(Updated) AMD Navi GPU to Offer GTX 1080 Class Performance at ~$250 Report Claims

[Ryujin2003]

So, according to WCCF Tech (insert salt as desired for taste), AMD's Navi is looking to produce GTX 1080 performance for its mainstream card, but this won't be until 2019. This will replace their current mainstream card, the RX580, and will apparently provide GTX 1080 performance for approximately $250. If this is true, that would be very cool. If they were able to drastically improve their mainstream card performance and still have gas left in the tank to provide enthusiast cards that will be competitive against Nvidia's next GPU lineup, then AMD could still be in business..... Though, I'm not too sure of the price point. I feel like it'll have to cost more in order for them to offset research and such. Otherwise, the price could be believable because for that performance at that price mark (assuming Nvidia's next lineup cannot perform the same at that price point), they would probably steal a huge market share from Nvidia (for about 2 weeks until Nvidia panic produces something in response /s).

 

This article refers to Navi as the Ryzen of GPUs. And as such, just remember the hype-train that came with Ryzen. However, even with that comparison, if Navi is like Ryzen, it'll hopefully bring back some competition.... Now.. If only to eliminate miners before the release of future GPUs, so that a $250 card doesn't cost $800 on Newegg/Amazon....

 

Quote

AMD’s next generation mainstream Navi GPU will reportedly offer GTX 1080 / RX Vega 64 class performance and replace the company’s current RX 580 cards as a mainstream class product in 2019.

AMD Navi Mainstream GPU to Have GTX 1080 Class Performance, Succeed RX 580

Navi is Radeon’s last Graphics Core Next based architecture and is expected to be the world’s first high performance GPU built on 7nm process technology. Little is known about Navi in the techspere to date. One detail that we seldomly see reported is that AMD is working on two Navi GPUs, we’ll call them Navi 10 and Navi 11 for the time being. One is designed for the desktop market and the other for the mobile market.

 

According to this report from Fudzilla, Navi will not be a large high-end GPU.  Although the report doesn’t specify which Navi GPU is being talked about, we’re going to assume that it’s the Navi 10 desktop part. The report further states that this Navi part will be a high performance, low power chip with the performance of today’s high-end GPUs and the positioning and power consumption of mainstream parts. An RX 580 successor in that ~$250 sweet spot in the market.

 

This makes sense from several perspectives. From a manufacturing point of view, it’s not feasible to produce a large GPU on a brand new cutting edge process like 7nm early in the node’s life-cycle. The yields and wafer costs make this prohibitively expensive. This is why NVIDIA and AMD were only able to introduce the GTX 1080 Ti and Vega in 2017, rather than 2016.

 

It also makes sense from a profitability point view, as mainstream and mid-range GPUs far outsell high-end GPUs, by a factor of 4 to 1 in fact. This is how AMD was able to double its market share in 2016, just with mainstream Polaris 10 & 11 GPUs.

Many (not the majority) of consumers will be letdown due to the lack of response from AMD with the soon to be release of Nvidia GPUs. AMD has been focusing and dedicating a lot of time on Navi, so I'm really hoping they nail this. There is a lot of pressure on them to start being more competitive with Nvidia's offerings.

 

It would be very beneficial for AMD if they are able to release BOTH desktop and mobile variants of their GPU at the same time. This would definitely increase profits rather quickly, as long as they don't run in to the cliff dive response of supply like they did with Vega.

Quote

Behind The Scenes at AMD’s Radeon GPU Labs

Over the past year we’ve been hearing whispers about a project at the company to bring horizontal die stacking technology and expertise over from the CPU department to the GPU department. With the intention of making high-end multi-die GPUs, akin to Ryzen Threadripper. We would have something like Navi 10 and Navi 20, with Navi 20 featuring two Navi 10 dies in the same package. This Navi 20 is what would theoretically go up against NVIDIA’s high-end parts, a la GTX 1180/Ti.

