NVIDIA Tunes GeForce RTX 5080 GDDR7 Memory to 32 Gbps, RTX 5070 Launches at CES

[StimpsonJCat]

It's about what you get for what you pay. This is the second time when Ngreedia tries to sell us lower-specified product with a sticker of "premium/high performance product". Same things as with two versions of RTX 4080 before. It's more like: How dare they? 12 GB VRAM for $600-700 GPU in 2024 is ridiculous, a ripoff. Of course, Nvidia does this on purpose so they can release 2 another versions of same card few months later. The RTX 4080 Super is fail among fails, that card is not even worth printing the boxes it's stored in.


DLSS and RT and similar stuff occupies noticeable space of VRAM for it's own caching purposes. VRAM is not only for textures. Yes, faster memory has higher bandwidth so it can make up for time lost with loading stuff into the slower memory. Having more VRAM means that sometimes there's no need for so many loadings and that enables disk, DRAM and CPU to focus on other operations.

Some games checks for VRAM size and don't let you ramp up certain graphical settings to the highest possible values due to not having enough VRAM.

Some games rely heavily on VRAM size in higher resolutions, as shown in the video above. Lows are much better with more VRAM. Especially take a look at Last of Us at 4K, RTX 4060 Ti 8 GB is completely messed up. Please, DO note that in order to compensate lack of enough VRAM, driver uses system memory (RAM) for this purpose. RAM not only is slower but might be required for other purposes. So, having more VRAM is better because the card will not parasite on other computer's resources.

As shown in the video above, sometimes takes more than 3 GB from the RAM. Gaming with 16 GB RAM and RTX 4060 Ti 8GB may easily become a stuttering festival past 1080p. Same logic applies to 12 GB, 16 GB, 20 GB, ... When there is not enough, it will look for it elsewhere.

nGreedia - I love that name, so fitting, so I'm only going to use that name from now on!

But nGreedia is only doing what it tried to get away with 2 years ago when the 40x0 series launched. They tried to pass off a 12GB version of the 4080 as a full 4080, and charge a premium for it, and consumers were outraged and actually got nGreedia to back down and cancel that card, only for them to later offer it for what it really was, a 4070 Super. So nGreedia have tried this product stack slide before, and they are desperate to try it again with the 50x0 series next year.

The 5080 leaked details are obviously what should be the 5070 12GB and 5070 super 16GB. The rumour is that the real 5080 with 24GB of VRAM will be launched later, and cost $1500+. A complete rip-off, and a middle finger to customers, but that's just how nGreedia thinks of us now.

We can only hope these rumours are fake. But I have a bad feeling about this.

 

[LittleBro]

They no longer are dependent on gamers. We are just a few single digit % of their income. Nvidia is all AI now, which as has been pointed out in this thread, is not really AI, it's more like accelerated computing for LLM and generative stuff. (By the way, I fully support this point, what Jensy calls "AI" is definitely not "AI".) Nvidia can have this disrespectful attitude towards gamers, because Nvidia now lives from orders of "AI" customers.

I remember before cryptocraze Nvidia was not so greedy, margins were lower. Since cryptocraze, Nvidia raised pricing enormously. Okay, there were crazy prices of GPUs during cryptocraze, as GPUs were not available at shops but who would not like to continue selling at twice the price as before? Especially when you see, that there are people who actually buy it. Nvidia got used to cryptocraze pricing and since then, the margins are much higher. Well, it hurts to sell again for less, I guess.

https://www.digitaltrends.com/computing/nvidia-says-falling-gpu-prices-are-over/

The problem is that model names and prices no longer scale with shaders, as some contributors in this thread have already pointed out. RTX 3080 has roughly 20% less shaders than RTX 3090 Ti, RTX 4080 has roughly 40% less shaders than RTX 4090 and RTX 5080 is rumored to have roughly half of the RTX 5090 shaders. The gap is widening, prices are growing. One may think that Nvidia intends RTX 5090 for China, which is probably the biggest consumer of RTX 4090 series. Of course, Nvidia can't sell it to China due to US restrictions, so they will lease it to them instead.

