• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

NVIDIA GeForce RTX 5070 Founders Edition

absolute amounts of VRAM don't matter
Season 9 Lol GIF by The Office
 
Other things the braindead are unable to understand:
  • ray- and path-tracing are the past and the future of graphics rendering, the only reason they dislike it is because AMD is bad at it
  • upscaling and frame generation are here to stay, again the only reason they dislike these features is because NVIDIA is better at them

Plenty of nvidia card owners also think ray-tracing is overrated or a misuse of resources, and/or aren't sold on the benefits of upscaling + frame generation FWIW. The latter has become a crutch for developers as it was always predicted to be.
 
So 5070 is still better then 7090 XT in upscaling ? 5070 is 700€ in Europe 12GB
Why there is no more games with upscaling test?
Please test more games with upscaling
@W1zzard please add pricing also for Europe
It's so hard to buy in Europe while pricing can be even 10-20% bigger first $ to € conversion is direct 1:1 and often in US is just much cheaper. Plus the shipping is 8-10€ + 10€-15€ with parcel forwarding to Slovenia
https://geizhals.eu/ (compare.eu) is price comparison for Europe with a lot of filters and technical details and
pricing. Even VRM details for motherboards for example
https://www.idealo.de/ is another but just for Germany

5070 is 700€ in Europe 12GB 7900 XT is 750€
5070 category
7900 XT is 750€
 

Attachments

  • Zotac 5070 700€.png
    Zotac 5070 700€.png
    99.3 KB · Views: 41
  • Zotac 5070 760€.png
    Zotac 5070 760€.png
    99.3 KB · Views: 35
Last edited:
HD 7950/70: these 3 GB VRAM don't do anything at the launch and when they do, these GPUs run out of juice anyway.
R9 290/X: GTX 980 isn't slower and needs much less power.
R9 Fury: nuff said.
RX 480: doesn't even remotely come close to Pascal in energy efficiency, not to mention it doesn't do anything to segments above the very bottom.
RX 5000 series: "Maybe this time? Please! Eh, black screen and crashes, dang it!"
RX 6000 series: made of pure unobtanium and still don't outclass any of NVIDIA GPUs other than the very worst ones.
RX 7000 series: overpriced, underspecced, failed chiplet implementation, non-existent RT, obscenely worse upscaling.

There were absolutely ZERO successful launches since early 2010s. All AMD do is "NVIDIA minus 50 dollars" or "NVIDIA minus 50 features." Even if RDNA4 is awesome they're showing up for work that does no longer exist.
Realize all AMD has to do is have something on the shelf and it's a better launch than the 5000 series.
 
You can keep on being wrong all you like. Doesn't change that you're wrong.
Have you seen the "super early, days before anyone else preview of 3080" by PF?

When they seconded Huang's 5070 = 4090 "3080 is 2 times faster than 2080".
Do you know how the actual 1.8 lead in one outlier benchmark was achieved?

By making it NOT fit in 2080's memory. :D
Which was even more hilarious given how low on VRAM the 3000 series is. (I blame Lisa)

The main reason 3070+ users want to upgrade IS VRAM. Mkay?

If we take spherical horse in vacuum, apply "94% compression" and something something AI something something, it is a lot of fun.


But it does not help 3000 users.
It does not make 5070 with its 12GB and shockingly poor performance any less terrible.
Right here, right now.

Spherical horses in vacuum are different, no disagreement here.

PS
1741185460181.png
 
This review has been updated with data from other non-MSRP cards: heat / noise, cooler comparison, overclocking
Legend.
 
DLSS = Deep Lying Strike Shoppers.
 
Other things the braindead are unable to understand:
  • ray- and path-tracing are the past and the future of graphics rendering, the only reason they dislike it is because AMD is bad at it
  • upscaling and frame generation are here to stay, again the only reason they dislike these features is because NVIDIA is better at them
  • absolute amounts of VRAM don't matter, what you do with it is what counts
  • criticising company A does not automatically mean an endorsement of company B
  • there is such a thing as "constructive criticism"
I'll raise ya one.
  • Ray/path-tracing are still best employed asynchronously, as it requires some amount of often lossy denoising in order for it to look coherent, realtime or not (see TAA/DLSS denoising/Optix) and despite generations of investment in ray accelerators, using it in realtime bears a heavy performance cost for little visual improvement. Ray acceleration is still nice, but it has trade-offs. Rasterization will not be replaced by it.
  • Upscaling/frame-gen are similarly nice-to-haves but it's an enhancement being employed as a crutch when your fancy-pants new card suddenly can't make metrics, and often the use of it as such makes the tech look noticeably worse and exaggerates shortcomings in competing solutions.
  • Lmfao. There have been and continue to be consistent and recurring issues with lacking video memory, the least affecting symptoms being the need to turn down memory-hungry settings (including raytracing/frame-gen) to stay under the ceiling. That issue will only persist the more a performance segment is permitted to have lopsided performance compared to memory allowance. 12GB is great for an entry level card (love ya 3060 12GB, B580) but the 5070 is sold and marketed as midrange. It needs to offer more than its older brother the 4070 SUPER, and no amount of software suites will get it there. Even a bare 2GB extra would have been better.
  • Yes and no. Ragging on Company A in opposition and furthermore comparison to Company B carries implicit weight. Explicit praises and criticisms are of course of much greater value, but there still exists a bias. It's better to be openly biased than to wax philosophical about how saying one thing does not imply another derived argument. With that said, go AMD. They offer a better product for my use case and semantic wants.
  • "Constructive Criticism". Hold on, let me just...
You can keep on being wrong all you like. Doesn't change that you're wrong.
:laugh:

Respect given is respect returned. Just saying 'You're wrong' instead of proposing a reasonable counter-argument is not constructive at all, so don't expect as much from anyone.
 
