How will Ampere scale on laptops?

I don't think thunderbolt is the way to go here, at least not currently. M.2 eGPUs yield much better performance, and I believe they get very close to native performance in a desktop. An external m.2 port would definitely be interesting.

 

Theoretically TB4 should be able to get pretty close to the m.2 egpus. The problem with those are they are a pain in the rear to install seamlessly and they take up a precious m.2 slot.

 

What's usually the latency for TB 3.0 eGPUs????

Well M.2 ports still use the DMI 3.0 which is x4 PCH PCIe 3.0 lanes, there shouldn't be much of a difference with TB eGPUs, the advantage should be that M.2 x4 PCIe 4.0 is much closer in time than TB 5.0, though I've never seen a proper way to get the M.2 port physically out of a laptop plus I don't think there is a way to re-route the video signal to the laptop's internal screen

 

This is true, if you want to use the internal panel you have to go tb I think. TB3 is slightly restricted in bandwidth with the 22Gb/s limitation. It works fine, and there is an extra penalty for using the internal panel, more notable at low resolutions like 1080p vs using a monitor hooked to the gpu, but yeah..

 

Shame to see the Nvidia dropped the laptop vga = desktop vga with lower core clock concept .
I guess this was inevitable and just lowering core clock to reach 150-200W tdp limit was not enough.

Oh well, even with 6144 cuda cores the 3080 laptop will easily beat the 2080Super laptop card.

 

They could have, there are many users who have undervolted their desktop RTX 3080 to 200w and still retain 90% performance. Theres even a guy who undervolted to 144w so its very much possible, nvidia just didn't bother and took the more profitable route.

 

We're just a couple of hours away from the rtx 30 announcement. I guess it's finally time for the moment of truth!

 

im putting in my chips.

it will be just barely faster than a 2080ti
(220) RTX 3080 Mobile benchmarks: faster than an RTX 2080Ti? - YouTube
(220) Laptop RTX 3080 Max-Q Benchmark Reveals +58% Gain vs RTX 2080 Max-Q Laptop (Leaked) - YouTube

and why do i say that? nvidia suprised all of us with pascal and they can do it again

Also those videos are based on the opencl benchmark leaks so nothing new just what people are saying as a whole, but then again the masses are not always right

And just imagine doing this!! 4k maxed re run of witcher 3 averaging 73fps

(220) The Witcher 3 Maximum Settings 4K | RTX 2080 Ti | i7 8700K 5.3GHz - YouTube
NVIDIA GeForce RTX 3070 video card benchmark result - AMD Ryzen 7 5800X,Gigabyte Technology Co., Ltd. B550 AORUS ELITE (3dmark.com)
2080=133fps
3080=174fps

not a huge jump but enough to justify an upgrade if you have the cash and want to play cyberpunk2077 at 60 fps locked vs 50

 

Absolutely, having RTX 2080ti level performance in a portable chassis is good enough for me to stick with for this gen of games-Especially since the Xbox Series S with its RT capable ~4tflop GPU will be the baseline for developers.

 

Are you getting an upgrade worthy of dropping a wad of cash is the pointy end of the stick question. Could have, should have, doesn't speak to that although it's always fun to slap around corporate types for their high level of scumbaggery. If you're like me and running a 1080 it MIGHT be a buy even if it isn't what I had hoped for back when I bought the current lappy. I'm no question paying attention...

 

your 1080 is still epic power for 1080p, however if your making the jump to 1440p/4k its justified.....i have a oc 1060 and dont mind 30/60fps but if i want to hook it up to a 4k tv im outta luck unless i drop all settings to low and do so on a pre 2014 game

 

Well if Nvidia's CES advertisement was to be believed it looks pretty decent. Too bad they didn't have any actual information.

 

My lappy is 2k 120 refresh native; the 1080 is well suited to the panel. In my mind this needs to bring a legit 4k machine to the table to attain no brainier status. Less than that is going to be a difficult call.