 

This die-stacking program we’re told is what the company meant by “scalability” in its Navi Roadmap. More recently we’ve been hearing that the future of this Navi die stacking project may be uncertain in 2019, as more die stacking engineering effort is poured into AMD’s entirely new 2020 post-GCN “Nextgen” architecture. A design that we’re told is as revolutionary as the company’s Zen CPU architecture.

 

If the Navi die stacking project has indeed been postponed then we will see AMD debut both its revolutionary new architecture, the Zen of GPUs if you will, and multi-die GPU stacking technology in 2020, after Navi. This would allow the company to address all segments of the market, from the entry level all the way to the ultra enthusiast segment just by employing a single GPU die that can be stacked to meet the needs of every market segment.

Even with the release of Ryzen, AMD focused on the "scalability" of their product. It would definitely be interesting to see if this method works well in the GPU space. I wonder if they will have "infinity fabric" or "denim" or whatever the fabric is going to be called... fabric limitations with the interconnected dies on a Navi 20, similar to what has been explained with Ryzen. However, with faster memory than DDR4, this could possibly be mitigated.

Quote

Wccftech	AMD 400 Series	AMD 500 Series	AMD 600 Series	AMD 700 Series
Architecture	Polaris	Polaris / Vega	Vega	Navi
Process Node	GF 14nm LPP	GF 14nm LPP	GF 12nm LP	GF 7nm LP
Memory	GDDR5	GDDR5 / HBM2	HBM2	HBM3/GDDR6
GPUs	Polaris 10, Polaris 11	Polaris 20, Polaris 21, Vega 10, Vega 11	TBA	Navi 10, Navi 11
Year	2016	2017	2018	2019

WCCFTech also states that they will be releasing another article for another AMD GPU adventure. They didn't give specifics, but their chart has AMD 600 series for 2018, under the Vega architecture..... So, I guess they ran out of salt, and had to run by the mine to grab some more. So, that one will have to wait.

 

From the original report on Fudzilla:

Quote

We have been sitting on this piece of information for a while, but maybe it's the right time to share it with you. Navi 7nm the 2019 chip will not be a high end GPU, it will be a quite powerful performance/mainstream chip.

Think of it as the Radeon RX 580 / 480 replacement. It will be small, and is likely to perform as well as the Vega 14nm that shipped last year. In the Nvidia performance world Navi should perform close to Geforce GTX 1080 which is quite good for the mainstream part but probably on part of the mainstream part planned after the high end part.

...

 

Navi 7nm won’t have two different SKUs, one that miraculously goes after the Geforce Turing edition planned for later this year. So, the long story short, AMD won’t have anything in the high-end space faster than Vega between now and end of 2019. In GPU world this is eternity. This is the product where Radeon Technology Group really spent some time to go after this highly competitive and profitable mainstream / performance market. Of course, AMD did sell every single RX 580 / 570 cards as well as every single Vega 56 and 64 manufactured to miners, as these guys were buying anything to get their hands on it, but with the current situation on the market, one can only hope that there will be a mining demand in mid 2019. 

Ii am hoping that Navi is very competitive, but I will not hold my breath. I've been teased enough to know not to get excited until I actually see what this can perform... And even after that, contain my excitement until I can actually get hands on a non-reference unit. Of course, the wait will be near infinite. Maybe AMD knows what they are doing, and are doing something very well calculated.... or.... they might just flop, and give up even more to NVidia.

 

(Update) So, due to some of the comments I've heard from some online YouTube personalities about how this is horrible because GTX 1080 be inept with Nvidia's future release, and how AMD is falling behind and laying everyone down..

 

Quote

The report on Fudzilla is claiming that the 7nm Navi GPUs will perform around the same level as a top-end 14nm AMD Vega card, which would put it around the same level as a GTX 1080. By that time we’re expecting to have a broad range of next-gen Nvidia graphics cards at our disposal - we should have a GTX 2080 (or GTX 1180, or GTX 1111111111118, who knows at this point...) as well as GTX 2070 and GTX 2060 cards kitting out the top tiers of the gaming market.

