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

Page 4 - 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
293
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.
 

Stuka87

Diamond Member
Dec 10, 2010
6,240
2,559
136
NVIDIA Will Be Announcing 7nm Ampere GPUs At GTC 2020 In March


looks like pro cards in March and consumer cards in June/July.

Being that Ampere is an HPC card, it will be for a new Tesla card. A consumer version will be quite a ways down the road. The article itself mentions 2H 2020 as a potential launch time.
 

jpiniero

Lifer
Oct 1, 2010
16,979
7,382
136
Being that Ampere is an HPC card, it will be for a new Tesla card. A consumer version will be quite a ways down the road. The article itself mentions 2H 2020 as a potential launch time.

WTFTech probably has no idea on an actual release date. The Pascal Tesla was released on April 5, 2016 and the first consumer card (1080) was released on May 27th.
 
  • Haha
Reactions: Saylick and maddie

GodisanAtheist

Diamond Member
Nov 16, 2006
8,398
9,801
136
A node shrink hopefully means better price to perf but the massive Turing dies makes me wonder if we'll even see that this time around...
 

Stuka87

Diamond Member
Dec 10, 2010
6,240
2,559
136
A node shrink hopefully means better price to perf but the massive Turing dies makes me wonder if we'll even see that this time around...

Node shrinks will never ever mean lower prices. Way back in the day it did result in lower prices. But as the cost for a node keeps going up, it basically is a wash.
 

GodisanAtheist

Diamond Member
Nov 16, 2006
8,398
9,801
136
Node shrinks will never ever mean lower prices. Way back in the day it did result in lower prices. But as the cost for a node keeps going up, it basically is a wash.


- But the last node shrink we had (28nm -16nm) did result in massively better price to performance. Even after the first mining boom and bust.

NV has just designed massive dies on the pseudo 12nm (really refines 16nm) process with a lot of R&D cost tacked on for RTX. AMD decided to price up as they were never going to go full volume production of GPUs when they needed wafer allocation for the far more lucrative CPU side of things. Why sell for $250 what people will buy for $450 dollars?

But without Price/Performance increases, even NV will have trouble moving new inventory to existing customers if they can't offer each of their price brackets a reason to upgrade. With Turing, they figured RTX would do it, but that's clearly not the case.
 
  • Like
Reactions: psolord

Glo.

Diamond Member
Apr 25, 2015
5,930
4,991
136
Nope, Ampere is mid 2020 at the most.
You know perfectly well that only GA chip that exists is GA100. And that is HPC chip.

Unless you want to say that Volta replacement with similar die size, and HBM2 will end up in consumer market. Then yes, you are correct.
 

Muhammed

Senior member
Jul 8, 2009
453
199
116
You know perfectly well that only GA chip that exists is GA100. And that is HPC chip.

Unless you want to say that Volta replacement with similar die size, and HBM2 will end up in consumer market. Then yes, you are correct.
EVERY Ampere chip will launch in 1H of 2020, 3080Ti will be delayed a little to 2H.
 

happy medium

Lifer
Jun 8, 2003
14,387
480
126
I say this every launch for the past 15 years.
It is not hard to figure out what Nvidia's doing after they announce their pro chips in March.
You can easily guess how fast the 3080ti,3080 and 3070 will be and you know consumer cards usually come about 2 or 3 months after.
It's as simple as that.

The only thing that will delay the launch is if AMD holds back big Navi.
Once big Navi is released the new Nvidia line will just crush it.
That's been Nvidia's strategy for years now.
 

maddie

Diamond Member
Jul 18, 2010
5,178
5,576
136
What I'm more interested in is where this will be produced and on what node.

What we know:
TSMC 7nm nodes are in high demand with AMD planning to double wafer orders with the upcoming expansion of capacity and the upcoming consoles taking even more.

Where is the available wafer volume at TSMC for Nvidia to keep their market share?
If not at Nvidia, are Samsung nodes mature enough for them to launch in volume mid-year as they will have to start production within the next month?
 

Stuka87

Diamond Member
Dec 10, 2010
6,240
2,559
136
I say this every launch for the past 15 years.
It is not hard to figure out what Nvidia's doing after they announce their pro chips in March.
You can easily guess how fast the 3080ti,3080 and 3070 will be and you know consumer cards usually come about 2 or 3 months after.
It's as simple as that.

The only thing that will delay the launch is if AMD holds back big Navi.
Once big Navi is released the new Nvidia line will just crush it.
That's been Nvidia's strategy for years now.

The post that I quoted was told in a matter of fact, this is the truth manner. Forum guidelines state that rumors cannot be posted as facts. Especially when the poster in question has been banned for such things.
 

Glo.

Diamond Member
Apr 25, 2015
5,930
4,991
136
EVERY Ampere chip will launch in 1H of 2020, 3080Ti will be delayed a little to 2H.
There is ONLY ONE AMPERE GPU.

One. GA100. Gaming, consumer GPUs are not Ampere. Ampere is HPC.

And gaming, consumer GPUs, of next generation will not come out before Q4 2020/Q1 2021. Period.

Ampere is ONLY for HPC. Put that in your minds guys. People who know better than you, who have seen deviceIDs say that Ampere "lineup" is constructed from only one GPU chip.

Why Q4 2020/Q1 2021?

Because Samsung hasn't solved their process, yet, and at TSMC they will not have enough 7 nm capacity to move enough GPUs for consumer AND HPC market, especially considering the die size of GA100.

All of 7nm capacity available for Nvidia in 2020, be it SS or TSMC, will be reserved for HPC chips. The end. There is a very good reason why Nvidia is readying RTX 2080 Ti Super, with full die, and fastest possible GDDR6 memory for Q2-Q3 2020. Because that is the best they will be able to offer in 2020.

And last thing. Next gen gaming GPUs are not named Ampere. Remember this.
 
Last edited:

maddie

Diamond Member
Jul 18, 2010
5,178
5,576
136
The post that I quoted was told in a matter of fact, this is the truth manner. Forum guidelines state that rumors cannot be posted as facts. Especially when the poster in question has been banned for such things.
It's not even as good as a rumor, but a "I have faith that 'XYZ' will happen"
 

happy medium

Lifer
Jun 8, 2003
14,387
480
126
What I'm more interested in is where this will be produced and on what node.

What we know:
TSMC 7nm nodes are in high demand with AMD planning to double wafer orders with the upcoming expansion of capacity and the upcoming consoles taking even more.

Where is the available wafer volume at TSMC for Nvidia to keep their market share?
If not at Nvidia, are Samsung nodes mature enough for them to launch in volume mid-year as they will have to start production within the next month?
Samsung EUV 7nm . This process will be more efficient than even 7nm TSMC.
I'm guessing the 3080 will be at least 20% faster than a gtx2080ti.
Save my posts if you like.
 
  • Haha
Reactions: Krteq

maddie

Diamond Member
Jul 18, 2010
5,178
5,576
136
Samsung EUV 7nm . This process will be more efficient than even 7nm TSMC.
I'm guessing the 3080 will be at least 20% faster than a gtx2080ti.
Save my posts if you like.
What time frame is the crystal ball saying? Also, will Nvidia be producing anything at TSMC?