI, and AI is the thing, which is and is not you, so your assumption(s) do not compute & therefore are completely invalid :D

12GB on a card which is likely to cost significantly over 600$... Oh Nvidia...

N/A5070 6400 Cuda ~~ $640
5080 10752 ~~ $1200
5090 21760 ~~ $2200

5080 Ti 28 GB in Q3 2025.

Lol no. The 5090 is going to double the price of the 5080 at the very least as no AMD card will come close to its performance. $2500 minimum.

Nostras12GB on a card which is likely to cost significantly over 600$... Oh Nvidia...


Lol no. The 5090 is going to double the price of the 5080 at the very least as no AMD card will come close to its performance. $2500 minimum.

if I was Jensen, I would get rid of the 5090 naming scheme, bring back the Titan name, make the 5080 $1500, and the new Titan $2400. i don't think Nvidia spends much time thinking about gamers anymore though, so you are prob right, they won't even try any tricks, just make it 2500 min for the 5090, even if its 1999, after tax its knocking on the door of 2500 depending where you live

Space Lynxif I was Jensen, I would get rid of the 5090 naming scheme, bring back the Titan name, make the 5080 $1500, and the new Titan $2400. i don't think Nvidia spends much time thinking about gamers anymore though, so you are prob right, they won't even try any tricks, just make it 2500 min for the 5090, even if its 1999, after tax its knocking on the door of 2500 depending where you live

Nah they won't. Most normies know shit about the Titan lineup but sure have heard of the 90 cards. There's a good reason why Nvidia went this route.
Nvidia, then bigger number better. In that order. Titan has no numbers last time I checked.

What 0 competition looks like when a super cut down 5080 has nothing to compete with this is what we get..... Should be interesting how it actually performs but the gap has only widened..... 3080 to 3090 12-15%, 4080 to 4090 25-30%.... Now is it going to be 40-60% seemingly.

Also AI is infinitely more lucrative right now than the pleb gaming market yet Nvidia still dominates in the gaming gpu market because there is no real competition regardless of the crumbs they send our way.

Space Lynxif I was Jensen, I would get rid of the 5090 naming scheme, bring back the Titan name, make the 5080 $1500, and the new Titan $2400. i don't think Nvidia spends much time thinking about gamers anymore though, so you are prob right, they won't even try any tricks, just make it 2500 min for the 5090, even if its 1999, after tax its knocking on the door of 2500 depending where you live

To me it seems the 90 class has been much more successful than the Titan Class so I doubt they will give up that naming, could I see them turning what should be an 80 class gpu and naming it XX90 whatever sure though. Basically with what they did last generation shifting everything up a tier in naming except arguable the 4090....

Marcus L$700 is the new "budget" mid-range, seriously I wish NV would FO, can't wait until the AI shit bubble bursts as it's only going to continue getting worse for the average gaming consumer, honestly the 1 company I would be happy with going bust, their arrogance knows no bounds

I very much doubt that AI is going to slow down or away, we're just scratching the surface what it can actually do, and even now result are amazing.
Imho it will last longer than cryptmining craze, or simply we will just adjust to it being always around same like smartphones never went away or internet.

Unless Skynet happens, but that's sci-fi, right?! ;) :D

kawiceI very much doubt that AI is going to slow down or away, we're just scratching the surface what it can actually do, and even now result are amazing.
Imho it will last longer than cryptmining craze, or simply we will just adjust to it being always around same like smartphones never went away or internet.

Unless Skynet happens, but that's sci-fi, right?! ;) :D

it's not even AI, it is BS that NV and lots of other companies are milking and making billions from, it is however not AI in the slightest or anywhere close to, it's a fuckin ponzi scheme, another dotcom bubble bollocks and it will bust, taking a lot of data and having the means to look at it and interpret so that it looks like intelligence it is not artificial intelligence, it's computing, it's faster computing than how we have been doing it up until now, but there's no intelligence involved, you can program it just the same as any other type of computer models to come up with any result you want it to, that's not artificial intelligence, it's nvidia/others realising they can do tasks on GPU's/LLM/"AI" thousands of times faster than you can on traditional CPU and telling you it's AI, we are so far away from true AI and I highly doubt it will even come to pass in our generations, the industry gets hyped up over some new technological BS every few years that is going to change everything and doesn't, just gets a select few very very rich and the rest left in the dirt, rinse and repeat

Pathetic amounts or ram, 5070 should be at 24 by now

kondaminPathetic amounts or ram, 5070 should be at 24 by now

There is only 16 Gbit density in production. And I hope the non binary 24 Gbit never gets any traction because I don't want a 192 bit bus on a 70 class either and let them get away with 18 GB.

My 2080 Ti never gets to use more than 6 GB any way, performance is too low. So my expectation is that 4080 cant possibly need more than 12 even with the DLSSFG shenanigans of 500% boost.

oxrufiioxoWhat 0 competition looks like when a super cut down 5080 has nothing to compete with this is what we get..... Should be interesting how it actually performs but the gap has only widened..... 3080 to 3090 12-15%, 4080 to 4090 25-30%.... Now is it going to be 40-60% seemingly.

