QuoteBest guess: The mobile "3080" will be GA104

You don't even have to guess, it's as good as confirmed already, though I wished for the mythical GA103 that so far at least didn't materialize.

Kinda sad that the 3080 Mobile will perform worse than a 3070, but then by normal Nvidia standards it would have been called a 3080 if they didn't put that label on what is normally a Ti-class chip, so it can't be helped.

Let's just hope they get on better terms with TSMC next time...

It'd be nice if they were more explicit in their naming schemes, the TDP range makes such a massive difference, but is hidden away by most manufacturers, so it is tougher to tell what you are considering buying.

There's no way GA102 is going into anything other than a DTR notebook design, and even then it would have to be so heavily underclocked as to make the necessary asking price look truly ridiculous.

Best guess: The mobile "3080" will be GA104, so a 3070, but underclocked so far that it performs closer to a desktop 2080 Super than the 2080Ti. The Max-Q version of it will perform more like a desktop 3060  / OG 2080. The cost measured as an increase from low notebook GPUs, meanwhile, will be higher than for the desktop 3080.

It's the same game Nvidia have been playing since Maxwell took the notebook GPU performance crown (aside from a brief respite with Pascal, though they dramatically increased prices to compensate), so why would they stop now?

Yeah, I wish there were laptops of Blade's build quality (sans battery issues lol) but 2x thicker for better cooling and more space for expansion. If only Clevo came out with a metal & glass screen chassis like that...

I don't need minor picture quality detail like RT on very small screen laptops. They should san those rarely use RT cores then cut the price, power consumption and heat by half.

While you won't easily cool a 300W gpu, a good cooling system can probably handle more than 200W on mobile, just forget about thin and elegant designs for these ones.
Since they would be desktop replacement anyway, why not?

I bet we'll get a 10-20% boost depending on the GPU, just like previous gens. Amazingly, eGPUs are as relevant as they've ever been now.

More than ever, lack of innovation with thermals and power are holding things back. But I feel like this last year has shown some progress in that area. For example, the XPS 13 2-in-1 has a vapor chamber, which signals that manufacturers see that even a general audience cares a little about throttling.

Sigh the trend of putting components that can't be cooled and powered in laptops continues.

I gave up on the pipedream and now am a happy desktop user. I use a laptop to remote in but I'm done fighting thermodynamics and paying a hefty price for it.

It's not so much Nvidia as the laptop makers focusing on thin and light laptops. The cooling is just not gonna cut it for anything more than a 90W chip.

There will still be fat Max-P options and even Max-Q can actually be configured for an almost Max-P tier TDP, as seen in a couple of bigger laptops.

But yeah, personally I expected some performance-per-Watt improvement for Ampere, and yet here we are again when I'm seriously considering not upgrading my six year old laptop for another year or two or just jumping ship to a desktop with all the staying at home thing as of late.

 >:( Did Nvidia seriously not learn anything from Turing? Or that the Max-Q are absolute garbage and are equivalent in performance to Pascal gen. GPUs? Great, now we have two more years of Max-Q laptops with Max-P only on 17 inch variants that cost an arm, a leg, and a kidney.