(Greyed text is my own speculation.)

Btw, I'd like to see how RTX 4070 fares with newest games on system with 16 GB RAM compared to 32 GB RAM.

 

[Legacy-ZA]

nGreedia - I love that name, so fitting, so I'm only going to use that name from now on!

But nGreedia is only doing what it tried to get away with 2 years ago when the 40x0 series launched. They tried to pass off a 12GB version of the 4080 as a full 4080, and charge a premium for it, and consumers were outraged and actually got nGreedia to back down and cancel that card, only for them to later offer it for what it really was, a 4070 Super. So nGreedia have tried this product stack slide before, and they are desperate to try it again with the 50x0 series next year.

The 5080 leaked details are obviously what should be the 5070 12GB and 5070 super 16GB. The rumour is that the real 5080 with 24GB of VRAM will be launched later, and cost $1500+. A complete rip-off, and a middle finger to customers, but that's just how nGreedia thinks of us now.

We can only hope these rumours are fake. But I have a bad feeling about this.

It's better to view them like this:

5080 = 5060
5070 = 5050

 

[TumbleGeorge]

A complete rip-off, and a middle finger to customers, but that's just how nGreedia thinks of us now

You're unhappy and won't get green popcorn.

 

[rtwjunkie]

On a related side-note, it seems the best value NV card next year is going to be the 4070ti, but we all know that NV will cancel that card ASAP.

I thought they already did cancel the 4070 Ti when they introduced the 4070 Ti Super? You can buy them from still in the pipeline, but they are cancelled, production-wise.

edited bc fat fingers typed 4970 and not 4070

 

[efikkan]

The main point that seems to escape you though, is one of longevity. We probably will have to wait 2 years with these cards, so do you think games in 2 years time will still work well with a 12GB frame buffer, or even a 16GB buffer?

No it hasen't escaped me, I've just seen the same cycle of nonsense repeat for over twenty years, and time after time we see graphics cards run out of computational power and/or memory bandwidth long before VRAM. For any who possesses basic knowledge of GPUs this should be plainly obvious; if future more demanding games will use more VRAM, they will also require more bandwidth and GPU performance to utilize it within a given frametime, which by logical deduction will make it impossible to scale further with just VRAM size alone. Or to put it in more simple terms; unless the GPU and memory bus has loads of untapped potential, slapping more VRAM isn't going to make it any more future proof, and thinking so would be laughable.

Now, go clean your room and learn some manners!

 

[wolf]

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

Considering 4070 has 15%~ more performance than 3070 Ti with less shaders and memory bandwidth, 5070 will obviously be even faster than that. You can't compare shader counts cross generationally

Core counts certainly are one of the indicators, but history has shown time again that there are far too many other factors than to just look at shader counts gen to gen for performance estimations. I certainly wouldn't hang my hat on a core count alone for a perf estimate.

 

[chrcoluk]

It's about what you get for what you pay. This is the second time when Ngreedia tries to sell us lower-specified product with a sticker of "premium/high performance product". Same things as with two versions of RTX 4080 before. It's more like: How dare they? 12 GB VRAM for $600-700 GPU in 2024 is ridiculous, a ripoff. Of course, Nvidia does this on purpose so they can release 2 another versions of same card few months later. The RTX 4080 Super is fail among fails, that card is not even worth printing the boxes it's stored in.


DLSS and RT and similar stuff occupies noticeable space of VRAM for it's own caching purposes. VRAM is not only for textures. Yes, faster memory has higher bandwidth so it can make up for time lost with loading stuff into the slower memory. Having more VRAM means that sometimes there's no need for so many loadings and that enables disk, DRAM and CPU to focus on other operations.

