Consumer 7nm GPUs from AMD are late 2019 to early 2020

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

happy medium

Lifer
Jun 8, 2003
14,387
480
126
My opinion, they scrapped there current 2018/2019 gpu's and are working on something to counter NVIDIA'S new features and performance.

Nvidia Will have time to counter with 7nm Ampere.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,483
14,434
136
My opinion, they scrapped there current 2018/2019 gpu's and are working on something to counter NVIDIA'S new features and performance.

Nvidia Will have time to counter with 7nm Ampere.
And where did you hear about this ? I think you are trolling here with this mention of "scrapping". Give us even a hint of proof.
 

piesquared

Golden Member
Oct 16, 2006
1,651
473
136
Oh how I wish AMD had been at all serious about 12nm dGPUs. It would give them a consumer filler product to put on the market now, so the cavernous void between RX Vega and Navi could be filled by something, anything.

It's stunning that AMD has allowed their consumer lineup an over-2-year gap between the launch of Vega last year, and Navi in 2019. Absolutely stunning. Talk about a wasted opportunity.

Hell if there's an HBM2 shortage, they probably could have arranged to use GF 12nm as a second source for that too. They're moving all their CPU production off 12nm anyway.



You would think the various APU releases would be excellent test-cases in fixing ngg path. Looks like the first Vega with fixed ngg path will be Vega10, which is I guess the 7nm Vega refresh?

Also, here's hoping for 7nm Vega Frontier Edition 2 (or whatever).

Clearly weighing their R&D in favor of the development of Zen and their compute roadmap wasn't a wasted opportunity. Why move to a stop gap 12nm node to gain 10% or 20% performance, wasting valuable resources when they are building products on 7nm. Even from a consumer standpoint, how would that benefit them?? These decisions to divert resources to a 12nm GPU would have had to be made at the time they were designing Zen and that seems like it would not be a wise decision.

The Vega architecture is quite strong for compute and beats Nvidia's latest in many cases. Focusing on the high profit compute and ML sector is the right business decision unquestionably. Look at the impact Zen is having on the market. With Zen 2 on 7nm due relatively shortly they are set to win a lot of market share in the very high margin server market. AMD do have gaming cards out there, they may not have the top spot but they are definitely capable. I use one and game on it all the time and am perfectly satisfied with it.

The HBCC does work and works quite well. Shortly after Vega launch, there was a demonstration of a user using it to address 256 GB of system memory on his FE, and it is designed to address 2 TB. So it works as intended. The 7nm Vega in the upcoming Radeon Instinct line should have a substantial compute performance increase. This will give more developers incentive to adopt the ROCm platform. I hope we see a 7nm Vega Frontier Edition also, but I have my doubts. It doesn't appear that it would be gamer focused and it would have to be priced prohibitively high initially so I don't think their would be any point in releasing it into the consumer market if it isn't number one in games, since that is the criteria the internet will judge it on. Of course, 7nm will allow them to increase the core count quite a bit so it is possible that gaming performance has improved significantly also. That would make a beast of a Vega FE, so here's hoping.

Maybe developers chose not to use the ngg fast patch and therefore AMD pulled resources. No point in throwing money into something if developers decide not to use it. I'm curious, how is it broken?
I'm not aware of GF developing HBM2 technology but I might have missed that. I thought just Samsung and Hynix were producing HBM modules.
 
  • Like
Reactions: ryan20fun and Ajay

happy medium

Lifer
Jun 8, 2003
14,387
480
126
And where did you hear about this ? I think you are trolling here with this mention of "scrapping". Give us even a hint of proof.
I dont ever remember Nvidia or AMD taking 2 or 3 years or mabe more ,to release a GPU.
If they release a new GPU in October of next year that means the rx480 will be on the market for 3 1/2 years and the 2 Vega cards over 2 years.
I dont think AMD planned is this way.
Thats where I form my OPINION, and I did say opinion in my other post.
But that's what I believe happened.

Or

Mabe they are having trouble with 7nm gaming gpu's?

If they dont have a GPU till 2020, what else could it be?
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,483
14,434
136
I dont ever remember Nvidia or AMD taking 2 or 3 years or mabe more ,to release a GPU.
If they release a new GPU in October of next year that means the rx480 will be on the market for 3 1/2 years and the 2 Vega cards over 2 years.
I dont think AMD planned is this way.
Thats where I form my OPINION, and I did say opinion in my other post.
But that's what I believe happened.