If our performance expectations of the next GeForce gen are vaguely correct that would put the top AMD Navi card somewhere between the GTX 2070 and GTX 2060 level. And if they price them right that wouldn’t be a massive problem for them - that’s where the volume market is and selling at a price that gives the performance-comparable Nvidia cards problems could really fill their coffers.

 

So, according to PCGamesN, the Navi cards should be estimated between the 1060 and 1070 replacement cards from Nvidia, which doesn't make Navi seem that bad. And if that is the case, and the scalability and architecture design for memory works out, you could potentially get a card between the **60 and **70, with potential pricing that could definitely make AMD's offerings much moar appealing.

 

Quote

Without competition at launch Nvidia are likely to price the next-gen GeForce GPUs pretty high, giving them chunky mark-ups, but giving AMD a target price to get below. Though that will also give Nvidia a lot of wiggle room to drop their own prices in response.

 

Again, it's hard to throw hate and salt either direction based on an unverified and immeasurable rumor.

 

Original Sources:

https://wccftech.com/rumor-amd-navi-mainstream-gpu-to-have-gtx-1080-class-performance/

https://www.fudzilla.com/news/graphics/46038-amd-navi-is-not-a-high-end-card

https://www.pcgamesn.com/amd-navi-mainstream-gpu

[Castdeath97]

Here we go again, boys and girls! It's the RX 480 hype train all over again! 

 

Honestly, at this point, I'd settle for 1070 performance at 250 USD and I'm sure many will agree. Nowadays you can't even get 1060 levels of performance at that price point that you used to get a year and a half ago ...

[ScratchCat]

*Please don't be another Vega*

 

Everyone should take this with a grain of salt, the hype for Polaris was just as high -"GTX 980 performance at $200".

Hopefully the extra revenue from the CPU department can boost the R&D budget for the GPU sector.

[samcool55]

For 250 bucks, probably only during the first hour or whatever until a few miners go on a shopping spree again...

[Spotty]

20 minutes ago, Ryujin2003 said:

AMD's Navi is looking to produce GTX 1080 performance for its mainstream card, but this won't be until 2019.

Cool, so in 2019 AMD will finally be able to compete with the performance of Nvidia's 2016 cards. If AMD's goal post for performance of their future generation cards is [what will soon be] last generation Nvidia cards, then I really doubt they will be able to compete with Nvidia at all over the next few years. With all the stuff Nvidia has been pulling lately (GPP), this is bad news.

I realise the story here is the price point it will be available for, however with the state of things in the GPU market and the memory market (particularly HBM) at the moment I really wouldn't trust anyones predictions on how much a GPU will cost in more than a years time.

[Jurrunio]

If it's HBM, prices will go even higher and supply, even less... Hope that doesnt happen.

[GoldenLag]

considering 1080 is a mid end card. i assume Navi will have a higher variant than the 1080 performance. a lot of people consider the 1060 and 1070 or the 570/580 to be mainstream. it wont be a hit if they dont have something to compete with at Nvidias xx80ti cards. People tend to look at the highest end card and then step down from there untill they hit a pricepoint they can pay. Navi needs to beat Nvidia at all pricepoint to be taken as the go to card. at least this is what i think about it. Just because you are best at the mid pricepoint doesnt mean you will allways grab marketshare. 

 

If this is  true that Navi can provide that at performance at that 250$ pricepoint, then Navi will be amazing. Hopefully that performance is hit using something like 6/8 computeclusters. remember that 1080 performance isnt huge in 2 years. its very good performance in the same sence the R9 390x is still good and the 780ti is still good. If Navi doesnt give top end performance in 2 years it wont shake the market as much as we want

[xAcid9]

RIP Novideo

 

2019? 1080 become low mid end card already if not high low end.

[asus killer]

i don't get this, AMD already has Vega 64 to compete with 1080 kind of, so what would be the point to be developing a card for a year to do the exact same thing? Yes it would be more power efficient with 7nm, still seems like a lot of work and money for nothing.

I don't believe they are developing Navi at 7nm just to come up with a Vega 64 performance clone. If it ends up to be true it's because Navi was a failure and couldn't be pushed further.

