Nvidia RTX 3060 Can Beat an RTX 3080?! AMD Navi 3X GPU Spotted?

[Gav]

You know how Nvidia has released many versions of their GPUs in the past? Well, they’re at it again. Their RTX 3000 series GPU’s for laptops contain not 4, not 8 but 28 different variants of cards. Computerbase.De managed to find a list of all of the variants for the RTX 3060, 3070 and 3080 Laptop GPU’s.

So why are there so many? Well, that’s thanks to all the different power requirements for each model. It goes anywhere from 60W for some RTX 3060 laptop GPU’s to 150W for some RTX 3080 laptop gpu’s.

While it seems like its not that big of a deal, it actually is. Depending on the GPU’s TGP the clock speeds will be altered meaning that an RTX 3070 with a higher TGP, could be faster than an RTX 3080 with a lower TGP. In fact, if you just look at the Max TFLOPS performance, this RTX 3070 with a TGP of 125W could be up to 30% faster than this RTX 3080. That’s because the latter is incredibly underclocked. A RTX 3060 laptop GPU could beat that RTX 3080!

 

[clshades]

Sounds like a lawsuit in the making.

 

[Soultribunal]

The removal of Max-Q / Max-P is just going to make the market and landscape so damn confusing. I know that Schenker is actually listing for their Series the actual TGP , so that way their buyers are well informed on what they are getting. If other manufacturers do not follow suit, its going to be a nightmare to navigate with this generation.

-ST

 

[sswilson]

Hasn't it always been like this, or is the difference now that it's going to apply to "real" dedicated graphics instead of the past when it applied to the naming conventions of the "just slightly better than integrated" offerings?

Maybe times have changed over the last time I looked into this (probably about 10 years ago), but I remember laptop graphics conventions being completely non-sensical with very little indication between different SKUs as to what was the better performer.

 

[Soultribunal]

Hasn't it always been like this, or is the difference now that it's going to apply to "real" dedicated graphics instead of the past when it applied to the naming conventions of the "just slightly better than integrated" offerings?

Maybe times have changed over the last time I looked into this (probably about 10 years ago), but I remember laptop graphics conventions being completely non-sensical with very little indication between different SKUs as to what was the better performer.

I think it was more nonsencial because you had to figure out what a 'Mobile' Version of your Graphics was comparable too in Desktop form so you could gauge how well your system would run the games you liked.
They had gotten better with Max-Q because that told you it was a cut down version of the 'full fledged mobile chip' so you knew you were getting a reduction in performance.

What they have done now, is taken chips that have a highly configurable TGP and not given them any labels. SO now its going to be a mucked up landscape where a lesser chip with a higher TGP can outperform its higher end 'sibling'.

What a disaster.

-ST

 

[Bond007]

I have 3 main thoughts with this.

1. I am completely fine with ditching the old Max q and normal gpus as separate lines. There is no reason.

2. I wish it was a single line of gpus running as efficiently as they can and with a fixed tdp per model. Just have more models. As much as I am sure everyone would this, I know it won’t happen because marketing and money. They make more money this way and marketing high model numbers sells.

3. Given the above 2 points, I accept this change as a bit of a middle ground....however, it should be mandatory to have something defining the actual performance level you are getting. Whether that is staying the tdp, or some kind of performance scaling between the lineup, or SOMETHING. It should be illegal to sell without the consumer knowing what you are buying.

 

[Soultribunal]

Interestingly enough I was reading a few articles this morning with Nvidia actually requiring the TGP of the GPU's to be listed on their products.

I think they realized the backlash of not letting that happen and have started to back peddle a bit.

-ST

 

[Bond007]

Interestingly enough I was reading a few articles this morning with Nvidia actually requiring the TGP of the GPU's to be listed on their products.

I think they realized the backlash of not letting that happen and have started to back peddle a bit.

-ST

I have read that now as well. Seems to be on the right track...though it will make for some interesting reviews. Will you be better with a higher tdp, but lower tier GPU? What tdp will be peak efficiency? I have a feeling that up to a point the tdp will play a bigger part than the model GPU.

 

[Soultribunal]

I have read that now as well. Seems to on the right track...though it will make for some interesting reviews. Will you be better with a higher tdp, but lower tier GPU? What tdp will be peak efficiency?

Given some of the initial reviews, I would take a 'full fat' 3060 over a cut down 3070/3080. I think it will all depend on how it is designed into the chassis.

-ST

 

[Marzipan]

yep, Nvidia has told their partners they must share the details of the GPU they use:

NVIDIA requires RTX 30 laptop makers to be transparent on specs

NVIDIA now requires that laptop makers using RTX 30 graphics be transparent about their GPU specs to avoid confusion.

www.engadget.com