AMD 7nm GPU News and Rumours 2018/2019

[MyBrainz]

The price will be similar to the ones today. £200-250 bracket.

Also it wouldn't surprise me it it was very close to the GTX1070 performance either.
Already in some games the 1070 is just 10%-15% faster than the 580, if that, at 1440p and 1080p.

Any more than £250 and they wont see my money, i think they will end up closer to £300 and i only buy Sapphire, maybe an 8GB Nitro+ will be around £275, and if that's the case, i'll keep using my RX 480.

 

[Gerard]

Always makes me scratch my head as to how they can be different sizes but at the same time be the same nm. Mostly marketing crap i guess? Or similar to how a company can claim they have "x" number of transistors because they count parts the others don't?

 

[Poneros]

I don't know. RTG shows very pronounced signs of extremely struggling entity, like not knowing what exactly to do, and what plans they actually have. Something similar to 3dFX, if I can recall it correctly. We all know what happened to 3dFX.

To the contrary, RTG knows exactly what it's doing. They have a gameplan and are executing it: focus on compute, focus on consoles, and keep developing the consumer GPU side to the best of their abilities while changing architectures. Not unlike what they did with the CPUs. Unfortunately that means you won't see them be as robust as Nvidia on the consumer side for quite some time. Let's not forget AMD is fighting with a fraction of Nvidia's R&D budget, they're not going to do miracles.

 

[Rroff]

Always makes me scratch my head as to how they can be different sizes but at the same time be the same nm. Mostly marketing crap i guess?

One thing that doesn't really convey well either is that Intel 10nm is a mixture - some stuff is closer to "12nm" some bits important to CPU design are (or were) closer to a second generation 7nm process - rumour is they've relaxed those aspects but for various reasons I'm not sure that it is quite as certain sites/people are claiming.

 

[4K8KW10]

To the contrary, RTG knows exactly what it's doing. They have a gameplan and are executing it: focus on compute, focus on consoles, and keep developing the consumer GPU side to the best of their abilities while changing architectures. Not unlike what they did with the CPUs. Unfortunately that means you won't see them be as robust as Nvidia on the consumer side for quite some time. Let's not forget AMD is fighting with a fraction of Nvidia's R&D budget, they're not going to do miracles.

RTG uses the same architecture for everything - how could you say they focus on these things differently?
Vega has too many transistors which are useless in games - that's it.

Always makes me scratch my head as to how they can be different sizes but at the same time be the same nm. Mostly marketing crap i guess? Or similar to how a company can claim they have "x" number of transistors because they count parts the others don't?

Well, look at the shape of the transistor - usually this marketing number shows the size of the smallest dimension, part of the transistor.

 

[Gerard]

Well, look at the shape of the transistor - usually this marketing number shows the size of the smallest dimension, part of the transistor.

I doubt its quite that simple. if it were then people wouldn't say "x companies 10nm is closer to x companies 7nm" etc. It's like how people dispute how companies count transistor numbers inside gpu's, apparently nvidia and amd both count them differently somehow or other.

 

[Poneros]

RTG uses the same architecture for everything - how could you say they focus on these things differently?
Vega has too many transistors which are useless in games - that's it.

Implementation is still different depending on the domain, that's not trivial just because it's not a new arch. Considering they don't have infinite resources that means time spent on X instead of Y or Z means they are focusing differently on each (what else can focus mean if not resource allocation?). Besides, you've changed the subject. The point being they very much have a plan and are working on executing it. You might not like it because it's not focusing on your needs/wants but it's nonetheless there.

 

[Hotwired]

Always makes me scratch my head as to how they can be different sizes but at the same time be the same nm. Mostly marketing crap i guess? Or similar to how a company can claim they have "x" number of transistors because they count parts the others don't?

Likely its very technical but dumbed down enormously to fit in the heads of investors and consumers.

Doubtless none of those companies give a dam about what it's called because they know the details and the dumbing down they are obliged to do for outsiders.

 

[D.P.]

To the contrary, RTG knows exactly what it's doing. They have a gameplan and are executing it: focus on compute, focus on consoles, and keep developing the consumer GPU side to the best of their abilities while changing architectures. Not unlike what they did with the CPUs. Unfortunately that means you won't see them be as robust as Nvidia on the consumer side for quite some time. Let's not forget AMD is fighting with a fraction of Nvidia's R&D budget, they're not going to do miracles.

But they are not really that focused on compute either. The monopoly Nvidia has in the HPC and DL markets makes the gaming GPUs look like a balanced market share.


AMD have a lot of potential but they are floundering, resources spread too thin, their GPU department seems to be in turmoil without any long term stability and no focused execution.

Looks like we will get another re-spin of Polaris, which no doubt means Navi is thr end of next year and a Vega replacement will be 2020.

 

[RBImGuy]

To the contrary, RTG knows exactly what it's doing. They have a gameplan and are executing it: focus on compute, focus on consoles, and keep developing the consumer GPU side to the best of their abilities while changing architectures. Not unlike what they did with the CPUs. Unfortunately that means you won't see them be as robust as Nvidia on the consumer side for quite some time. Let's not forget AMD is fighting with a fraction of Nvidia's R&D budget, they're not going to do miracles.