Or

Mabe they are having trouble with 7nm gaming gpu's?

If they dont have a GPU till 2020, what else could it be?
Maybe AMD has been too busy WIPING THE FLOOR OF INTEL WITH RYZEN AND EPYC to bother with Nvidia.

Just my OPINION
 

GodisanAtheist

Diamond Member
Nov 16, 2006
6,719
7,016
136
So long as AMD shows up with something great at reasonable price points, this timeline is not concerning.

Nvidia really controls this market and it's cadence at this point. Having something now, or early next year is meaningless if it's just going to slam into that NV brand wall because it's too hot or power hungry or short on performance.

If they end up releasing another Polaris tier card that delivers Vega 56-64 performance for $300 bucks then it's a problem.

I will likely be close to a card pruchase at that point so I sincerely hope they give me a reason to upgrade off my 980ti (such a good card, it might last me a while longer than my usual 3 year upgrade cycle).
 

Qwertilot

Golden Member
Nov 28, 2013
1,604
257
126
He's assuming AMD are still on a regular upgrade cycle, which frankly they fell off around Tonga sometime.

Their next generation consumer parts most likely in sync with next gen consoles for obvious reasons.

Compute stuff they can push a bit as such a huge market.
 

happy medium

Lifer
Jun 8, 2003
14,387
480
126
I posted before that I'm expecting them to target holiday '19 (i.e. Nov - mid Dec) for Navi. Could slip to early 2020 though.

Do you think Navi will have Ray tracing features or some type of DSSL to compete with or do you think Navi was finalized too late to implement it in time.

I think Navi will need to be a MONSTER of a gpu to compete with a gtx2080ti late next year when I'm sure plenty of games will be using DSSL. Who knows there might be a 7nm Nvidia card by then.
7nm Navi would need to be like 70% faster than a Vega 64 to compete with a 12 nm 2080ti without DSSL.

It might bring prices down, thats a plus.
 

beginner99

Diamond Member
Jun 2, 2009
5,208
1,580
136
Maybe AMD has been too busy WIPING THE FLOOR OF INTEL WITH RYZEN AND EPYC to bother with Nvidia.

Exactly. And realistically in current state intel is the easier target mainly because of software (CUDA). In the deep learning space? You are forced to use NV unless you are a masochist. And on consumers side driver features and upgrades also consume resources.

With CPU / x86 AMD has no software hurdle to overcome plus die sizes tend to be smaller especially with MCM and hence higher yields = higher margin = more profit.

With limited cash, it was certainly a good decision.
 

coercitiv

Diamond Member
Jan 24, 2014
6,151
11,686
136
I dont ever remember Nvidia or AMD taking 2 or 3 years or mabe more ,to release a GPU.
If they release a new GPU in October of next year that means the rx480 will be on the market for 3 1/2 years and the 2 Vega cards over 2 years.
The GTX 1060 (GP106) is what we call the mainstream chip and has been around for over 2 years, with no replacement in sight for 2018. It may actually end up being a 3 year old GPU before it gets replaced.
 
  • Like
Reactions: crisium

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
Well there is no replacement for any Pascal bellow GTX1080 for 2018. Dont say RTX2070 because those will be at higher price points.
So both AMD and NVIDIA will have dGPUs up to $500-550 for close to 3 years old at the end of 2018.
 

DrMrLordX

Lifer
Apr 27, 2000
21,583
10,785
136
Clearly weighing their R&D in favor of the development of Zen and their compute roadmap wasn't a wasted opportunity. Why move to a stop gap 12nm node to gain 10% or 20% performance, wasting valuable resources when they are building products on 7nm. Even from a consumer standpoint, how would that benefit them?? These decisions to divert resources to a 12nm GPU would have had to be made at the time they were designing Zen and that seems like it would not be a wise decision.

You're assuming that producing a refresh card on 12nm would have cost them too much money. An optical shrink or near-shrink with tweaks would have helped tame Vega's heavy power usage near "normal" clockspeeds (1585 MHz).