[Coaxialgamer]

This would be impressive if this wasn't a 7nm chip trying to match a 3 year old 16nm gpu that barely qualifies as high end in the first place. 

[cj09beira]

2 minutes ago, asus killer said:

i don't get this, AMD already has Vega 64 to compete with 1080 kind of, so what would be the point to be developing a card for a year to do the exact same thing? Yes it would be more power efficient with 7nm, still seems like a lot of work and money for nothing.

I don't believe they are developing Navi at 7nm just to come up with a Vega 64 performance clone. If it ends up to be true it's because Navi was a failure and couldn't be pushed further.

because it would be a card on a different price bracket, a shrinked vega 64 on 7nm might be able to reach 1800 mhz or even a bit more which would put it more times closer to a 1080 ti 

so it makes sense to have that card made but it would also make sense to make a bigger die, the problem is that until now amd hasn't made anything with more than 64 rops and 64 Cus, my hope is that "Scalable" (word used to describe navi on a slide) is able to have more shader engines and so it will be able to have more Rops and Cus, it could also be a mcm (would serve the same purpose)

[Taf the Ghost]

45 minutes ago, Castdeath97 said:

Here we go again, boys and girls! It's the RX 480 hype train all over again! 

 

Honestly, at this point, I'd settle for 1070 performance at 250 USD and I'm sure many will agree. Nowadays you can't even get 1060 levels of performance at that price point that you used to get a year and a half ago ...

Agreed. Navi was going to replace Polaris and be on 7nm, though it's not a given that it'll be on GloFo's 7nm. We don't know that for certain. Only Vega on 7nm is public information. (AMD can produce chips at TSMC now.)

[Trixanity]

Here we go again.

 

I just find the whole notion of targeting existing performance and bringing it down in price is a natural phenomenon and not a goal in itself. It seems wholly unambitious and I really hope that's not what RTG is talking about internally because that would be sad.

Of course, it's WCCF so take it with a grain of salt. I mean they're talking about 12nm GPUs which we pretty much with 98% certainty know to be cancelled. Vega Instinct 7nm is all that's on the roadmap for 2018. Anything else is a pipe dream unless AMD are pulling a fast one and surprises everyone by pulling a bunny out of the hat out of nowhere. 

 

Them working on two dies is not surprising at all. It's the same AMD has done for years now and it's just not good enough. It can't satisfy the market. AMD brought out Polaris 12 much later because 10 and 11 wasn't enough. They also brought out Vega 10 and that still isn't enough. So at this rate they'd need 5 dies (if we discount the MCM/die-stacking) to get a top to bottom stack. 

 

I'll see it before I believe anything.

 

[leadeater]

23 minutes ago, GoldenLag said:

considering 1080 is a mid end card.

What? No it isn't.

[cj09beira]

2 minutes ago, leadeater said:

What? No it isn't.

in terms of die sizes yes it is, thats probably what he means

[Taf the Ghost]

1 minute ago, cj09beira said:

because it would be a card on a different price bracket, a shrinked vega 64 on 7nm might be able to reach 1800 mhz or even a bit more which would put it more times closer to a 1080 ti 

so it makes sense to have that card made but it would also make sense to make a bigger die, the problem is that until now amd hasn't made anything with more than 64 rops and 64 Cus, my hope is that "Scalable" (word used to describe navi on a slide) is able to have more shader engines and so it will be able to have more Rops and Cus, it could also be a mcm (would serve the same purpose)

My hope is they've broken down the GCN and rebuilt whatever systems were causing bottlenecks so it scales properly by CUs. Pretty much neither the Vega 56 or 64 are bottlenecked by Clocks or CU, but by the amount of memory throughput they have.  Nvidia has been tweaking things, shrinking things and just adding more SMs, and scaling pretty well from that. AMD makes a brilliant iGPU Architecture that doesn't scale great beyond 32 CUs. That's the big problem.