Some games checks for VRAM size and don't let you ramp up certain graphical settings to the highest possible values due to not having enough VRAM.

Some games rely heavily on VRAM size in higher resolutions, as shown in the video above. Lows are much better with more VRAM. Especially take a look at Last of Us at 4K, RTX 4060 Ti 8 GB is completely messed up. Please, do note that in order to compensate for lack of enough VRAM, driver uses system memory (RAM) for this purpose. RAM not only is slower than VRAM but might be required for other purposes. So, having more VRAM is better because the card will not parasite on other computer's resources.

As shown in the video above, sometimes it eats more than 3 GB from the RAM. Gaming with 16 GB RAM and RTX 4060 Ti 8GB may easily become a stuttering festival past 1080p. Same logic applies to 12 GB, 16 GB, 20 GB, ... When there is not enough video memory in the graphics card, driver will look for it elsewhere.

Looks like he is just looking at performance and reported VRAM usage, what he should be doing is analysing stutters, dynamic level of detail, and if any textures are missing.

 

[Knight47]

Do we know what exclusive features will Rtx 5000 series bring for gamers?

Not sure if I should wait for 5070 or get a 4070 Ti Super now. My current gpu can barely handle 1440p with 60-65 fps.

 

[StimpsonJCat]

Do we know what exclusive features will Rtx 5000 series bring for gamers?

Not sure if I should wait for 5070 or get a 4070 Ti Super now. My current gpu can barely handle 1440p with 60-65 fps.

There have been rumours about DLSS 4 - Probably some A.I. BS artificially limited to these and possibly the previous gen cards. I've also heard DisplayPort 2.1 - Don't know if we can count on UHBR support though, it would be a real shame if it didn't, as nGreedia has held the entire monitor industry back with DP 1.4.

I've also heard that RT perf is not increased much either 15-25% is what I heard. I hope this is wrong, because it's too slow still by far.

I heard about some kind of better texture and colour compression, because a 12GB card is really gonna rely on that! But I don't think that's possible without introducing artifacts.

I don't think there are many or any architectural changes other than maybe a little more of everything, so don't expect much.

 

[Knight47]

There have been rumours about DLSS 4 - Probably some A.I. BS artificially limited to these and possibly the previous gen cards. I've also heard DisplayPort 2.1 - Don't know if we can count on UHBR support though, it would be a real shame if it didn't, as nGreedia has held the entire monitor industry back with DP 1.4.

I've also heard that RT perf is not increased much either 15-25% is what I heard. I hope this is wrong, because it's too slow still by far.

I heard about some kind of better texture and colour compression, because a 12GB card is really gonna rely on that! But I don't think that's possible without introducing artifacts.

I don't think there are many or any architectural changes other than maybe a little more of everything, so don't expect much.

Thanks. I only care about DLSS 4(since 3 is locked to 4000 series).

I don't trust the 12GB is 16GB, that's some bs that Apple tried to pull by saying 8GB in Mac is like 16GB in windows.

I'm already near to max out 8GB vram while doing some casual photo editing stuff while having Firefox open in background. I not liking the idea of getting a 12GB for the next 3-5 years.

 

[StimpsonJCat]

I don't trust the 12GB is 16GB, that's some bs that Apple tried to pull by saying 8GB in Mac is like 16GB in windows.

I'm already near to max out 8GB vram while doing some casual photo editing stuff while having Firefox open in background. I not liking the idea of getting a 12GB for the next 3-5 years.

Nobody but a handful of arrogant old men who just use office and weird browser distros behind proxies are happy about the memory status of the upcoming 50x0 series.

 

[LittleBro]

Looks like he is just looking at performance and reported VRAM usage, what he should be doing is analysing stutters, dynamic level of detail, and if any textures are missing.

What is to analyse when you have 10-15 FPS less, or even 4-5 FPS instead of 50? Even if we don't take VRAM usage into account, stutters are much lesser with more VRAM, as his software is reporting better lows with more VRAM.