New Zen microarchitecture details

[Arachnotronic]

I will suggest to revisit the perf/watt again with RX 470 vs GTX 1060 (Reference $249 cards) in DX-12 games throughout the year. Then we can talk IF and how much behind is GCN 1.4 architecture and the 14nm LPP process.

It's always the next product, it seems. RX 480 is the only 14nm product that AMD has out now, it's the company's flagship Polaris part, and it has an equivalent from a competitor in terms of positioning.

Let's stop shifting the goal posts to defend AMD whenever they field an inferior product.

 

[therealnickdanger]

I will suggest to revisit the perf/watt again with RX 470 vs GTX 1060 (Reference $249 cards) in DX-12 games throughout the year. Then we can talk IF and how much behind is GCN 1.4 architecture and the 14nm LPP process.

Perhaps your post was a typo, but the 470 is not comparable to the 1060 as it's supposed to be a ~$150 card. I would argue that the 480 is also not comparable. The 480 is a $199 card (4GB unlockable to 8GB) with $230 variants (8GB out of the box). I wouldn't necessarily place it as a competitor to the $249 1060 (6GB) that will have $280 variants. The 480 can be CF'd while the 1060 can't be SLI'd. For $50 more, it had BETTER be BETTER.

If you didn't mistype, I'd like to know what you're planning. 😉

 

[frozentundra123456]

I will suggest to revisit the perf/watt again with RX 470 vs GTX 1060 (Reference $249 cards) in DX-12 games throughout the year. Then we can talk IF and how much behind is GCN 1.4 architecture and the 14nm LPP process.

Well, problem is, there are a lot of games that are not DX12, and a lot of other cards in the line-up for both companies. Anyway, 1060 competes with 480, not 470, and indications are very strong that perf/watt will strongly favor 1060. Your scenario, is a cherry picked, best case scenario for AMD. Maybe we should examine all cards in each lineup, in a wide variety of games.

 

[cytg111]

If the process has problems, like AMDs 65nm, AMD always fudges the performance figures to be best case in a one-off corner-case benchmark. So the general figure ends up VERY different to their claims.

As with Polaris.

From what I'm seeing, 14nm is their 65nm all over again.

Or G80 vs R600.

I did say Phenom, right from the beginning... I don't see how Zen can possibly be competitive with this process.

So there is lots of room for process improvement.
Did anyone expect AMD to hit a homerun first batter?
From where I am standing AMD rolled out a competetive product on GLOFO 14nm, first try. Sure some metrics could be better but common... On glofo? A competetive product, right down the main street/WSA street? I call that a couple of steps in the right direction. Here is to watching the next couple of steps fumble or carry on.

 

[Fanatical Meat]

What's with all the 460 hate? I'm not a huge AMD fan but look at its price point its meant to be a general purpose & light gaming card. Hell the 480 is just $200 and its good enough for 1080p gaming. I fail to understand the hate about new gpu that performs good enough costing $200 being a bad thing.

 

[frozentundra123456]

What's with all the 460 hate? I'm not a huge AMD fan but look at its price point its meant to be a general purpose & light gaming card. Hell the 480 is just $200 and its good enough for 1080p gaming. I fail to understand the hate about new gpu that performs good enough costing $200 being a bad thing.

This is getting off topic in the cpu forums, but nobody argues that the 480 offers a lot of performance for the price. Again though, it was overhyped by AMD, and especially by some in these forums.

It clearly failed to meet the performance per watt that people expected, and is basically over-volted in order to reach acceptable frequencies, with rapidly increasing power use with minimal overclocking. AMD is covering their butts now by saying "Oh, we will get 2.8x performance per watt with Polaris, but not *this* polaris. Wait for the 470." So there is a lot of criticism, rightly so IMO.

 

[coffeemonster]

From where I am standing AMD rolled out a competetive product on GLOFO 14nm, first try.

that's what I've been thinking. It's like a miracle ()🙂

 

[Fanatical Meat]

This is getting off topic in the cpu forums, but nobody argues that the 480 offers a lot of performance for the price. Again though, it was overhyped by AMD, and especially by some in these forums.

It clearly failed to meet the performance per watt that people expected, and is basically over-volted in order to reach acceptable frequencies, with rapidly increasing power use with minimal overclocking. AMD is covering their butts now by saying "Oh, we will get 2.8x performance per watt with Polaris, but not *this* polaris. Wait for the 470." So there is a lot of criticism, rightly so IMO.

I see regardless I say either is a good value card and its nice to see something release at the $200 point that is pretty capable of gaming with. I do agree AMD has a bad habit of over promising & under delivering.
Sorry for the side track.

