First NVIDIA Ampere mobile benchmarks: Up to 18% faster than Turing mobile, but don't expect Ampere desktop-class performance

[Redaktion]

We have obtained exclusive first synthetic benchmarks run on pre-sample gaming laptops powered by NVIDIA's new Ampere mobile GPUs, the RTX 3060 Mobile, RTX 3070 Mobile, and the RTX 3080 Mobile. These preliminary numbers show Ampere mobile offering anywhere between 7% to 18% performance improvements over their Turing predecessors. However, they do seem to lag behind their desktop versions considerably.

https://www.notebookcheck.net/First-NVIDIA-Ampere-mobile-benchmarks-Up-to-18-faster-than-Turing-mobile-but-don-t-expect-Ampere-desktop-class-performance.514698.0.html

[S.Yu]

Hmmm, the gap seems nearly equal to that of the 20 desktop-laptop series, slightly higher but better than expected. That is, if they didn't bump the TGP of the Mobile cards.
We shall see if the Max Q versions have a greater deficit to regular Mobile than last generation.

[John Doe]

Better than expected? Looking at the 3dMark Time Spy chart, 2080 Super Desktop was like 10% faster than 2080 Super Mobile, while 3080 Desktop is 45% faster than 3080 Mobile. Given the results, the 3080 Mobile should have been labeled as 3070 Mobile instead.

[AllahuAkbar!]

As long as there will be the following laptops this year, I won't riot in Capitol:

1) Lenovo ThinkPad X1 Extreme/Dell XPS 15" with
AMD Ryzen 9 5900HS + GTX 1660 Ti
2) Levion T15 with
AMD Ryzen 7 5800HS + RTX 3060 Max-Quran
(Half the chassis is empty!)
3) Asus G14 with webcam, better keyboard and
AMD Ryzen 9 5900HS + RTX 3060 Max-Quran
4) Lenovo Legion 5 slim with better keyboard and
AMD Ryzen 7 5800H + RTX 3060-3080
5) Dell XPS 17 with better keyboard and
AMD Ryzen 9 5900HS + RTX 3060 Max-Quran
6) Alienware slim 15" with matte frame/chassis, better keyboard and
AMD Ryzen 9 5900H + RTX 3080
7) Lenovo ThinkPad T14 with
AMD Ryzen 7 5800HS + onboard GPU

[DF]

Pretty likely we also get HDMI 2.1 outputs from the new mobile cards.  That will come in handy down the road as the data rate is quite a bit higher than Displayport 1.4 as well as supporting higher audio channel standards.

[Halah]

The performance increase from RTx 2 series is way below than expected for those mobile cards imho. Such a shame..

[S.Yu]

Better than expected? Looking at the 3dMark Time Spy chart, 2080 Super Desktop was like 10% faster than 2080 Super Mobile, while 3080 Desktop is 45% faster than 3080 Mobile. Given the results, the 3080 Mobile should have been labeled as 3070 Mobile instead.

Yeah now if you'll look past the 3080 then the others are all much closer, and remember that a previous famous estimate was that the 3080 would scale linearly down to 70W, this is far better than that, being less than half the TGP at 70% the performance.

[fsdfsdf]

So... 3060 mobile scores a bit less than top implementations of 2070 Super mobile 115W. Not impressive, but still i wonder what wattage is this 3060 actually?

[Rrx3080]

What TDP was this chip running at?

[Zoltán Laky]

I'll take a moderate step forward (10-20%) over last gen as long as efficiency is improved and the laptops run quiter and cooler, instead of a big performance leap.

[Temerarious]

I was going to replace my 2070 Super Alienware M15 R3 I got for $1,300 but after looking at this, I wonder is the 3060 Mobile really only 4% better at $999.99 low MSRP (with probably bad screen, low hertz and ms I'm guessing)?  If that's the case I don't think I need to replace my laptop anytime soon.

The 3070 though on a black friday deal for a higher end machine might be worth it.  I feel like the price of the laptops with 2000 gen is the same as the 3000 gen with the difference that 3000 gen is 18% - 30% improvement.  Depending on the price a person can get the 2000 RTX laptops for, the 2000 RTX laptops seem like they are still viable.

[ilikegames]

Nvidia's Ampere Architecture really amped me up, lol. Jokes aside, it is evident that the mighty beast 320 watt RTX Geforce 3080 is now showing its weakness. When you built the product on a older node just to save "money", the burdens of "performance hit" & "supply shortage" will be carried onto customers as a result. Rumours insinuate that Nvidia wanted a price cut from TSMC and that's why, they stated to produce their GPUs under Samsung 8nm Node but TSMC being TSMC didn't budge and Nvidia is paying for that blunder. Well, technically, consumers are facing the problems at the end of the day. (High power required, low performance per watt, lower yield due to chips being very large and other countless issues that comes when not calculating the after-effects.)

[S.Yu]

Nvidia's Ampere Architecture really amped me up, lol. Jokes aside, it is evident that the mighty beast 320 watt RTX Geforce 3080 is now showing its weakness. When you built the product on a older node just to save "money", the burdens of "performance hit" & "supply shortage" will be carried onto customers as a result. Rumours insinuate that Nvidia wanted a price cut from TSMC and that's why, they stated to produce their GPUs under Samsung 8nm Node but TSMC being TSMC didn't budge and Nvidia is paying for that blunder. Well, technically, consumers are facing the problems at the end of the day. (High power required, low performance per watt, lower yield due to chips being very large and other countless issues that comes when not calculating the after-effects.)

Samsung probably never had the capacity in the first place but I don't believe the supply situation would be relieved on more advanced nodes, yield issue obviously.