We are approaching the ideal of the golden ratio, where each next tier is 60% faster. It's actually not bad.

I don't care if the 5070 is good, launching it with 12GB is an utter disgrace at this segment of the market. What a load of crap. Delay it until 3GB does are ready and increase the cuda cores 20%.

As for the 5080 I'll wait for 24GB variant undoubtedly coming as the 5080 Super (Ti). With AI upscaling on 61MP images it's easy to fully utilise more than 16GB. My 6800XT is maxxed out and my old 2080 Super can't even handle the 2x upscale, as it lacks enough memory.

Marcus L$700 is the new "budget" mid-range, seriously I wish NV would FO, can't wait until the AI shit bubble bursts as it's only going to continue getting worse for the average gaming consumer, honestly the 1 company I would be happy with going bust, their arrogance knows no bounds

Until AMD releases 8800XT at $550-600 with 7900XT-7900XTX beating performance at much lower power. 8800XT is ready to go, AMD just needs to sell off huge stocks of 7900's that no one wants.

Daven5070 might be more like the 4070 Ti Super


The problem is always the performance gap. The difference between the 4080 and 4090 is around 30%. If the 5080 cannot match the 4090 in performance then it will be less than 30% faster than the model it is replacing. That is too small of a difference and acting more like CPU generational differences.

Battlemage is targeting 4070 ti in gaming; 4080 in compute. I am not buying an Nvidia card for gaming or productivity, when both Intel & Amd can outperform them at every price point they compete at.

Looks like the only ones I'll be getting going forward are some of those interesting ones I have seen and purchased off of Aliexpress.

(I swear, I can stop buying Aliexpress cards at any time, honest.)

Minus InfinityI don't care if the 5070 is good, launching it with 12GB is an utter disgrace at this segment of the market. What a load of crap. Delay it until 3GB does are ready and increase the cuda cores 20%.

As for the 5080 I'll wait for 24GB variant undoubtedly coming as the 5080 Super (Ti). With AI upscaling on 61MP images it's easy to fully utilise more than 16GB. My 6800XT is maxxed out and my old 2080 Super can't even handle the 2x upscale, as it lacks enough memory.


Until AMD releases 8800XT at $550-600 with 7900XT-7900XTX beating performance at much lower power. 8800XT is ready to go, AMD just needs to sell off huge stocks of 7900's that no one wants.

RTX 2080ti with 22gb of ram can be had for $484, shipped to your door. I am debating whether to get one of these for Stable Diffusion.

Don't forget Nvidia NEVER limits the present, only the future. The fact at hand here is that they can quite deliberately consider as "future" the next gen, and that it's therefore perpetually ping pong with nVidia on each side of the table and never missing the rebound. That's the place they carved for themselves.

So what is it? You can be sure they'll have leveled a performance increase just enough. So the 5070 will be slightly better than the 4070S *at something (take your pick: RT, AI, and/or a slight generation increase)* and replace its pricepoint (+50$) or something like that. The 4070S will see a price drop by mid 2025 to 4070 pricepoint, which will take most of 6 months before it even starts to show when you're regularly browsing prices, by which time you'll be a couple months away from the next gen. The 4070 will be evacuated OR will become the "5060Ti" as is or something along those lines depending on how nVidia needs to target a pricepoint with "old" performance. This leaves ample space to make a 5070S or a 6070, as they'll see best, with 16GB vRAM. Perhaps it will even become a ping-ping-pong just like Intel once upon a time went from tick-tock to tick-tick-tock,

By then nVidia will have calculated the market shares of each Blackwell line and will have come up, as needed, with a 5070Ti, a 5080Ti, a real 5060Ti, or will wait for the next gen to just upgrade each the 70 and 80 as a Super/Ti edition. If anything I would figure either the "Ti" editions for any card might be eventually evacuated, or that they'll launch "Ti"s a little early on a following Super gen and play with prices so that it fits the market shares with just the minimal expense (again take your pick, could be vRAM, could be frequency increase, could be better RT or AI implementation, or of course opening a specific chip in full, and/or a bit of all of these).

As of now nVidia seems to make a point of offering "about" as many lines as in the past, but we can also see how they're skipping on their "traditional" array of cards in the mid/high end so that there is less of them to compare with/upgrade to, but also that each of them is a significant step into leveling or improving market shares still. It's difficult to see an obvious trend in where they decide that a line should come up or be skipped for any gen around, but one thing for sure is that they do re-use any lines they have in waits strictly as needed instead of feeling obligated to make them all within six months of a launch. I don't believe they will strictly follow any trend neither. I mean, all they need to do now is to have a next gen offering with the slightest step-up and see how it goes until they can suck up some money with any next series or upcoming Ti as they feel is needed to fullfil their intended margins AND just not lose a single necessary step in scraping all of the available extra for a similar cost ratio.

