The mobile GeForce RTX 3080 is 40 percent slower than the desktop GeForce RTX 3080. What the hell happened?

[Hakim]

How high up the list is thickness in people's priority when buying a gaming laptop?

[Bennyg]

It's all to do with power

In the Pascal era the mobile 1080 had similar performance because the power limits were similar to the desktop 1080. NVIDIA sandbagged and sold this as their 'high end' GPU as they didn't need a more powerful GPU to own the competition at the time, and they had the 1080ti waiting for when they did.
... Therefore Nvidia's laptop gaming GPU marketing.

With the 3xxx generation the desktop chips eat more than double the power, that's why mobile variant performance is lower.

The only reason anybody would think ZOMG mobile 3080 is slower!??' is because they haven't read and understood the specs and believe that NVIDIA naming it 3080 means something.

And this level of power - 300W plus - of GPU is not possible in anything thinner than a 5kg 2 inch thick DTR with 60dba fans like the Clevo P870.

And given the shift to thinner and lighter across all segments including gaming, the market has voted that it doesn't want full desktop performance in thick heavy expensive laptops, it wants reduced performance in reduced weight expensive laptops.

[Bennyg]

It's all to do with power

In the Pascal era the mobile 1080 had similar performance because the power limits were similar to the desktop 1080. NVIDIA sandbagged and sold this as their 'high end' GPU as they didn't need a more powerful GPU to own the competition at the time, and they had the 1080ti waiting for when they did.
... Therefore Nvidia's laptop gaming GPU marketing.

With the 3xxx generation the desktop chips eat more than double the power, that's why mobile variant performance is lower.

The only reason anybody would think ZOMG mobile 3080 is slower!??' is because they haven't read and understood the specs and believe that NVIDIA naming it 3080 means something.

And this level of power - 300W plus - of GPU is not possible in anything thinner than a 5kg 2 inch thick DTR with 60dba fans like the Clevo P870.

And given the shift to thinner and lighter across all segments including gaming, the market has voted that it doesn't want full desktop performance in full bodied thick and heavy expensive laptops, it wants reduced performance in lighter thinner expensive laptops.

[SERGEY DALECHIN]

THE HECK? why zero mention on the main reason it's so much slower - the fact that the 3080m is using an enitely different chip alltogether, with a fricken 42% difference in cuda core count? If they used exactly the same chip and memory the desktop 3080 uses, their performance would be much closer together even with a much lower TDP on laptops.
I thought you guys are experienced in this. Whereas 1000 and 2000 series used exactly the same GPU-s in laptops and desktops, now laptops have cut down versions of them EXACTLY like it was before Pascal - M GPU-s, which is THE MAIN reason why 3080m is so pathetic. Power contraints hurt performance, sure, and the gap would be bigger no matter what they did, but they could at least use the same chips. But I guess greed ruined the utopia of laptops being equal to desktops.

[TabalugaDragon]

You cannot defy physics: power supply, cooling and space is limited on laptops, so it's ok.

What's more important: is the RTX 3080 (150W, 16GB) worth getting over the RTX 3070 (140W, 8GB) e.g. in the Tongfang Schenker XMG Neo 15 (AMD Ryzen 5800H)?

apparently you can't break the laws of greed either. The main reason why 3080m is SO MUCH behind the desktop 3080 is not the power, but the fact that it uses a 3070 chip, NOT GA102 chip that the desktop version uses, hence a whapping 42% cuda core difference and crazy difference in performance. I am not saying power doesn't matter - it does of course, but the difference in chips matters even more. The more cuda cores and the better the memory - the more energy efficient a GPU becomes, even at lower clockspeeds. But Nvidia was way too greedy to use the desktop 3080 chip and here's the result.