Polaris 11 - 1500€ customs value card spotted! [zauba listing info&3dcenter info]

[Tic-Tac]

Let the games begin!

 

https://www.zauba.com/import-102-c99398-hs-code.html

 

http://www.3dcenter.org/news/ein-weiteres-lebenszeichen-von-amds-polaris-generation-mit-einer-1500-euro-karte

 

Next-gen AMD GPU's Polaris info finally showing up! and it's with HBM2!

 

Related news backing this up are Samsung's mass productions of HBM-Fastest DRAM : https://news.samsung.com/global/samsung-begins-mass-producing-worlds-fastest-dram-based-on-newest-high-bandwidth-memory-hbm-interface

 

 

 

Sources:

 

https://www.zauba.com

http://www.3dcenter.org/

[byalexandr]

This is old news. Polaris was being shipped around long before February 1st.

[Tic-Tac]

Those 3Dcenter news are from yesterday though.

[Link4]

I think those codenames are incorrect though, C981 should be Polaris 11 and C920s Polaris 10. AMD already had Polaris 11 chips at Computex and the Zauba data for C981 matches it. C990s are most likely Vega 10 as that is the high end GPU and it will launch later than the Polaris chips.

[Notional]

Considering AMD showcased a working model back in December, this is hardly news. Also since Samsung just announced production of HBM 2 this January, after the Polaris test cards were out, and the fact that AMD has priority access to SK Hynix' HBM (due to their partnership in inventing the damn thing), I'd say it's safe to say that Polaris uses SK Hynix modules, not Samsung's.

 

I am however as excited as you are OP

[Watermelon Guy]

33 minutes ago, Link4 said:

I think those codenames are incorrect though, C981 should be Polaris 11 and C920s Polaris 10. AMD already had Polaris 11 chips at Computex and the Zauba data for C981 matches it. C990s are most likely Vega 10 as that is the high end GPU and it will launch later than the Polaris chips.

AMD are talking about polaris being "VR" ready. Meaning Better than 970/290x performance. 

[Doobeedoo]

Somewhat expected, but 1500€ Q.Q

[rattacko123]

I think the 1500 Euro card is the dual fury X

the 650 Euro card is likely the new high end polaris

[That Norwegian Guy]

Shipping manifests are what they are insured for, not what it'll cost at Newegg /SMH

[Tic-Tac]

It's customs value of that card and not the actual market value. In Germany it's a range from 17% and all the way up to 48.5% (anti dumping policy). So an estimated market price of Polaris 11 would be somewhere around 770€.

[CTR640]

Those prices are getting fucking outrageous now.

[Tic-Tac]

Well, all I know is that those new GPU's Polaris and Pascal will NOT be any cheaper than GTX 980 Ti for example, that's for sure. Combine that with 1200+ $/€ monitors and you get yourself a decent small car! LOL

[Miguel Batista]

Well, all I know is that those new GPU's Polaris and Pascal will NOT be any cheaper than GTX 980 Ti for example, that's for sure. Combine that with 1200+ $/€ monitors and you get yourself a decent small car! LOL

True. No reason to believe this new gen will drive prices down... Quite the opposite I think. With all these technologies, prices could go up. There are no price shortcuts... You want performance, you pay performance.

[Valentyn]

13 hours ago, Doobeedoo said:

Somewhat expected, but 1500€ Q.Q

Could easily be a dual GPU card, or the much much needed new FirePro line. Now that AMD have access to CUDA, it makes sense for them to push out the new cards.

 

Either that, or AMD is going to try and pre-empt NVIDIA with as halo card similar to the Titan series. If it's true that NVIDIA won't have any offerings until the second half of the year AMD could gain some traction.

[samcool55]

Well, fuck. That's all i can say.

As i said before, the price will decide if polaris is a succes or not. Performance won't be an issue.

 

We all know polaris is going to be a huge step, FINfet, 16nm, and the little we know looks really promising.

So performance will probably we awesome. But if it is going to cost 1500 bucks, does it even matter?

I mean it's waaay above most people's budget. and even if it performs like a monster, it's not going to be a succes at that price, it just won't.

 

If that is the real price, we are screwed, we just are. Almost nobody is going to buy it, and for most of us it would just give us some nice reviews on websites, but almost nobody will end up getting it.

[Texhnlyze]

Can't they come any sooner?!! My wallet is ready!

 

Pascal needs to come sooner too so I don't have any regrets when it comes available and becomes more powerful than Polaris.

[SurvivorNVL]

58 minutes ago, Texhnlyze said:

Can't they come any sooner?!! My wallet is ready!

 

Pascal needs to come sooner too so I don't have any regrets when it comes available and becomes more powerful than Polaris.

I am still hoping those Titans are out in April, if so - goodbye $2000 and hello Pascal Titan SLI.

[looncraz]

On 2/7/2016 at 7:06 PM, samcool55 said:

Well, fuck. That's all i can say.

As i said before, the price will decide if polaris is a succes or not. Performance won't be an issue.