 

This bodes well for maybe 200w class cards in Alienware and potentially Clevo models:

 

Bam..
They were too focused on DLSS and max-q, although dynamic boost 2.0 looked promising.
Not a word on the specs but a 3060 performing 30% better than a ps5, 3070 outdoing a 2070 by 50%..well its a start.

 

I thought there was a 3050 budget mobile/regular card? Maybe later on. I don't believe you'll get a 3060 laptop for $1k though sounds preposterous. Almost double the cuda cores from a 2060 wow.

https://videocardz.com/newz/nvidia-...0-rtx-3070-and-rtx-3060-mobile-graphics-cards

 

Double the CUDA cores doesn’t mean double the performance. On Ampere, Nvidia is counting the hybrid FP/INT cores as CUDA cores too. The true number of dedicated FP-only CUDA cores is half.

 

Plus there would need to be enough power to actually use those cores at full capacity...

 

Ah, well another example of just changing the meaning of names to get away with lying ugh.

 

This is the route I think Nvidia should've taken, even on desktops. I know desktops aren't power limited like laptops and that we want maximum performance from desktops, but I don't fancy a 60% increase in power draw for only a 10% improvement in performance, even on desktops. We have to pay for electricity, and I don't want an expensive power bill from a GPU operating at a very inefficient point. For those who wish to extract every possible ounce of performance from the GPU, they can still do that.

 

Notebookcheck updated the info of the RTX 30 series from today's event and how it compares the RTX 20 Series.

https://www.notebookcheck.net/NVIDIA-GeForce-RTX-3080-Mobile-GPU-Benchmarks-and-Specs.497450.0.html
https://www.notebookcheck.net/NVIDIA-GeForce-RTX-3070-Mobile-GPU-Benchmarks-and-Specs.497451.0.html
https://www.notebookcheck.net/NVIDIA-GeForce-RTX-3060-Mobile-GPU-Benchmarks-and-Specs.497453.0.html
https://www.notebookcheck.net/NVIDIA-GeForce-RTX-3050-Mobile-GPU-Benchmarks-and-Specs.513790.0.html

RTX 3080 Max-P = RTX 3070 (Desktop)
RTX 3070 Max-P = RTX 2080 Super Max-P
RTX 3060 Max-P = RTX 2070 Super Max-P
RTX 3050 Max-P = RTX 2060 Max-P

 

I'm really hoping to see the next-gen Dell XPS 15 with either Tiger Lake H or Ryzen 5000U and an RTX 3050 or 3060.

 

Yayy external Pcie ports! That's something I've been wanting to see in laptops for a while.

The only part I don't like is that the GPU enclosure is proprietary. This would have been better had the enclosure been capable of housing off the shelf desktop GPUs.

 

That's exciting but it looks like it would be loud with that footprint. But I really like what they did there, and it looks like it's also a mobile gpu no doubt. It would be interesting to see them create an egpu that could sport a desktop card but I highly doubt they will go that route. It's niche enough already. Can't wait to see a real review later this month.

 

Pretty meh tbh. The 3080 Mobile in their pre-sample unit was 50% faster than my 1080N in Time Spy Graphics and only 15% faster in Fire Strike Graphics. Hopefully that's just a Max-Q version.

 

Looks like on desktop, 3060 is only about 10-15% faster than 2060 when using light or no RT.

 

i fold....but its fairly close to a 2080ti...in between a 2080 super and 2080ti

i should have known the masses go figure but nbr members are usually bang on

 

Clicked around for a 3080 performance laptop but all I can find are the Q variants. The price on those is similar to the previous gen 2080s, gotta figure they'll be more proud of the 16 gig part but I wonder exactly how proud? A lot of the usual suspects like Sager didn't even show a 30 series yet but I noticed the Gigabyte stuff was supposed to be shipping early February on HIDs site. Anyone even know if the 16 gig piece is going to be available at the same time as the Meh Q part?

 

Apparently not

https://www.notebookcheck.net/NVIDIA-GeForce-RTX-3080-Max-Q-GPU-Benchmarks-and-Specs.513789.0.html

The Max-Q version is around the same performance as the RTX 3070 Max-P

I need to see real world performance rather then just benchmarks

 

What's weird is the 3080 Mobile is slower than the 3070 Mobile in Fire Strike Graphics, which is why I thought the 3080 was Max-Q.

