• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

NVIDIA ‘Ampere’ 8nm Graphics Cards

Seems the 3080 Mobility utilises 16GB VRAM on a 256-bit bus and equivalent to a desktop 2080 Ti... so basically very similar in terms of performance and capabilities to a desktop 3070 with 16GB VRAM. This means that we will likely see a desktop part with 16GB VRAM. https://wccftech.com/nvidia-geforce...ifications-benchmark-leak-on-par-rtx-2080-ti/

This only highlights how unwise a decision and mistake it was by Nvidia to put 10GB VRAM on the desktop 'flagship' 3080 at launch. It's my belief that we are inevitably going to see higher VRAM desktop models (not just the 3080Ti) soon.
 
Surly a mobile GPU is going to run out of chuff long before that extra 6gb of ram becomes useful. The desktop 3070 has even less than that but both desktop parts have much faster memory.

Particularly where power and thermal limits are concerned in most laptop designs (ignoring the silly desktop replacements).
 
Surly a mobile GPU is going to run out of chuff long before that extra 6gb of ram becomes useful. The desktop 3070 has even less than that but both desktop parts have much faster memory.

Particularly where power and thermal limits are concerned in most laptop designs (ignoring the silly desktop replacements).
The 3070 has too little VRAM and the 3080 mobility has too much. However as far as I am aware that's the nature of the architecture... you generally have to go double or nothing the memory in order to keep the same bus width and in the case of the current AMpere lineup that is: 6GB or 12GB (192-bit bus), 8GB or 16GB (256-bit bus), 10GB or 20GB (320-bit bus) etc.
 
Surly a mobile GPU is going to run out of chuff long before that extra 6gb of ram becomes useful. The desktop 3070 has even less than that but both desktop parts have much faster memory.

Particularly where power and thermal limits are concerned in most laptop designs (ignoring the silly desktop replacements).

Most gaming laptops have 1080p high refresh screens so yeah high vram is pointless for 1080p gaming
 
The 3070 has too little VRAM and the 3080 mobility has too much. However as far as I am aware that's the nature of the architecture... you generally have to go double or nothing the memory in order to keep the same bus width and in the case of the current AMpere lineup that is: 6GB or 12GB (192-bit bus), 8GB or 16GB (256-bit bus), 10GB or 20GB (320-bit bus) etc.


Though the 3080 uses GDDR6 and the 3080 uses 6X. I wonder if these proposed desktop variants (Ti/Super) that have more VRAM will use 6 and not 6x as it's much cheaper. Why did Nvidia use 6X which is way more expensive? Does 10GB of 6x do the same as 16GB of 6?
 
6x is faster but nothing like twice as fast. It was 30% IIRC.

Oh right - I've had a good look around and cant find any real world comparisons/tests.

No, speed and capacity are two different things. It was a strange design choice by Nvidia and in practical terms seems to have caused more headaches than it has solved.

I know speed and capacity are different and 6x is more to do with bandwidth isn't it? as it can cycle 2x the instructions 6 can.

I take it then that VRAM usage in games is the same on 6x memory as 6? I couldn't find any comparisons.

I just find it odd that AMD dropped their HBM high bandwidth memory in favour of GDDR6 and Nvidia went for an expensive high bandwidth 6X. Can't find anything concrete anywhere. And apart from your 10GB & 20GB etc for bandwidth which makes sense - nothing concrete as to what Nvidia will use on the Ti/Super. We'll have to wait and see I suppose. JUst odd why Nvidia went for something so expensive.
 
Memory needs to be both sufficient capacity and have sufficient bandwidth to move data.

In theory with faster ram you can cycle out assets more quickly. Without direct access to storage, this has to go via the CPU and can cause other performance issues It’s also slow compared to direct access which is coming soon. Stuttering is a common sign that the graphics memory is at capacity or doesn’t have sufficient bandwidth.

I guess 6X is the future someone has to bite the bullet at some point and start using it.

Vram usage in games is not a great measure. Many games operate on the basis of allocating as much as possible, even if they don’t need it. At the end of the day, game makers make games to the hardware that is available. If most GPUs on the market have less than 10Gb available, they’ll make games to utilise that amount as a maximum. It’s pointless then developing something that doesn’t work or people can’t play.

Games have become increasingly bloated in recent years and outside RTX there is not much to show for the huge bump in hardware requirements. Developers need to go back to the drawing board and refocus on optimising their content rather than trying to just brute force everything with better hardware.
 
I know speed and capacity are different and 6x is more to do with bandwidth isn't it? as it can cycle 2x the instructions 6 can.

I take it then that VRAM usage in games is the same on 6x memory as 6? I couldn't find any comparisons.
Capacity is finite in his context. It doesn't matter if GDDR6X is 30% faster, if a card has 8GB VRAM and needs more, then it has hit a wall.
 
Back
Top Bottom