AMD is selling nearly every video card they produce right now. Yes, they can keep schlepping Vega and Polaris for another year, or they can up market share a bit with 12nm Vega (and maybe even 12nm Polaris). Either way they are going to sell whatever they push out to market. The R&D expenses would likely pay for themselves. Not that we'll ever know, since AMD chose not to go that route.

The Vega architecture is quite strong for compute and beats Nvidia's latest in many cases.

So? What does that have to do with 12nm consumer cards? We already knew AMD was going to go for 7nm Vega for compute, so there was no need to address that market with 12nm parts.

Focusing on the high profit compute and ML sector is the right business decision unquestionably.

I am reasonably confident that AMD has/had the resources necessary to prepare a 12nm consumer Vega and 7nm professional Vega, at the same time. They simply skipped 12nm altogether. AMD bailed out of the consumer market. I question whether or not they will ever seriously return to it.

Look at the impact Zen is having on the market.

Irrelevant to the topic at hand. Zen's success does not preclude AMD from continuing to serve the consumer dGPU market. At. All. They maintained an entirely separate RTG division under Raja Koduri while developing Zen in the first place. Continued work on Zen+ and Zen2 could have continued while RTG refined Vega and moved it to GF's 12nm process, which according to many is not much more than a refined 14nm process anyway . . .

AMD do have gaming cards out there, they may not have the top spot but they are definitely capable. I use one and game on it all the time and am perfectly satisfied with it.

It is not rational for them to continue selling the same consumer dGPUs for over two years.

I hope we see a 7nm Vega Frontier Edition also, but I have my doubts. It doesn't appear that it would be gamer focused and it would have to be priced prohibitively high initially so I don't think their would be any point in releasing it into the consumer market if it isn't number one in games, since that is the criteria the internet will judge it on. Of course, 7nm will allow them to increase the core count quite a bit so it is possible that gaming performance has improved significantly also. That would make a beast of a Vega FE, so here's hoping.

Vega FE wasn't #1 in games when it came out last year, but AMD launched it anyway. What's stopping them now?

Maybe developers chose not to use the ngg fast patch and therefore AMD pulled resources. No point in throwing money into something if developers decide not to use it. I'm curious, how is it broken?

The feature is non-functional in Vega9, period. Whether or not developers intended to support it is another matter altogether.

I'm not aware of GF developing HBM2 technology but I might have missed that. I thought just Samsung and Hynix were producing HBM modules.

AMD could likely have arranged production of HBM2 modules at GF in the event of supply shortages from Samsung and Hynix. Though I am not sure what process Samsung used for production of HBM2 last year for Vega.

He's assuming AMD are still on a regular upgrade cycle, which frankly they fell off around Tonga sometime.

AMD still managed to deliver fairly regular releases from Fiji to Polaris to Vega. Polaris obviously did not replace Fiji, but it was a new uarch that was a significant upgrade for their midrange lineup. And obviously Vega replaced Fiji. 2018 was the right time for either a 12nm Vega or small Vega, or a 12nm Polaris update at least . . . something, anything.

Well there is no replacement for any Pascal bellow GTX1080 for 2018. Dont say RTX2070 because those will be at higher price points.
So both AMD and NVIDIA will have dGPUs up to $500-550 for close to 3 years old at the end of 2018.

Nvidia is under very little pressure to update the 1070 or 1080. AMD is the only party that could apply that pressure, and obviously they are doing nothing this year.
 

Paul98

Diamond Member
Jan 31, 2010
3,732
199
106
Considering some of the major features that were supposed to be part of Vega don't work... Moving forward things should get better next gen. As hopefully they will learn from their mistake
 

Kenmitch

Diamond Member
Oct 10, 1999
8,505
2,248
136
Do you think Navi will have Ray tracing features or some type of DSSL to compete with or do you think Navi was finalized too late to implement it in time.

I think Navi will need to be a MONSTER of a gpu to compete with a gtx2080ti late next year when I'm sure plenty of games will be using DSSL. Who knows there might be a 7nm Nvidia card by then.
7nm Navi would need to be like 70% faster than a Vega 64 to compete with a 12 nm 2080ti without DSSL.

It might bring prices down, thats a plus.