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

 

Aren't all these benchmarks based on just engineering samples?

 

Yes

 

The new scar has a 3080 with 16 gigs which should be the up tempo version right? There are a few first looks at it popping up on youtube.

 

I don't trust the scores at all. The 3080 mobile scores being faster and slower than the 3070 mobile raises a red flag immediately.

Its Max-p

I don't even think these are actual numbers, looks more like speculation to me. Wait for actual reviews/benchmarks

 

I'm just curious as to what these usually cost at new on release? The cheapest 2080 laptop I could find was about 2700$, and a 2070mq at 1250$. I'm guessing well into the $3k range for 3080?

 

ASUS 15.6" Republic of Gamers Strix SCAR 15
MSI GP66 Leopard 10UH-275 - 15.6" 240 Hz IPS - Intel Core i7-10870H - NVIDIA GeForce RTX 3080 Laptop GPU 8 GB GDDR6 - 16 GB Memory - 1 TB NVMe SSD - Windows 10 Home - Gaming Laptop - Newegg.com
MSI GS66 15.6" 300hz Gaming Laptop Intel Core i7 NVIDIA GeForce RTX 3080 Max Q 1TB SSD 32GB Memory Black GS66254 - Best Buy

 

Wow so there's really no price difference from a 2080 laptop at current best buy prices I saw. I won't hold my breath for much of a clearance firesale on 2080 laptops. Much rather have that ryzen/1440p for the same price lol All of this is way way out of my league though lol

 

That would entirely defeat the portability aspect of what Asus is offering.


Sent from my iPhone using Tapatalk

 

Interesting that mobile 3060 has more CUDA cores than desktop 3060, 3840 vs. 3584, although clocked lower (due to lower power limit) and half the VRAM.





Seems that 3060M with 115W TDP (up to 130W with Dynamic Boost 2.0) will be the closest performer to its desktop namesake.

 

(220) These RTX 3080 Gaming Laptops will Blow Your Mind! (40% FASTER!?) | The Tech Chap - YouTube

 

All i want is a MAX-P that works with a modded BIOS in my P775DM3-G.

Screw MAX-Q

 

I'm hopeful they'll perform similarly. Despite the reduced TDP, the extra cores should make up the difference. I don't think the increased VRAM on the desktop variant will really do anything since the 3060 is not a card aimed at 4K. I was expecting the RTX 3060 to only come in a 6 GB variant, both for laptops and desktops.

 

It's not the first time Nvidia has given more CUDA cores to the mobile GPU compared to the desktop ones. Off the top of my head, the mobile GTX 1070 and GTX 1650 have more CUDA cores than their desktop counterparts, but they run at lower core speeds.

 

Also the mobile GTX 1060 3GB.

 

How likely do you think this is? If they push an update for Clevo x170sm-g then the card is going to be on the same form factor as the 20 super and compatible with p775 with a heatsink change.

 

lol, the Nvidia presentation was hilarious... that video of a guy plugin in 3 screen to his laptop and playing FS2020 at like 140 fps, oh Nvidia you always bring joy to my heart. So leaked specs were 100% accurate nothing new today, probably the new and most interesting part was that video of a board where the CPU and GPU and VRAM are like communicating to be more efficient, but that should be available for BGA models only? I'm not sure.

Performance wise, I'm still expecting the 3080 max-p to perform in games similar to a desktop 3060 Ti and the 3080 max-q to perform similar to a potato.

 

Hopefully I'll be able to purchase one off Ebay. We'll see. But I believe it's a worthy upgrade from a GTX 1080 (the MAX-P)

 

I can't find any reviews with real world numbers despite youtubers having these in hand so I'm assuming we're under embargo; anyone confirm, deny, know when it lifts?

 

Resizable bar? Nah that's just part of the PCI spec. Be interesting to see if that comes to turing though since it technically supports it.