 

[cytg111]

that's what I've been thinking. It's like a miracle ()🙂

I know, right?

 

[therealnickdanger]

...Again though, it was overhyped by AMD, and especially by some in these forums...

...It clearly failed to meet the performance per watt that people expected...

Let's be realistic. People in the rumor mill hyped it beyond all known logical standards. From the beginning, cooler heads anticipated 390-ish performance for $199 and that's what we got.

Aside from technical achievement or PR, who cares about perf/W? Miners running 100% 24/7? Perf/$ is the single most important metric to most consumers/gamers. The 1060 won't win that fight.

 

[AtenRa]

It's always the next product, it seems. RX 480 is the only 14nm product that AMD has out now, it's the company's flagship Polaris part, and it has an equivalent from a competitor in terms of positioning.

Let's stop shifting the goal posts to defend AMD whenever they field an inferior product.

There is no competitor to the RX 480 as of now and from the first DX-12 benchmark licks there will be no RX 480 DX-12 competitor from NVIDIA in 2016.
RX 470 has already been announced and we just waiting the official release date (late July ??). There is no goalpost changing, we know the specs of RX 470 before the licks of the GTX 1060. We do know that RX 470 is a 110W TDP card, so lets see how the perf/watt will be against the 120W TDP GTX 1060 in DX-12 games.

 

[The Stilt]

According to AMD, 2.8X is between RX470 110W TDP vs R9 270X 180W TDP in 3D Mark Fire Stike.

Right.

 

[Dresdenboy]

Why is it the process's fault? Why is it never the fault of the company that actually designed the product?

Anyway, Polaris got a big jump in perf/watt over Tonga, no doubt process related. But AMD's architecture was so behind NVIDIA's to begin with that expecting to hit the perf/watt levels that NVIDIA did with Pascal on 16FF+ (which is a superior process, but not by THAT much) was just unrealistic wishful thinking.

I think, you're right on the money here. Polaris seems to be an evolutionary step in the GCN family. Given AMD's R&D situation, FinFET costs, and cost targets (mainstream markets), which seems to have driven them to a high volume die strategy both for GPUs and CPUs, it would be reasonable to expect them to also keep design costs for lower margin products in check. This means: improved GCN and not a completely reworked GCN (e.g. faster clockable pipeline -> lower FO4/stage) for Polaris.

So what AMD could get out of Polaris was limited from day one. With a smaller die than Nvidia's DT Pascal GPUs (to keep costs down), the only option to get more performance out of it is ... upping the clock frequency into the less efficient range.

Here is my view on this:

Well, problem is, there are a lot of games that are not DX12, and a lot of other cards in the line-up for both companies. Anyway, 1060 competes with 480, not 470, and indications are very strong that perf/watt will strongly favor 1060. Your scenario, is a cherry picked, best case scenario for AMD. Maybe we should examine all cards in each lineup, in a wide variety of games.

I didn't check, but if DX12 is easier to use, offers more performance, is in more widespread use (Win10 adoption thanks to the upgrade offer and XBox One, etc.), DX11 (where AMD's drivers lack anyway, regardless if it's a FinFET GPU or not) will fade away over time. This might also be an effect of game project schedules.

This is getting off topic in the cpu forums, but nobody argues that the 480 offers a lot of performance for the price. Again though, it was overhyped by AMD, and especially by some in these forums.

It clearly failed to meet the performance per watt that people expected, and is basically over-volted in order to reach acceptable frequencies, with rapidly increasing power use with minimal overclocking. AMD is covering their butts now by saying "Oh, we will get 2.8x performance per watt with Polaris, but not *this* polaris. Wait for the 470." So there is a lot of criticism, rightly so IMO.

I think the whole back and forth in opinions here and elsewhere is a to be expected result of people with a more or less complete picture of semiconductors discussing a complex topic with only part of needed information at hand. 🙂

E.g. some information that might have gone missing might be this one:

Many people seem to be so easily triggered, that they begin to run before knowing the direction... 😉 This is true both for sceptics and "hypesters".

 

[AtenRa]

Perhaps your post was a typo, but the 470 is not comparable to the 1060 as it's supposed to be a ~$150 card. I would argue that the 480 is also not comparable. The 480 is a $199 card (4GB unlockable to 8GB) with $230 variants (8GB out of the box). I wouldn't necessarily place it as a competitor to the $249 1060 (6GB) that will have $280 variants. The 480 can be CF'd while the 1060 can't be SLI'd. For $50 more, it had BETTER be BETTER.

