Wednesday, September 27th 2017

AMD "Navi" GPU Architecture Successor Codenamed "Arcturus"?

Arcturus is the fourth brightest star in the night sky, and could be the a new GPU architecture by AMD succeeding "Navi," according to a Phoronix report. The codename of Navi-successor has long eluded AMD's roadmap slides. The name "Arcturis" surfaced on Phoronix community forums, from a post by an AMD Linux liaison who is a member there. The codename is also supported by the fact that AMD is naming its GPU architectures after the brightest stars in the sky (albeit in a descending order of their brightness). Polaris is the brightest, followed by Vega, Navi, and Arcturus.

AMD last referenced the Navi-successor on a roadmap slide during its 2017 Financial Analyst Day presentation by Mark Papermaster. That slide mentioned "Vega" to be built on two silicon fabrication processes, 14 nm and "14 nm+." We know now that AMD intends to build a better-endowed "Vega" chip on 7 nm, which could be the world's first 7 nm GPU. "Navi" is slated to be built on 7 nm as the process becomes more prevalent in the industry. The same slide mentions Navi-successor as being built on "7 nm+," which going by convention, could refer to an even more advanced process than 7 nm. Unfortunately, even in 2017, when the industry was a touch more optimistic about 7 nm, AMD expected the Navi-successor to only come out by 2020. We're not holding our breath.
Source: Phoronix
Add your own comment

43 Comments on AMD "Navi" GPU Architecture Successor Codenamed "Arcturus"?

#1
R0H1T
Don't know about the $ wars but AMD seems to be winning the code name wars by a landslide, against both of its major competitors :nutkick:
Posted on Reply
#2
londiste
Polaris is the brightest, followed by Vega, Navi, and Arcturus.
I would certainly hope not... :D
Posted on Reply
#3
NC37
R0H1T said:
Don't know about the $ wars but AMD seems to be winning the code name wars by a landslide, against both of its major competitors :nutkick:
It's all fun and games till some crazy engineer names one betelgeuse...
Posted on Reply
#4
Camm
Guess we needed another GPU series for the AMD hopeful to wait for.
Posted on Reply
#5
R0H1T
NC37 said:
It's all fun and games till some crazy engineer names one ...
And then I googled Betelgeuse aka Alpha Orionis, surprisingly Arcturus aka Alpha Boötis sounds mildly better :toast:
Posted on Reply
#6
EntropyZ
R0H1T said:
And then I googled Betelgeuse aka Alpha Orionis, surprisingly Arcturus aka Alpha Boötis sounds mildly better :toast:
Hmm, AMD Lisa Su Thicc Booty Special Edition™. No, I didn't make this up while I was drunk.
Posted on Reply
#7
Octopuss
I don't even know what is what anymore. Architecture names, gpu code names, then architectures that are pro and not pro.... I don't even care anymore :eek:
Posted on Reply
#8
hat
Enthusiast
Right... and Navi is still nowhere to be seen. Though I had no idea these were named after stars. When I hear Polaris, I think of our own @Polaris573, and nukes (thanks Megadeth). When I hear Vega... I think Scolar Visari, for some reason... and when I hear Navi, a number of things come to mind, including the North Star, navigation, and Navi from Zelda: Ocarina of Time.
Posted on Reply
#9
TheGuruStud
hat said:
Right... and Navi is still nowhere to be seen. Though I had no idea these were named after stars. When I hear Polaris, I think of our own @Polaris573, and nukes (thanks Megadeth). When I hear Vega... I think Scolar Visari, for some reason... and when I hear Navi, a number of things come to mind, including the North Star, navigation, and Navi from Zelda: Ocarina of Time.
With the Sony talk surrounding Navi, I'm not even sure it's anything but a low power, cheap die to go into the PS5 for Fall launch. Sure, they can sell it on desktop as a Vega replacement, but that doesn't do anything performance-wise (as rumored). That means there's no reason for it to be released earlier. Also, AMD needs wafers for Zen 2 as they will continue to sell more and more.