That's one of the underlying issues whey Lisu Su decided to just give the gpu arena the middle finger for now. Why waste the the time and effort on something in the end that for the most part is just to keep the others guys pricing in check? You can bury your head in the sand if you like but you know how it works. Too little, too late, can barely match a XXXX, and the list goes on. The never ending shifting of goal posts to justify purchasing the other guys product no matter what anyways. Wanting competition for no other reason than getting a better deal is what got the gpu market in it's current predicament. Yes there are those who buy and would buy AMD's offering but sadly they're arsed out because of the majority.

AMD has tried fighting the battle on 2 fronts for a decade and it didn't work out so good. They survived and have chosen to focus on one battle at a time. Looks like the best strategy in the current market and the most profitable for them.

It's probably a coincidence but it looks like Zen pretty much landed just as Intel hit it's wall of despair. Future Zen products will only bring on more hurt for Intel it looks like. Look at the bright side you guys get to still reward Intel with your $$$'s when you find the silly metric that keeps you going back....At least you get more for less which is all that matters to you anyways.

Sit back and wait while allowing the gpu buyers to willingly allow nvidia to jack up pricing looks like the current strategy. When the time is right I expect AMD to strike like a cobra....It's just a matter of time.

My opinion is....Your opinions are of no real value too me in the end. That's the safe to post here version.

I'm also not dumb and can see what your really looking for in the end.
 

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
Nvidia is under very little pressure to update the 1070 or 1080. AMD is the only party that could apply that pressure, and obviously they are doing nothing this year.

AMD has cards that are competitive Up to GTX1080 territory, they are also not in a rush to launch anything new and spend resources they would other wise put for 7nm cards.
So both are sitting comfortable on their ass waiting to see how consumers will react to new high prices of the RTX cards , gathering data for the new 7nm cards that will follow the next 12-18 months.
 
Last edited:
  • Like
Reactions: beginner99

jpiniero

Lifer
Oct 1, 2010
14,511
5,159
136
If they end up releasing another Polaris tier card that delivers Vega 56-64 performance for $300 bucks then it's a problem.

That's pretty much what the rumors are on Navi 10. Might even be cheaper.

Vega 20 btw is rumored to be same core count (4096) but includes hardware changes for 1/2 DP plus probably some sort of Deep Learning functionality and allows for quad HBM2 stacks. And of course any clock speed improvement they manage to get out of 7 nm.
 

DrMrLordX

Lifer
Apr 27, 2000
21,583
10,785
136
AMD has cards that are competitive Up to GTX1080 territory,

They do, but look at who is selling more cards in the consumer space right now.

AMD's counter to the 1080 right now is a $500 card that has had availability problems since launch until recently. The counter to the 1070 and 1070Ti is, what, a $360 card?

What do you suppose happens if AMD comes out swinging with a 12nm card for $300 or less that can challenge the 1070Ti and/or a card in the $400 range that can challenge the 1080? NV changes their prices, updates their lineup, or suffers the consequences. Not that we'll ever know about that now, since it isn't going to happen . . .

they are also not in a rush to launch anything new and spend resources they would other wise put for 7nm cards.

No idea why 12nm cards would really hurt their 7nm compute launch this year. At worst, AMD spends a comparatively trivial amount on R&D moving something Vega-based to 12nm and gains 0 market share because NV reacts by slashing prices. Maybe that is what AMD feared the most? Nvidia has a lot of leeway in how they can counter AMD's product refreshes. NV's prices are high and their performance is good enough that they can keep on trucking after just dropping prices some.
 

GodisanAtheist

Diamond Member
Nov 16, 2006
6,719
7,016
136
That's pretty much what the rumors are on Navi 10. Might even be cheaper.

Vega 20 btw is rumored to be same core count (4096) but includes hardware changes for 1/2 DP plus probably some sort of Deep Learning functionality and allows for quad HBM2 stacks. And of course any clock speed improvement they manage to get out of 7 nm.

-Well no one is going to accuse AMD of not stretching their dollar on GCN, if that's the case!

I do recall an old AMD marketing slide where they stated Navi was bringing "scalability". After so many iterations of the Fury core config l, I have to assume they mean reworking their graphics processing engines to go above 4 (others took an MCM Ryzen type interpretation out of it).

Lisa Sure stated she wasn't content being the second player in the graphics market, which means bringing something, top notch performance for example, to the market.

If all they plan on competing on is price, then they deserve the value brand label.