Last edited:
embrace your goldfish memory enthusiasm!
Hey, don't be jerk to me just because I offered an alternate perspective that offends your fragile sensibilities.

but today it's largely an AMD fanboy community
While there is some of that, it's not as severe as you make it out to be. And there's just as much NVidia fanboying..

is because AMD is bad at it
Not anymore, or have you not seen the latest reviews? While AMD's offering don't beat NVidia outright, they're close and those two new GPUs offer very reasonable competition this round. You know me, I'm an NVidia guy, and still going to be one, but I have to tip my hat to AMD, they're coming up.

If you want to feel good about buying Nvidia, why bother going on the internet at all to read/watch reviews and comments? Just buy the most expensive Nvidia card you can afford every time and call it a day. That would be the end result for you either way, right?
No, I might just stick it to NVidia this round and do something else.

AMD 9070 should have no problem kicking the crap out of this POS card. That bad.
And the reviews are out. They're competitive and solidly so. However "kicking the crap out of" is a big stretch.

Not considering a 9070XT?

Looks like the better offer thus far...
Nope. As mentioned earlier, I'm "locked" into the NVidia side of things. I don't just game. Compute is important and AMD doesn't have any comparable options currently. However, I might just separate my gaming/compute rig into separate systems if AMD can pull out all the stops. Been thinking about that kind of move anyway. Often find the two tasks overlapping and it's getting cumbersome.
Even with the new reviews about AMD's offerings, I'm not switching sides. The advances offered by AMD are very good and bring them on par with NVidia in those classes of cards, they're not good enough for me to abandon a great deal of work and have to redo it all just to focus on a new hardware platform. I might just skip this round altogether.. It'll be the first time I've skipped an entire generation of GPU's from camp NV, let alone two. Although, it evens out a bit. I have had a card from every NV gen from the Geforce 6000(mid 2000's) series forward.
 
Last edited:
This piece of crap gpu is 699.- here on average, and the only in stock is msi for 709.-, lmao! on day one, this garbage cannot get worse
 
Stop posting this deliberately misleading bulls**t. Compare the 5070 Ti to the 3070 Ti, or the 5070 to the 3070, and tell me what those performance differences are.


It used to be, but today it's largely an AMD fanboy community, most of whom neither understand the realities of silicon engineering and why it's no longer possible to get massive generational performance improvements, nor want to understand it. Because it's far easier to post "ngreedia" and get circle-jerk agreed with by other similarly braindead members.

Other things the braindead are unable to understand:
  • ray- and path-tracing are the past and the future of graphics rendering, the only reason they dislike it is because AMD is bad at it
  • upscaling and frame generation are here to stay, again the only reason they dislike these features is because NVIDIA is better at them
  • absolute amounts of VRAM don't matter, what you do with it is what counts
  • criticising company A does not automatically mean an endorsement of company B
  • there is such a thing as "constructive criticism"


Good luck with that, considering almost every PC game is a console port nowadays.
How about just no? You're completely missing the mark here. Nvidia is simply pulling all the stops to offer minimum hardware for maximum gain. They always did. You're just deluding yourself. They create a software illusion to hide the harsh hardware reality. Even Nvidia isn't denying it anymore, calling itself a software company. But somehow, you are. Curious indeed. 'Braindead', you say... interesting pov on customers making their own choices.
 
Another failed GPU made by Nvidia yours truly, nvidia just want money nowadays, it was the day when nvidia cared about giving its supporters something worth the price. GPU market is done for, something big needs to happen to change the spiral death.
 
Maybe that's all they could really get from Nvidia, are chips to build 5070s.

Right now my local Micro Center only shows 1 model of a RTX 50x0 card available - an ASUS 5080 (there are 2 in stock, yesterday afternoon there were 6) listed for $1650. 4 people actually paid nearly $1800 (after taxes) for a 5080 yesterday. Holy hell....what is wrong with these people?
 
This coming from someone with the AMD logo for an Avatar? Really? :slap:
It is a very nice logo. Is there anything else to this story?


Maybe that's all they could really get from Nvidia, are chips to build 5070s.

Right now my local Micro Center only shows 1 model of a RTX 50x0 card available - an ASUS 5080 (there are 2 in stock, yesterday afternoon there were 6) listed for $1650. 4 people actually paid nearly $1800 (after taxes) for a 5080 yesterday. Holy hell....what is wrong with these people?
FOMO is what is wrong with these people. Nevertheless, you could imagine situation in EU when you have bad situation in US regarding prices and stock.
 
You missed the context and implication there. I'm not going to outright call you a "fanboy", but come on, really with that statement? From YOU?
Yep. I am a AMD Radeon fanboy. With nVidia GeForce RTX 4080 GPU in my main rig, in which Radeon was not in since 2004 and 9700Pro.

Back to subject - still puzzles me though, why the hell 19 models od 5070? It is perfectly clear that GPP is fully active at MSI but still... 19 models. Why? For whom?
 
It is a very nice logo. Is there anything else to this story?



FOMO is what is wrong with these people. Nevertheless, you could imagine situation in EU when you have bad situation in US regarding prices and stock.

Surprisingly, there even is a stock of RTX 5070 in EU shops right now, with a lot of models available in many shops.

But the cards came so pre-scalped there isn't much interest - cheapest you can get is 730 EUR, about $800. Not really a good price / performance point compared to offerings before Blackwell launch.
 
Back to subject - still puzzles me though, why the hell 19 models od 5070? It is perfectly clear that GPP is fully active at MSI but still... 19 models. Why? For whom?
How is that puzzling? At all? Over a dozen companies and they create 2 or 3 models of card for each GPU line. Par for the course. I'm surprised there aren't more models.
 
Back
Top