Glofo screwing them, again, just pushes back desktop parts even more. Enterprise is increasingly taking priority for cash (Lisa's game plan).

Turding is a joke, so I don't think AMD cares. Vegas/580s will continue to sell like hotcakes as long as prices are low. Nvidia has priced themselves out of the market lol.
Posted on Reply
#10
hat
Enthusiast
TheGuruStud said:
With the Sony talk surrounding Navi, I'm not even sure it's anything but a low lower, cheap die to go into the PS5 for Fall launch. Sure, they can sell it on desktop as a Vega replacement, but that doesn't do anything performance-wise (as rumored). That means there's no reason for it to be released earlier. Also, AMD needs wafers for Zen 2 as they will continue to sell more and more.

Glofo screwing them, again, just pushes back desktop parts even more. Enterprise is increasingly taking priority for cash (Lisa's game plan).

Turding is a joke, so I don't think AMD cares. Vegas/580s will continue to sell like hotcakes as long as prices are low. Nvidia has priced themselves out of the market lol.
Well... just as a single design can be segmented across an entire product stack, like nVidia Pascal (that is, all chips ranging from GP100 through GP108), I'm sure Navi can be scaled up and down to meet both the requirements of the PS5 as well as a kickass graphics card. Long gone are the days where consoles are made up of totally custom hardware designed specifically for that console. Instead, they take existing hardware, tweak it to fit their design and use it. As of now, we can only wait to see how exactly AMD (and TSMC I suppose, or GloFo, whoever) handles this.

I'm not sure where GloFo stands here. It seems like they've given up on 7nm and have relinquished themselves to older process nodes. Judging from comments I've read over the years, GloFo was never really good for parts where high performance was critical, anyway, and now they're even further behind. So, that leaves a lot of weight on TSMC.

As for Turding, :laugh: while it does offer a tangible performance increase over Pascal, I do agree that the prices are way too high. The cheapest RTX2080 in Newegg is about a Whopper meal away from the $800 mark, and that's not even the top tier card. The cheapest 2080Ti could get you a half decent vehicle (from a private seller, anyway)! Used to be you could build an entire decent gaming PC for around that price. nVidia is acting like they're in Breaking Bad in the high performance segment - they produce the best product, and there is no competitor. And I don't even know what to think of RTX (yet). So far, it seems like the most opportune time for nVidia to experiment with not only a new technology in graphics, but also to see how high they can push prices, all while clearing out as of yet unsold Pascal stock (at lower prices, lower than Turing anyway). At the same time, though, they're ripe for a good :nutkick:from AMD, or possibly even Intel, if either one can deliver early enough.
Posted on Reply
#11
Frick
Fishfaced Nincompoop
NC37 said:
It's all fun and games till some crazy engineer names one betelgeuse...
Ix would be a good name
Posted on Reply
#12
efikkan
hat said:
Well... just as a single design can be segmented across an entire product stack, like nVidia Pascal (that is, all chips ranging from GP100 through GP108), I'm sure Navi can be scaled up and down to meet both the requirements of the PS5 as well as a kickass graphics card.
Well, it's possible to craft a GPU design that can be easily scaled down, but I've not yet seen any design that can be scaled up later.
What's worrying about Navi is that Sony is sponsoring the development and AMD stated previously this year that it was targeting "Vega level performance".

hat said:

As for Turding, :laugh: while it does offer a tangible performance increase over Pascal
If an architecture offering 39% performance gains and 18% gains in energy efficiency is a turd, then I guess almost every past generation is as well. Not to mention the piles of cow dung AMD have been making lately…:rolleyes:
Posted on Reply
#13
Caring1
efikkan said:
If an architecture offering 39% performance gains and 18% gains in energy efficiency is a turd, then I guess almost every past generation is as well...… :rolleyes:
I would be happier if those figures were reversed.
Posted on Reply
#14
HTC
A turd in performance? Absolutely not.