This is all based on there hearsay/rumor that Navi is going to be a floater (and after the hype bullet train that was Polaris and Vega, RTG could stand a anti-hype train).
 

beginner99

Diamond Member
Jun 2, 2009
5,208
1,580
136
So both are sitting comfortable on their ass waiting to see how consumers will react to new high prices of the RTX cards , gathering data for the new 7nm cards that will follow the next 12-18 months.

I'm gonna say it's pretty predictable. Decreasing sales. There are only so many who can spend >$600 and the rest already mostly bought into polaris / 1000-series. Yes there will always be some that upgrade or need a new card but in general sales will go down as for a large part of the market there is nothing new and affordable available.

Maybe we see a small surge of 1000 series sales this and next 1-2 month, eg people that waited but got disappointed or simply can't afford the new cards.
 
Last edited:
  • Like
Reactions: ryan20fun

maddie

Diamond Member
Jul 18, 2010
4,723
4,628
136
I am reasonably confident that AMD has/had the resources necessary to prepare a 12nm consumer Vega and 7nm professional Vega, at the same time. They simply skipped 12nm altogether. AMD bailed out of the consumer market. I question whether or not they will ever seriously return to it.
Just addressing this point.

One point continually forgotten is the amortizing of R&D over a large production run. You see it in a lot of industries where CEOs take the decision to concentrate on high margin, high value products that are expensive to design. A short term strategy to improve profits, but it often backfires as revenue falls off appreciably reducing their competitiveness vs a more expansive producer as there is less funds for the critical R&D needed for future products.

If AMD does this (abandon the high volume gamer market), they will soon be out of graphics completely. Witness the demise of the specialty CPU manufacturers vs Intel. There are many more examples from autos, etc.

A possible middle ground is to ignore the extreme top end and just concentrate on the low-end to high mid-range where most of the products are sold. Not the best, but certainly not the worst situation. They have done this and its profitable for them.
 
  • Like
Reactions: beginner99

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
You're assuming that producing a refresh card on 12nm would have cost them too much money. An optical shrink or near-shrink with tweaks would have helped tame Vega's heavy power usage near "normal" clockspeeds (1585 MHz).

AMD is selling nearly every video card they produce right now. Yes, they can keep schlepping Vega and Polaris for another year, or they can up market share a bit with 12nm Vega (and maybe even 12nm Polaris). Either way they are going to sell whatever they push out to market. The R&D expenses would likely pay for themselves. Not that we'll ever know, since AMD chose not to go that route.



So? What does that have to do with 12nm consumer cards? We already knew AMD was going to go for 7nm Vega for compute, so there was no need to address that market with 12nm parts.



I am reasonably confident that AMD has/had the resources necessary to prepare a 12nm consumer Vega and 7nm professional Vega, at the same time. They simply skipped 12nm altogether. AMD bailed out of the consumer market. I question whether or not they will ever seriously return to it.



Irrelevant to the topic at hand. Zen's success does not preclude AMD from continuing to serve the consumer dGPU market. At. All. They maintained an entirely separate RTG division under Raja Koduri while developing Zen in the first place. Continued work on Zen+ and Zen2 could have continued while RTG refined Vega and moved it to GF's 12nm process, which according to many is not much more than a refined 14nm process anyway . . .



It is not rational for them to continue selling the same consumer dGPUs for over two years.



Vega FE wasn't #1 in games when it came out last year, but AMD launched it anyway. What's stopping them now?



The feature is non-functional in Vega9, period. Whether or not developers intended to support it is another matter altogether.



AMD could likely have arranged production of HBM2 modules at GF in the event of supply shortages from Samsung and Hynix. Though I am not sure what process Samsung used for production of HBM2 last year for Vega.



AMD still managed to deliver fairly regular releases from Fiji to Polaris to Vega. Polaris obviously did not replace Fiji, but it was a new uarch that was a significant upgrade for their midrange lineup. And obviously Vega replaced Fiji. 2018 was the right time for either a 12nm Vega or small Vega, or a 12nm Polaris update at least . . . something, anything.



Nvidia is under very little pressure to update the 1070 or 1080. AMD is the only party that could apply that pressure, and obviously they are doing nothing this year.
Ngg path was driver specific and ought to have worked without developer support.
Some of the same features as in maxwell btw.
So its not a small deal its not working. Look at the raw shader numbers to give you an idea of the potential.
 

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
  • Like
Reactions: ryan20fun