They went all in on 7nm.
The right call for them.
still play my games on 1440p and a 480.
Looking forward amd´s 7nm cards

 

[Armageus]

I don't know. RTG shows very pronounced signs of extremely struggling entity, like not knowing what exactly to do, and what plans they actually have. Something similar to 3dFX, if I can recall it correctly. We all know what happened to 3dFX.

Not really - they may have had to "reset" because Fury/Vega/HBM hadn't turned out as well as expected, but it's better than carrying on down that path for the sake of it.
Until "new" products i.e. Navi arrives, you still have to be able to sell something, so a cheap/easy respin of Polaris makes sense at the low-mid range, and possibly we may still see 7nm Vega parts at the start of next year, depending on the expected release of Navi (which I think is at least Q3 2019 for "low-end" parts and Q2 2020 for "high-end" parts)

 

[GAC]

To the contrary, RTG knows exactly what it's doing. They have a gameplan and are executing it: focus on compute, focus on consoles, and keep developing the consumer GPU side to the best of their abilities while changing architectures. Not unlike what they did with the CPUs. Unfortunately that means you won't see them be as robust as Nvidia on the consumer side for quite some time. Let's not forget AMD is fighting with a fraction of Nvidia's R&D budget, they're not going to do miracles.

i agree they know what they are doing, but lets be blunt its nowhere near enough or fast enough. the problem is amd dropped the ball and haven't been able to pick it back up as of yet when it comes to being competitive on performance, sure people will say they can beat a 1070 and a 1080 but in the world of 1080ti's and now the 2080ti that means very little to the vast majority of people as they will just look at the head line cards and dive in to that brand as its winning the overall race. yes on here most of us know different and will research whatever we buy, but even on here there seems to be some members looking at the 2000 series as the latest example and looking at dropping £600+ ona 2070 or 2080 for a couple of frames per second and seeing it as a good investment because its a newer card.

 

[G J]

OH goody another card similar to the 290(x) performance but will likely be sold for £350+, what a bargain!

Yep and it's funny as some posters here keep shouting about mindshare and nvidiots or whatever term they use and why dont we help out poor lil AMD while glossing over that AMD has played its part in completely stagnating the mid range for years now.

 

[Armageus]

i agree they know what they are doing, but lets be blunt its nowhere near enough or fast enough.

Not sure they could really do anything quicker (or certainly better in the short term).

Navi is supposed to be a clean slate architecture, but obviously that takes time to design/sample/debut/produce. In the mean time what else could they do?

Pretty much the options were something like:
A) Carry on and evolve another Vega based product (but 4096 shaders etc is basically the limit) and is a dead end / wasted development resources
B) Fix/Develop Polaris - but to go much further with that it needs a wider memory bus (384-bit or 512-bit ala 280X or 290X) but that then makes the chips/boards costly to produce (and also would be publicly admitting Vega/HBM was a failure)
C) Stop all existing development and "reset", focus resources on new architectures with intention that Navi will be competitive new product. In the interim, use newer fab tech (12nm/7nm) for quick/easy respins of existing products, that don't need much development resource.

Option "C" is where we seem to be at.

 

[GAC]

the problem they are tweaking stuff when they could have just piled the time and money in to navi, the only reason i think they are tweaking polaris a contract they already had from sony/ms for console stuff although i thought they where supposed to be using vega with ryzen but who knows.

hopefully navi will be a move forward and not another side step like they have done in the past just to get that perf/watt down. just gotta wait for jan and see whats announced, and pray they hit market with them before summer.

 

[Armageus]

the problem they are tweaking stuff when they could have just piled the time and money in to navi

The amount of resource being devoted to a Polaris respin is likely tiny in relation to what is needed for Navi (may even be driven by engineers at the Foundry to do the 14nm->12nm adaption - I don't know if that's possible), it's unlikely to have brought Navi forward enough to negate the gap in having any products available.

 

[nashathedog]

Not just 1080p, 1440p too. Hell, I use an RX480 at 1800p even. Not everyone needs max settings 144fps. So the difference between mine and the RX680 could end up at like 20%. That's significant.

I spent a few months using an RX 480 at 3440x1440 this year and I could get perfectly playable performance for only a small difference visually. Usually you don't have the time to stop playing a game and start staring at things to spot the difference.

 

[bru]

The price will be similar to the ones today. £200-250 bracket.

Also it wouldn't surprise me it it was very close to the GTX1070 performance either.
Already in some games the 1070 is just 10%-15% faster than the 580, if that, at 1440p and 1080p.

Whereas I do agree there are some titles where the 580 does quite well against the 1070, 10-15% is far from the norm, which is just under the 40% mark.

 

[LePhuronn]

If that WCCFtech article is accurate, I'm surprised the respin isn't using GDDR6

 

[bru]

I would be very surprised to see a £200-250 price bracket, seeing as that at the end of the 580's life they are going for £215-£419 looking at the prices from here. So £200-250 might be asking a bit too much.