A turd in price? Abso-fvcking-lutely yes!!!!!

EDIT

Unless AMD comes up with a way to have Zen like chips to form a single GPU, i don't see AMD catching up to nVidia in performance any time soon.
Posted on Reply
#15
NC37
R0H1T said:
And then I googled Betelgeuse aka Alpha Orionis, surprisingly Arcturus aka Alpha Boötis sounds mildly better :toast:
Man I miss the 80s...

Please tell me there's another cool person in here who got the reference!
Posted on Reply
#16
efikkan
HTC said:
A turd in performance? Absolutely not.

A turd in price? Abso-fvcking-lutely yes!!!!!
GTX 2080 Ti is way too expensive, but GTX 2080 is not too bad, shave off $50 and it's just fine. Condemning a whole generation due to the pricing of one product is not justified.
Also, let's see if the price is tweaked after Pascal runs out.

Don't forget that AMD tried to sell Vega at $100 over advertised MSRP.
Posted on Reply
#17
R0H1T
hat said:
Well... just as a single design can be segmented across an entire product stack, like nVidia Pascal (that is, all chips ranging from GP100 through GP108), I'm sure Navi can be scaled up and down to meet both the requirements of the PS5 as well as a kickass graphics card. Long gone are the days where consoles are made up of totally custom hardware designed specifically for that console. Instead, they take existing hardware, tweak it to fit their design and use it. As of now, we can only wait to see how exactly AMD (and TSMC I suppose, or GloFo, whoever) handles this.

I'm not sure where GloFo stands here. It seems like they've given up on 7nm and have relinquished themselves to older process nodes. Judging from comments I've read over the years, GloFo was never really good for parts where high performance was critical, anyway, and now they're even further behind. So, that leaves a lot of weight on TSMC.

As for Turding, :laugh: while it does offer a tangible performance increase over Pascal, I do agree that the prices are way too high. The cheapest RTX2080 in Newegg is about a Whopper meal away from the $800 mark, and that's not even the top tier card. The cheapest 2080Ti could get you a half decent vehicle (from a private seller, anyway)! Used to be you could build an entire decent gaming PC for around that price. nVidia is acting like they're in Breaking Bad in the high performance segment - they produce the best product, and there is no competitor. And I don't even know what to think of RTX (yet). So far, it seems like the most opportune time for nVidia to experiment with not only a new technology in graphics, but also to see how high they can push prices, all while clearing out as of yet unsold Pascal stock (at lower prices, lower than Turing anyway). At the same time, though, they're ripe for a good :nutkick:from AMD, or possibly even Intel, if either one can deliver early enough.
They're in the Empire business, comrade (or was it comprende) :pimp:
NC37 said:
Man I miss the 80s...

Please tell me there's another cool person in here who got the reference!
Sorry I'm not that old, am from the 80's but late 80's :ohwell:
Posted on Reply
#18
hat
Enthusiast
efikkan said:
Well, it's possible to craft a GPU design that can be easily scaled down, but I've not yet seen any design that can be scaled up later.
What's worrying about Navi is that Sony is sponsoring the development and AMD stated previously this year that it was targeting "Vega level performance".
Well, generally, you produce the top end product first, and you scale down, either by design or by "accident" when you get defects in your top end products.

efikkan said:
If an architecture offering 39% performance gains and 18% gains in energy efficiency is a turd, then I guess almost every past generation is as well. Not to mention the piles of cow dung AMD have been making lately…:rolleyes:
I take no issue with the raw performance of Turing. That, it does well. I take issue with the price, though. I'd call it good if they slashed prices in half. If they can't because it's too expensive to make, then that's nVidia's failure for producing a product that's too expensive.

efikkan said:
GTX 2080 Ti is way too expensive, but GTX 2080 is not too bad, shave off $50 and it's just fine. Condemning a whole generation due to the pricing of one product is not justified.
Also, let's see if the price is tweaked after Pascal runs out.