 

We all know polaris is going to be a huge step, FINfet, 16nm, and the little we know looks really promising.

So performance will probably we awesome. But if it is going to cost 1500 bucks, does it even matter?

I mean it's waaay above most people's budget. and even if it performs like a monster, it's not going to be a succes at that price, it just won't.

 

If that is the real price, we are screwed, we just are. Almost nobody is going to buy it, and for most of us it would just give us some nice reviews on websites, but almost nobody will end up getting it.

 

First, and vitally, this is NOT the retail price - it's a value assigned to a limited-run pre-production part that is probably already partly over-inflated in value.  It is MUCH less than the declared value AMD made for Fiji GPUs - and those came out at $650 on the market. This GPU could easily be a $650 on the market... or less... in fact, $1500 is CHEAP... which suggests AMD made a larger production run, sent them to their Canada HQ for driver development, and now has sent just a couple to India for more testing and development.

 

Next, Polaris is believed to be 14nm FinFet now - giving AMD a (minor) process advantage over nVidia - but also giving them two foundries from which to choose, making it far more likely that they may be able to capitalize on any lead to market (get more sales, higher volumes, and keep inventories flush - probably pre-planning a reduction in volume for when Pascal releases).

[Prysin]

On 2/5/2016 at 0:50 AM, Watermelon Guy said:

AMD are talking about polaris being "VR" ready. Meaning Better than 970/290x performance. 

not really "hard" to achieve.

on average, a die shrink has yielded 20-30% performance increase (flat increase) over older generation. So if we substract 30% from GTX 970 you end up around GTX 950/R9 270X...

 

Also, GTX 970 is not remotely comparable with R9 290X these days. 290X is almost at 390X levels which is almost 980. Only a heavily OCd 970 will match a 290X. I know the "X" and "non X" can be confusing. but really, atleast TRY to keep things apart.

[Prysin]

38 minutes ago, looncraz said:

 

First, and vitally, this is NOT the retail price - it's a value assigned to a limited-run pre-production part that is probably already partly over-inflated in value.  It is MUCH less than the declared value AMD made for Fiji GPUs - and those came out at $650 on the market. This GPU could easily be a $650 on the market... or less... in fact, $1500 is CHEAP... which suggests AMD made a larger production run, sent them to their Canada HQ for driver development, and now has sent just a couple to India for more testing and development.

 

Next, Polaris is believed to be 14nm FinFet now - giving AMD a (minor) process advantage over nVidia - but also giving them two foundries from which to choose, making it far more likely that they may be able to capitalize on any lead to market (get more sales, higher volumes, and keep inventories flush - probably pre-planning a reduction in volume for when Pascal releases).

density wise, 14nm GLOFO/Samsung = 16nm TSCM

the thing about these "nodes" are that just like TDP rating, everyone is doing it ever so slightly differently. For instance, IBMs 10nm is smaller and denser then intels 10nm. but due to how intel does their process, there isnt any real performance edge. In practice, its the same. But once you start going into the fine details and start nitpicking, its not. Same for GLOFO/Samsung 14nm vs TSCM 16nm.

 

the more you know

 

 

[Notional]

33 minutes ago, Prysin said:

density wise, 14nm GLOFO/Samsung = 16nm TSCM

the thing about these "nodes" are that just like TDP rating, everyone is doing it ever so slightly differently. For instance, IBMs 10nm is smaller and denser then intels 10nm. but due to how intel does their process, there isnt any real performance edge. In practice, its the same. But once you start going into the fine details and start nitpicking, its not. Same for GLOFO/Samsung 14nm vs TSCM 16nm.

 

the more you know

 

 

 

 

This is of course Intel's claims "true 14nm". But bear in mind that Samsungs/GloFo's SRAM should be 3D and thus better than Intel's.

 

Either way this year is going to get interesting. We know launch dates of AMD. The question is when NVidia will launch anything. The April number still sounds impossible as Samsung just started HBM production this January. I hope for AMD to get a leg up. This insane market bias against AMD is bad against said market.

[looncraz]

1 hour ago, Prysin said:

density wise, 14nm GLOFO/Samsung = 16nm TSCM

the thing about these "nodes" are that just like TDP rating, everyone is doing it ever so slightly differently. For instance, IBMs 10nm is smaller and denser then intels 10nm. but due to how intel does their process, there isnt any real performance edge. In practice, its the same. But once you start going into the fine details and start nitpicking, its not. Same for GLOFO/Samsung 14nm vs TSCM 16nm.

 

the more you know

 

 

 

I read your comment and went to go find the very chart Notional posted

 

They're certainly close, but 14nm still has the advantage (I specifically mentioned the advantage was minor as well).

 

Density wise, the advantage is the probably smallest.  14nm LPP has the advantage for leakage and switching speed as well, IIRC. It may only give AMD a 5 or 10W benefit over using 16nm @ TSMC, but combined with dual-sourcing, slightly better density, and AMD's long-standing relationship (and wafer commitment) with Global Foundries, it makes the most sense - while also providing a competitive edge.