If you didn't mistype, I'd like to know what you're planning. 😉

Because they are 110W (RX 470) and 120W TDP (GTX 1060), we can easily compare perf/watt between the two, also almost same die size (RX 470 is cut down so even closer to GP106 die size).

 

[AtenRa]

Well, problem is, there are a lot of games that are not DX12, and a lot of other cards in the line-up for both companies. Anyway, 1060 competes with 480, not 470, and indications are very strong that perf/watt will strongly favor 1060. Your scenario, is a cherry picked, best case scenario for AMD. Maybe we should examine all cards in each lineup, in a wide variety of games.

I have explained before that GCN is designed for Mantle/DX-12 and Vulkan, it is not DX-11 optimized and thus it doesnt have the same efficiency in DX-11 workloads (unless specifically optimized). We know GCN is not the best DX-11 architecture, we dont care, most of the major AAA game titles this year and next are going to be DX-12 and that is what people should care more.

As i have mentioned above, RX 470 = 110W TDP and GTX 1060 = 120W TDP. We can directly compare them and draw any conclusions about perf/watt and perf/size.

 

[AtenRa]

Right.

Yea, do you have the slide deck and see note 2 ???

 

[The Stilt]

Yea, do you have the slide deck and see note 2 ???

Why would I have it available?
The fact is that they claimed up to 2.8x (180%) higher performance per watt improvement for RX 480. Yet they delivered ~ 10% higher performance per watt over their last generation 28nm GPU (Fiji as Fury X) and ~85% higher performance per watt over their most inefficient first generation GCN GPU (Tahiti as 280X), which is over four and half years old already.

 

[Sweepr]

As i have mentioned above, RX 470 = 110W TDP and GTX 1060 = 120W TDP. We can directly compare them and draw any conclusions about perf/watt and perf/size.

GTX 1060 competitor is RX 480 8GB, GTX 1050 will compete with RX 470.

According to AMD, 2.8X is between RX470 110W TDP vs R9 270X 180W TDP in 3D Mark Fire Stike.

Nowhere close to 2.8x in actual games, try 1.35x:

This is mediocre for an improved architecture on a FinFET node.

 

[ElFenix]

will you dorks get back on topic

 

[AtenRa]

GTX 1060 competitor is RX 480 8GB, GTX 1050 will compete with RX 470.



Nowhere close to 2.8x in actual games, try 1.35x:

This is mediocre for an improved architecture on a FinFET node.

RX 480 has ~2x perf/watt vs R9 280X not 1.35.

RX 480 vs 280X = 92%
RX 480 vs 285 = ~82%

GTX 1070 only has 70% higher perf/watt than GTX 970 according to the slide you posted.

 

[AtenRa]

will you dorks get back on topic

yea my bad, that was last OT sorry guys.

 

[el etro]

I think, you're right on the money here. Polaris seems to be an evolutionary step in the GCN family. Given AMD's R&D situation, FinFET costs, and cost targets (mainstream markets), which seems to have driven them to a high volume die strategy both for GPUs and CPUs, it would be reasonable to expect them to also keep design costs for lower margin products in check. This means: improved GCN and not a completely reworked GCN (e.g. faster clockable pipeline -> lower FO4/stage) for Polaris.

So what AMD could get out of Polaris was limited from day one. With a smaller die than Nvidia's DT Pascal GPUs (to keep costs down), the only option to get more performance out of it is ... upping the clock frequency into the less efficient range.

Here is my view on this:

Great post Dresdenboy.

 

[Zstream]

E.g. some information that might have gone missing might be this one:

Many people seem to be so easily triggered, that they begin to run before knowing the direction... 😉 This is true both for sceptics and "hypesters".

Wow, I haven't actually seen that slide. I wonder why people reference the 2.8x to the 480?

 

[LTC8K6]

Wow, I haven't actually seen that slide. I wonder why people reference the 2.8x to the 480?

Because AMD puts out confusing slides? Slides that have to be deciphered and footnoted, and interpreted.

Not that they are alone in that...

 

[coercitiv]

Wow, I haven't actually seen that slide. I wonder why people reference the 2.8x to the 480?

Because AMD puts out confusing slides? Slides that have to be deciphered and footnoted, and interpreted.

Because of this.

Top left "RX 480 Built on 14nm FinFET"
Middle of the screen "UP TO 2.8X" PPW with a tiny footnote sign.

If they want to play the footnote game... they get the PR and credibility to match.

Personally I see nothing confusing about this: they lied. As I may have said before on the VC&G forum, from now on AMD better back any claim they make with a live demo: zero credibility policy from me.

will you dorks get back on topic

Oops, missed an entire page of comments from my last post. I'll leave the post as is even if redundant.