Don't forget that AMD tried to sell Vega at $100 over advertised MSRP.
The cheapest 2080 is a Whopper meal short of $800 at Newegg. That's just stupid. All the prices are too damn high. Shave off $50? If they shaved off $300 I'd still call it too expensive. The 2080Ti maybe should be $500. I still wouldn't buy it, but I'd at least be okay with the price being where it is. And while I do remember the Vega debacle, that won't stop me from calling a spade a spade. The 9900k is a great chip, for example, but the estimated price of $450 is insane. Turing prices are just lunacy. Nobody wants to get into PC gaming when a good graphics card ALONE (let's pick 2080 at $800) costs 3x as much as an entire console.
Posted on Reply
#19
Caring1
NC37 said:
Man I miss the 80s...

Please tell me there's another cool person in here who got the reference!
Not sure about cool, but that is a classic.
Posted on Reply
#20
efikkan
hat said:

The cheapest 2080 is a Whopper meal short of $800 at Newegg. That's just stupid.
Then blame the stupid shops for that, it's not Nvidia taking a premium.
Posted on Reply
#21
hat
Enthusiast
efikkan said:
Then blame the stupid shops for that, it's not Nvidia taking a premium.
I don't know who to blame, really. There's a lot of factors at play which aren't limited to nVidia alone... but I definitely know the price is too high. When I see that price, I do the same thing described in the immortal words of somebody somewhere on the internet who didn't like the Xbox 360: turn 360 degrees and walk away.

Posted on Reply
#22
Frick
Fishfaced Nincompoop
hat said:
I don't know who to blame, really. There's a lot of factors at play which aren't limited to nVidia alone... but I definitely know the price is too high. When I see that price, I do the same thing described in the immortal words of somebody somewhere on the internet who didn't like the Xbox 360: turn 360 degrees and walk away.


And the immortal reply is "that means you walk towards it". I don't know how many layers of irony and meta we're at now, but probably too many.
Posted on Reply
#23
efikkan
hat said:
I don't know who to blame, really.
Really? It should be fairly well known how this works, it works the same way as with most other goods (TVs, cars, hair dryers, etc.). The product maker sells a product at a certain price to wholesalers along with a suggested MSRP for the product. And then the wholesalers and the retailers can choose to sell it above or below that MSRP, depending on if they want to make more or less money per product. The whole price spike on GPUs we saw last year and early this year was a spike on the retail side.
Posted on Reply
#24
R0H1T
efikkan said:
Really? It should be fairly well known how this works, it works the same way as with most other goods (TVs, cars, hair dryers, etc.). The product maker sells a product at a certain price to wholesalers along with a suggested MSRP for the product. And then the wholesalers and the retailers can choose to sell it above or below that MSRP, depending on if they want to make more or less money per product. The whole price spike on GPUs we saw last year and early this year was a spike on the retail side.
Yes & that was the retailers, in some cases the wholesaler, being greedy. In this case however if RTX models are still selling above their "suggested MRP" then it boils down to 2 more plausible scenarios ~ they don't have enough margins on RTX (% wise or perhaps due to lower volume) &/or they're having problems clearing the inventories of the erstwhile Pascal lineup. The latter was mentioned by more than one reputable source IIRC, so no this pricing fiasco is totally on Nvidia this time around & miners can't take the blame for this!
Posted on Reply
#25
theoneandonlymrk
efikkan said:
Really? It should be fairly well known how this works, it works the same way as with most other goods (TVs, cars, hair dryers, etc.). The product maker sells a product at a certain price to wholesalers along with a suggested MSRP for the product. And then the wholesalers and the retailers can choose to sell it above or below that MSRP, depending on if they want to make more or less money per product. The whole price spike on GPUs we saw last year and early this year was a spike on the retail side.
Yet you blamed Amd for +£100 msrp a few posts ago, Nvidias set their prices too high to start with.
Posted on Reply
Add your own comment