Question 'Ampere'/Next-gen gaming uarch speculation thread

Page 125 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Ottonomous

Senior member
May 15, 2014
559
292
136
How much is the Samsung 7nm EUV process expected to provide in terms of gains?
How will the RTX components be scaled/developed?
Any major architectural enhancements expected?
Will VRAM be bumped to 16/12/12 for the top three?
Will there be further fragmentation in the lineup? (Keeping turing at cheaper prices, while offering 'beefed up RTX' options at the top?)
Will the top card be capable of >4K60, at least 90?
Would Nvidia ever consider an HBM implementation in the gaming lineup?
Will Nvidia introduce new proprietary technologies again?

Sorry if imprudent/uncalled for, just interested in the forum member's thoughts.
 

DisEnchantment

Golden Member
Mar 3, 2017
1,747
6,598
136
If it is really 7LPP it shows that Samsung can handle huge dies with EUV. (I don't know if they will call 8LPU+ 7nm but they just might)

Really looking forward to details. Density and performance.
Getting tired of Digitimes FUD about Samsung's 7nm.
Most likely Samsung 6LPP or even 5LPE would be available for NV 30 series refresh like the 12 FFN, if NV really use 7LPP.
Also for not so power constrained devices performance at a slightly lesser efficiency is very much tolerable.
Also TSMC's density figures comes with so much fine print for high performance devices that makes their numbers outside of smartphone SoCs not really meaningful.
 

Tup3x

Golden Member
Dec 31, 2016
1,086
1,085
136
If it is really 7LPP it shows that Samsung can handle huge dies with EUV. (I don't know if they will call 8LPU+ 7nm but they just might)

Really looking forward to details. Density and performance.
Getting tired of Digitimes FUD about Samsung's 7nm.
Most likely Samsung 6LPP or even 5LPE would be available for NV 30 series refresh like the 12 FFN, if NV really use 7LPP.
Also for not so power constrained devices performance at a slightly lesser efficiency is very much tolerable.
Also TSMC's density figures comes with so much fine print for high performance devices that makes their numbers outside of smartphone SoCs not really meaningful.
Also if Kopite's price comment is true, maybe they got better deal from Samsung and that helps to reduce the price. In any case, not a long wait now.
 

Glo.

Diamond Member
Apr 25, 2015
5,803
4,777
136
84 SM Ampere GPU - 40% Rasterization perf. improvement, over RTX 2080 Ti, 375W TGP - RTX 3090.
68 SM GPU - 10-15% rasterization improvement over RTX 2080 Ti, 320W TGP - RTX 3080.
48 SM GPU - 10% rasterization improvement over RTX 2080 Super - RTX 3070 Ti.
40 SM GPU - RTX 2080 performance - RTX 3070.
36 SM GPU - RTX 2070 Super performance - RTX 3060 Ti
30 SM GPU - RTX 2070 performance - RTX 3060
24 SM GPU - RTX 2060 performance - RTX 3050 Ti.
20 SM GPU - GTX 1660 Ti performance - RTX 3050.

RTX 3050 - 159$.
RTX 3050 Ti - 199$.
RTX 3060 - 249$
RTX 3060 Ti - 299$
RTX 3070 - 349$
RTX 3070 Ti - 449$
RTX 3080 - 599$.
RTX 3090 - 999$.

This is my prediction on price, and performance targets for Ampere GPUs.
I would now change the price of RTX 3080 from 599% to 699$ now...
And 3070 from 349 to 499$.
And 3060 fron 249 to 349...
And 3050 to 199$...
etc, etc...
 

DDH

Member
May 30, 2015
168
168
111

Looks like Nvidia won't go nuts with pricing this time around, as I expected (at least, aside from the 3090 monstrosity). :)
Good and expensive are subjective. I really doubt NVIDIA will price the 3090 lower than the 2080ti. And they will need to keep a slot there for the 3080ti. And he says pricing for ampere as a whole, so perhaps the 3060, and 3070 are less price gougy.

3060 - 399
3070 - 499 (FE+50)
3080 - 799 (FE+100)
3090 - 1299 (fe+200)

Performance prediction
3060 - 2070S + 5%
3070 - 2080S +10%
3080 - 2080ti +15%
3090 - 2080ti +45%
 
Last edited:

beginner99

Diamond Member
Jun 2, 2009
5,233
1,610
136
I was really considering going PS5 (still am, tbh)

Yea or xbox. In general console. Not paying $500 for a gpu I can get console with equal or better performance. So performance/$ should be mandated by expected console performance. So $499 for an 3070 will simply be way too much as actual price here will be even more than that.
 
  • Like
Reactions: kurosaki

maddie

Diamond Member
Jul 18, 2010
4,881
4,951
136
Good and expensive are relative terms.

He almost certainly means in relation to where we are now, so good is a decrease in price, at worse, they stay stagnant. Fantastic news for all of us as AMD will also have to be priced well.
 

jpiniero

Lifer
Oct 1, 2010
15,223
5,768
136
Videocardz got a hold of an Asus PR about their cards, says "up to 1.9X performance-per-watt over the previous generation". Not sure I buy that, maybe close in RT?
 

NTMBK

Lifer
Nov 14, 2011
10,324
5,360
136
Nearly 400w dumped in the case is big trouble. Might be the generation where reference card is smarter choice than 3rd party.

IF NV cooling is exhausting large percentage of the heat via backplate, and does so at reasonable noise levels i don't really care about power as long as performance lead is there.

What i am worried is 3rd party cards, that are "cool" by virtue of having 3 large fans and massive heatsinks, but dump 100% of heat in the case...

Time to break out the old Fermi memes...

image.png.3de8d258596d8b9c3286062902c1e141.png
 

ozzy702

Golden Member
Nov 1, 2011
1,151
530
136
Yes, upping the performance per watt with 1.9x AND adding 100W might give us a good ride. Sounds too good to be true though.

Agreed. It has to be 1.9x in some very specific situation like RT or DLSS. Outside of that I'd be happy with a 50% bump across the stack. We'll see...
 

kurosaki

Senior member
Feb 7, 2019
258
250
86
I believe that the 1.9 has been already mentioned previously in this thread, as 1.9 with RTX and DLSS.
I'm a bit saddened. "With DLSS" dosent tell us anything useful. Are they upscaling from 540p? of course they could show figures like this then. I feel they have found a loophole to circumvent the need to tell us in which resolution they are measuring things. Hope that Ian can poke a bit in this issue in an upcoming review.