 

I also kind of hope they can fully split their die designs for Mainstream & Compute. As much as you can't buy a 1080 Ti competitor from AMD, at least they can find a way to fight Nvidia in the really high margin spaces. There's optics but not a lot of money in the high-end GPU space.

[asus killer]

15 minutes ago, cj09beira said:

because it would be a card on a different price bracket, a shrinked vega 64 on 7nm might be able to reach 1800 mhz or even a bit more which would put it more times closer to a 1080 ti 

so it makes sense to have that card made but it would also make sense to make a bigger die, the problem is that until now amd hasn't made anything with more than 64 rops and 64 Cus, my hope is that "Scalable" (word used to describe navi on a slide) is able to have more shader engines and so it will be able to have more Rops and Cus, it could also be a mcm (would serve the same purpose)

i can only conclude that you read nothing of the quoted article. There will be no Vega 7nm (edit: for gaming) and there will be no gpu faster than vega according to them, and that's what we are commenting.

[Taf the Ghost]

5 minutes ago, leadeater said:

What? No it isn't.

 

2 minutes ago, cj09beira said:

in terms of die sizes yes it is, thats probably what he means

Realistically it shouldn't be, by Die Size and relative performance to the previous generation, but Nvidia's "Compute First, Consumer Second" approach paid off a few years ago. As a result, AMD can't dislodge them at the performance level, so they still charge a premium.

[Taf the Ghost]

1 minute ago, asus killer said:

i can only conclude that you read nothing of the quoted article. There will be no Vega 7nm and there will be no gpu faster than vega according to them, and that's what we are commenting.

Vega on 7nm will be out by the end of 2018, which I believe was from Lisa Su herself, but it's a Machine Learning only card. It's for AMD's Instinct line. It's also the "testing" process for GloFo's 7nm.

[Ryujin2003]

Just now, asus killer said:

i can only conclude that you read nothing of the quoted article. There will be no Vega 7nm and there will be no gpu faster than vega according to them, and that's what we are commenting.

The initial release of Navi will target mainstream because that's where the bulk of the income is from. However, their goal is to bring much higher performance to mainstream. So if this will be on part with 1080, then it should be relatively similar, but hopefully better than Vega. Their Navi 20 is supposedly designed to be their enthusiast lineup, using dual Navi 10 does, similar to the construction of Ryzen and Threadripper.

[Ryujin2003]

Just now, Taf the Ghost said:

Vega on 7nm will be out by the end of 2018, which I believe was from Lisa Su herself, but it's a Machine Learning only card. It's for AMD's Instinct line. It's also the "testing" process for GloFo's 7nm.

Fudzilla started the same thing on their report. Maybe I will add that when I get a chance.

[leadeater]

4 minutes ago, cj09beira said:

in terms of die sizes yes it is, thats probably what he means

Die size there is only 1 above it, discounting Volta as that is not in the GeForce lineup. This is a case of enthusiast blinders, too much looking at the high end of the market.

[GoldenLag]

7 minutes ago, leadeater said:

What? No it isn't.

some that has looked at Nvidia in the past consider the 1070/1080 to be mid end and the 1080ti and Titan to be high end. i myself consider the 1080 to be mid end. the Vega 64/liquid is AMD`s high end. In performance you might consider it to be and high end card, but in reality it is just a mid-end card. Then again that wasnt the point i was making. also i knew someone was going to comment on me calling it mid-end. 

[Princess Luna]

15 minutes ago, Coaxialgamer said:

This would be impressive if this wasn't a 7nm chip trying to match a 3 year old 16nm gpu that barely qualifies as high end in the first place. 

Hey it only took AMD 3 years to make Vega have the same performance of the TITAN X Maxwell... seems like a pattern, AMD's often 3 years behind on the highest end.

[asus killer]

2 minutes ago, Taf the Ghost said:

Vega on 7nm will be out by the end of 2018, which I believe was from Lisa Su herself, but it's a Machine Learning only card. It's for AMD's Instinct line. It's also the "testing" process for GloFo's 7nm.

yes, i was talking in terms of gaming that i guess is what interest us, still i could have mentioned it, it is in fact in the article the OP quoted.