Nvidia's 3000 Series RTX GPU [3090s with different memory capacity]

[CarstenS]

Maybe not so much: https://hiveos.farm/statistics/
scroll down until you reach the "GPU Brands" diagram (at the time of me writing this, it's AMD 61%, Nvidia 39 %). Funny sidenote: The most prevalent RTX is the 3070 (8% of GeForces), which in turn is mostly absent from Steam.

 

[neckthrough]

Funny sidenote: The most prevalent RTX is the 3070 (8% of GeForces), which in turn is mostly absent from Steam.

That observation is consistent with AnandTech's data (https://www.anandtech.com/show/16493/nvidia-launches-cmp-dedicated-mining-hardware) showing the 3070 to have the highest MH/s/W efficiency of all NV cards.

 

[CarstenS]

Some of those are taken from minerstat.com though (as AT note as well) and at least one seems very low (2080 Ti, which is way underrated there. 59,6 MH/s is easily achievable at TBP of 150 watts): https://minerstat.com/mining-calculator

 

[DegustatoR]

ASUS confirms GeForce RTX 3050 Ti Laptop GPU with 4GB GDDR6 memory

 

[Clukos]

I put some liquid metal on the die and managed to hit more than 20k GPU score on Timespy: https://www.3dmark.com/spy/18867500, 12.9K on Port Royal: https://www.3dmark.com/pr/927367".

First time I play with liquid metal but I'm impressed with the difference it makes.

 

[Deleted member 2197]

Blender 2.92 Linux & Windows Performance: Best CPUs & Graphics Cards – Techgage
March 3, 2021

Blender’s latest version, 2.92, has just released, and as usual, we’re going to dig into its performance and see which CPUs and GPUs reign supreme. For something a bit different this go-around, we’re adding Linux results to our rendering and viewport tests, and not surprisingly, the results are interesting!
...
A happy accident occurred during our testing, where our testing scripts thought the GTX 1660 Ti was an RTX card, and proceeded to run the OptiX tests. Since the GTX series doesn’t include RT (or Tensor) cores, that API was typically unavailable. At some point, that changed, because OptiX works on this RT-coreless GTX 1660 Ti, and actually improves performance by doing so. If you have an NVIDIA card, but one that’s not RTX-enabled, you should try out the OptiX option as well and see if performance improves.

 

[DavidGraham]

Still no AMD 6000 series visible on Steam Survey.

That's because AMD sold so few cards, Q4 2020 share of AMD dGPUs is at an all time low of 17%. NVIDIA stands at 83%.

https://www.3dcenter.org/news/die-grafikchip-und-grafikkarten-marktanteile-im-vierten-quartal-2020

I wouldn't hold my breath for the 6000 series after all the unexplained issues Steam Survey has had with AMD products before, nor do I think those numbers would be anywhere near accurate for NVIDIA either.

The survey is a very good indicator of current trends, AMD cards are a no show because they are simply a no show at the market as well, see above.

 

[HLJ]

That's because AMD sold so few cards, Q4 2020 share of AMD dGPUs is at an all time low of 17%. NVIDIA stands at 83%.

https://www.3dcenter.org/news/die-grafikchip-und-grafikkarten-marktanteile-im-vierten-quartal-2020


The survey is a very good indicator of current trends, AMD cards are a no show because they are simply a no show at the market as well, see above.

Yeah, this kinda suprised me:


Seems like NVIDIA going Samsung was a smart move...out of the congestion of TSMC's 7nm

 

[Rootax]

Proposing a better product help too

But yeah, in 2020-2021, having a product available is key...

 

[Putas]

Navi21 does not make sense even for AMD alone. They can have 7 Zen3 CCX for that die space.
For now, they exist only to stay in the game, not to win it.

 

[Kaotik]

Navi21 does not make sense even for AMD alone. They can have 7 Zen3 CCX for that die space.
For now, they exist only to stay in the game, not to win it.

You assume each Navi chip means less Zen chiplets, which isn't necessarily true.
We all know TSMCs 7nm currently is a mess, with no-one but their customers actually knowing what kinda of different 7nm processes there are.
Originally there was N7, N7P and N7+. At that time, Zen-chiplets were supposedly N7, while Navi1x were N7P.
Now they're talking about more or less customized 7nm processes without referring to N7 or N7P (I think MS said N7E?), but no-one has said anything about whether they're actually the same process with each other, or not.

 

[HLJ]

Those numbers acutally make Steam's number kinda valid.
I keep hearing that "Steam is not valid"

Proposing a better product help too

But yeah, in 2020-2021, having a product available is key...

But, but...."Raytracing is not important"
It reminds me of AMD's "Too much tessellation!!!"
I guess we soon will see "Too much upscaling!!!" too

 

[Man from Atlantis]

https://twitter.com/x/status/1369599411320414210

 

[HLJ]

~9M vs ~1.9M is almost a 5:1 ratio, no wonder NVIDIA's revenue is skyrocketing.

 

[yuri]

So for example reporting a 30% month-to-month increase of Chinese players is not a credibility warning sign, right?

I'm not disputing the 83:17 nV vs AMD ratio - its probably even steeper, like 90:10. I'm disputing the Steam Survey since it's obviously a system full of flukes & flaws lacking even basic sanity checks on the provided data...

 

[HLJ]

So for example reporting a 30% month-to-month increase of Chinese players is not a credibility warning sign, right?

I'm not disputing the 83:17 nV vs AMD ratio - its probably even steeper, like 90:10. I'm disputing the Steam Survey since it's obviously a system full of flukes & flaws lacking even basic sanity checks on the provided data...

For a System full of flukes, flaws, lack of sanity-checks etc. it's does quite the good job...

 

[yuri]

it's does quite the good job...

This lies in the eye of the beholder. Too bad for people who passed Statistics 101.

 

[HLJ]

This lies in the eye of the beholder. Too bad for people who passed Statistics 101.

I find it amazing that such a "flawed" metric can be so close to the reality...it's almost as it defies reality /shrugs

 

[swaaye]

Why do I never see the 2013/2014 coin frenzy mentioned? I bought a R9 290X for cheap off fleaBay after that crashed.

And that is some impressively low volume for AMD 6x00 GPUs! Yikes.

 

[Davros]

I'm disputing the Steam Survey since it's obviously a system full of flukes & flaws lacking even basic sanity checks on the provided data...

Every month valve index is at worst the 3rd best selling item on steam has been as far back as I can remember I dont believe it also valve released their best selling charts for the year and it wasnt even in the top 100
The Year's Top 100 Games as Measured by Gross Revenue
https://store.steampowered.com/sale/BestOf2020

Edit : nearly always top 3 a couple of dips to 6 then disapears out of the top 10 in sept but is back in august (I'm working back from dec)
you have to go right back to feb for there to be a time it wasnt in the top ten
https://steamdb.info/topsellers/2020W52/