First NVIDIA Ampere mobile benchmarks

The new mobile Nvidia graphics cards are not faster than I thought.

https://www.notebookcheck.net/First...mpere-desktop-class-performance.514698.0.html

 

Makes you wonder unless cost is a consideration, what would be the selling point for upgrading to the new cards?

 

Pathetic gains over Turing, as expected. It’s too power hungry of an architecture, fitting for its name Ampere....needs a lot of current to make it happy. I wouldn’t call this a generational leap for mobile, not in the slightest. 2080M -> 3080M feels like 680M -> 780M all over again.

 

http://forum.notebookreview.com/thr...scale-on-laptops.834039/page-25#post-11066379

 

Your prediction of

3080M = desktop 2080S
3070M = desktop 2070S
3060M = desktop 2060S

was spot-on.

 

I’ve been around the mobile scene long enough to comfortably make my predictions

 

yeah but 2080 super desktop is nothing too scoff at thats still alot of power but not the 2080ti level everyone wanted.

and yes raidriar you were bang on super accurate and impressive rep


only good news is the 3060 is 41 percent faster than a 2060....thats impressive....looks like the 3060 unlike ever is the sweet spot

 

Welp there is always next year...

 

well 7 percent behind a 2080ti in a laptop is insane like me....and i think me and my new 3080 thin and light bga crapbook will be a good match lol

Welp there is always next year...
@Reciever I think you mean in two years...but yeah if your not happy with 2080ti in a laptop and already have a high refreshrate screen and 2060 and up its kinda completely pointless

also raidriar isnt the 3080 just as close to a 2080 super as it is to a twenty eighty ti....5 percent give or take 1 percent

 

From your point of view it probably is a big upgrade as you have 1060 right? if you already have a 2080 or 2080 super though it's defnitely not worth it at all.

 

yeah its a big upgrade for me! but for you guys with 2080s its unreal how crappy this news is 15 percent are you kidding me....so your getting 30 fps now gasp 34.5fps unreal....bga turdbooks are killing it

 

These aren't for Turing owners. Wait for Hopper.

 

*Lovelace

 

no owner of a 1080 and up should upgrade

 

Cyberpunk 2077 says otherwise lol.

 

Nah Pascal is old, has no current technological features.

The 2060 and up is good to wait.

 

isnt the 1080 better than the 2060? but yeah dlss is pretty nifty..

 

As far as I see, there's no legitimate reason Cyberpunk 2077 should be getting such poor performance (with raytracing off that is) on hardware that is only 2 generations old. A GTX 1060 should be enough to play any games being released this year at 60 fps on at least low or medium settings at 1080p.

Yes.

 

It's because Pascal isn't a modern architecture. It doesn't have dedicated integer cores, RT cores, or Tensor cores. I speculated on why Pascal is so much worse off in new games like Cyberpunk here.

Also look at Doom Eternal, where the 1080 Ti is beaten by 2060 Super:

 

Thanks, I completely glazed over that. The performance in Cyberpunk still feels way too low for the Turing and Ampere cards on that chart though (regarding the presets with no raytracing).

 

Regardless, Turing and Ampere have DLSS, so Pascal stands even less of a chance.

 

yeah dlss is kinda a big deal when dlss 4k looks better than native 4k that is 2.0

 

I look at the 1080 as something I will use for maybe 2018 and older titles, things like Monster Hunter or in general things I would play on a controller but thats only circumstance for it so it doesnt bother me me much about it starting to show its age

 

These benchmarks are way off and not final...perfect example is the Fire Strike score...3070 beats 3080? Yea right...

 

3080M beats 3070M in TS. FS probably has some Dynamic Throttle Boost 2.0 shenanigans.

 

Probably some alpha driver shenanigans. Real result for the 3080 mobile is likely to be 30-31K.

If I can score 26K with my RTX 2080, an 18% uptick is 31K.

 

Man, my mxm 1080 does 25k FS graphics on the daily oc. We just got to wait and see when some benchers from here get their hands on these cards, then we can judge whether it makes sense as an upgrade.

 

Firestrike is a bit long in the tooth now though and doesn't contain a lot of the modern rendering methods that Ampere and Turing are better at.

 

I tend to focus in on TimeSpy to get an idea of where GPUs stand for synthetic benchmarks.

 

RTX 3070/3080 vs RTX 2070/2080:
Performance Test: GeForce RTX 3070 Laptop & RTX 3080 Laptop - NotebookCheck.net Reviews

 

What a mess

 

yep, Nvidia made chaos the word of the day..

 

I guess as a 2080 laptop owner I can rejoice and not upgrade. I'll save $1500 from not having to upgrade I suppose. And that ^notebookcheck really solidifies this for me. I actually wasn't too impressed with the desktop 3080 performance gap considering the power consumption difference over the 2080. Currently have a gs75 stealth with 2080 max q. To add insult to injury your mobile 3080 will vary in TDP and might run slower than a 2080 max-q. I should of seen this coming with the power consumption of a desktop 3080 consuming 100 watts more than a 1080 Ti. There was just no way they were going to get even close to desktop level 3080 performance in a reasonably thin gaming laptop. It's nice though as I can keep my GS75 for another year or two and not have the upgrade itch.