They'll start to throw all they've got in for a couple of gens when/if AMD or Intel make any significant threathening leap. I hoped they would do it now while they can achieve and secure a definitive step-up in performance, and I guess they would have if their market share wasn't so dominating already, but as it stands they do not need to do anything much but ensure the xx90 card will reign supreme for about two years, which isn't very difficult to achieve when they have created and gained complete control over a market where any other consumer card they make is basically just skimming some performance increase, and where there is no direct competition to be had anyhow.

SSGBryanBattlemage is targeting 4070 ti in gaming; 4080 in compute. I am not buying an Nvidia card for gaming or productivity, when both Intel & Amd can outperform them at every price point they compete at.

Looks like the only ones I'll be getting going forward are some of those interesting ones I have seen and purchased off of Aliexpress.

(I swear, I can stop buying Aliexpress cards at any time, honest.)


RTX 2080ti with 22gb of ram can be had for $484, shipped to your door. I am debating whether to get one of these for Stable Diffusion.

2080 Ti only has 130 TOPS IIRC, I'd rather 5080 which should be > 1000 TOPS. Even 4080 is 780 TOPS. I'd get a second hand 4090 if they were available and not insane prices as that is 1380 TOPS. For comparison my 6800XT is 68 TOPS and is hopeless for stable diffusion.

The memory bandwidth improvement is not going to save the RTX 5080 when you consider the fact that the RTX 4090 is already starting to struggle in new graphic heavy games. There is no CUDA core increase, and I don't believe we will see any significant improvement in clockspeed either since Nvidia is still keeping to 4nm. Apart from the usual improvement in RT and Tensor cores, I suspect Nvidia is going to do something more to make frame generation better to make up the lack of native performance, while also try to reduce latency.

Which is better 32gig of gddr6 or 16 gig of gddr7

Also nasty TDP regression, feel I did right not waiting.

Take a £800 320w 4080 super over a £1200 400w 5080.

Anyone interested in 5000 series wait for super cards at least they improve value.

watzupkenThe memory bandwidth improvement is not going to save the RTX 5080 when you consider the fact that the RTX 4090 is already starting to struggle in new graphic heavy games. There is no CUDA core increase, and I don't believe we will see any significant improvement in clockspeed either since Nvidia is still keeping to 4nm. Apart from the usual improvement in RT and Tensor cores, I suspect Nvidia is going to do something more to make frame generation better to make up the lack of native performance, while also try to reduce latency.

When I removed Nvidias underclock on my VRAM there was no measurable performance improvement. I be surprised if it has a big effect, I feel Nvidia push out faster VRAM mainly for specs looking better, helps justify their prices. Seen many 4090s with memory bandwidth utilisation under 50% when their core is at 99%.

If they don't release this current 5080 leak as the 5070 (at sane x70 series prices), and pull out a "real" 5080 with 24GB VRAM, this gen is DOA.

Fingers crossed that Jensen had enough nudging from the more in-touch people on his advisory board to attempt to undercut the 40 series instead of ramping prices further. I don't care if it even makes significant improvements over the 40 series, if it manages to ameliorate the price distribution it's enough. They have a unique opportunity to chase AMD and Intel down the price curve and keep their dominant position.

The RTX 4080 also has faster/premium GDDR6X modules rated for 24 Gbps over the 21 Gbps modules used in the RTX 4090. As we have seen with the 4080 - this faster memory alone is nowhere even close to being enough to make up for the narrower 256-bit bus, let alone the hardware's performance deficiency in comparison to the 4090 - however, if Nvidia sells this card for $999 at launch, effectively replacing the 4080 Super at ~4090 performance level, it might be a pretty decent buy. Not great, but a decent buy.

Price makes or breaks it, and I genuinely think this shouldn't cost more than $900 to get an unreserved recommendation from me.

if the 5080 was priced at 1000usd, +10% faster than 4090, cooler dimensions stay the same as 4080/4080S then it's a worthy upgrade for RTX2000/3000 owners

FiredropsIf they don't release this current 5080 leak as the 5070 (at sane x70 series prices), and pull out a "real" 5080 with 24GB VRAM, this gen is DOA.

It's all DOA except 5090 that will sell like hot cakes. All the rest are refreshes with 10-20% improvement. 5080 with only 5% more Cuda than 4080 Super is slower than 4090 for sure.

phintsI'll wait for 6070 and try again. By then PS6 will be coming out

No. Most rumours tagged 2028 for launch. Probably around RTX 7000 series launch.

N/AIt's all DOA except 5090 that will sell like hot cakes. All the rest are refreshes with 10-20% improvement. 5080 with only 5% more Cuda than 4080 Super is slower than 4090 for sure.

4080 has 10% less CUDA cores than 3090Ti, yet beating 3090Ti by 20% at 4K