Vega/Navi Rumors (Updated)

[parvadomus]

I expect Vega10 to be 2xPolaris 10 in terms of resources: 64 Rops/4608 SP, that would follow the same basic strategy they did with Tonga-> Fiji
It should be very close to Titan XP (this is exactly 2xGP106), and about 20% better than 1080.
Obviously AMD will wait until nvidia launches 1080Ti to adjust its price acordingly..

 

[Dave2150]

I expect Vega10 to be 2xPolaris 10 in terms of resources: 64 Rops/4608 SP, that would follow the same basic strategy they did with Tonga-> Fiji
It should be very close to Titan XP (this is exactly 2xGP106), and about 20% better than 1080.
Obviously AMD will wait until nvidia launches 1080Ti to adjust its price acordingly..

It may be many months before Vega's release - I expect the 1080ti to have been out at least 3-4 months before we see any Vega cards (HBM2, expensive and low supply).

 

[Arachnotronic]

My guess is May. Last year we got the first public demo of Polaris later in January and it released in June. This year we got a demo earlier in December so if thats any indication it seems like we are looking at a May release.

AMD was showing Polaris silicon working to the press in private demos back in December of 2015, AFAIK, so it's about the same timeline.

Even if the silicon is good to go, the drivers need to be all set, the board designs from the AIBs need to be good to go (or very close to it), and of course it takes a while to actually build the wafers. You put a wafer in today and it's not coming out for another three months. Add another 2-3 weeks for packaging and test (HBM2 packaging is not easy), and then you have the time that it takes to box and ship these cards around the world to the various PC OEMs/distributors.

So yeah, I'm expecting a Computex 2017 announcement with availability shortly thereafter.

 

[HurleyBird]

It was nice to see that Vega is up and running at the AMD Horizon event today (near the end of presentation). Hopefully, Vega launches around the same time as RyZen

I got the impression that they weren't very confident in Vega, or it would have been a good time to show it off more. Hope I'm wrong.

 

[swilli89]

AMD was showing Polaris silicon working to the press in private demos back in December of 2015, AFAIK, so it's about the same timeline.

Even if the silicon is good to go, the drivers need to be all set, the board designs from the AIBs need to be good to go (or very close to it), and of course it takes a while to actually build the wafers. You put a wafer in today and it's not coming out for another three months. Add another 2-3 weeks for packaging and test (HBM2 packaging is not easy), and then you have the time that it takes to box and ship these cards around the world to the various PC OEMs/distributors.

So yeah, I'm expecting a Computex 2017 announcement with availability shortly thereafter.

Earliest example I believe we have is January 2016. http://www.trustedreviews.com/news/amd-demos-14nm-finfet-polaris-gpu-confirms-mid-2016-release-date

I remember this is when we were first even told the Polaris name. I'm thinking we see Vega a bit earlier in the year compared to Vega.

 

[Arachnotronic]

Earliest example I believe we have is January 2016. http://www.trustedreviews.com/news/amd-demos-14nm-finfet-polaris-gpu-confirms-mid-2016-release-date

I remember this is when we were first even told the Polaris name. I'm thinking we see Vega a bit earlier in the year compared to Vega.

Tech press got to see it before we did:

But showing off 16nm pre-production parts is not a big deal, SemiAccurate held an AMD 14nm device in December. Why do we say that Nvidia has a massive lead over AMD then?

https://semiaccurate.com/2016/01/11/nvidia-pascal-over-a-year-ahead-of-1416nm-competition/

 

[swilli89]

Tech press got to see it before we did:



https://semiaccurate.com/2016/01/11/nvidia-pascal-over-a-year-ahead-of-1416nm-competition/

Amd showed Polaris working to the public in January 2016. AMD showed Vega working to the public in December 2016.

 

[Dave2150]

Amd showed Polaris working to the public in January 2016. AMD showed Vega working to the public in December 2016.

Correct. Though remember the RX480 is a rather 'simple' card to produce, compared to the logistical nightmare that 300W Vega will be. HBM2, interposer and TSV's etc are much more complicated, though AMD have hopefully gained great experience with their Fury lineup.

I wouldn't be at all surprised if Vega gets delayed and ends up launching after Volta.

 

[tential]

Correct. Though remember the RX480 is a rather 'simple' card to produce, compared to the logistical nightmare that 300W Vega will be. HBM2, interposer and TSV's etc are much more complicated, though AMD have hopefully gained great experience with their Fury lineup.

I wouldn't be at all surprised if Vega gets delayed and ends up launching after Volta.

That's essentially saying it will do worse than Fiji. That's not what I want to hear. I mean I know you're being real, but it's not making me happy.

I'm getting bored

 

[Dave2150]

That's essentially saying it will do worse than Fiji. That's not what I want to hear. I mean I know you're being real, but it's not making me happy.

I'm getting bored

It's not what I want to hear either, though as you say we have to be realists. I really can't see Vega launching before June.

 

[Bacon1]

It's not what I want to hear either, though as you say we have to be realists. I really can't see Vega launching before June.

They've already shown it to be working. Now they just need to work on mass production and final board / cooler design.

I just hope they don't release sub par cooling this time, or heck, just drop the reference all together and let the partners do their own cooling a'la 390 vs 290 release.

Many online reviewers keep those throttling, hot, power hungry launch day reference cards years later, please don't release them like that again AMD!

 

[Qwertilot]

That's essentially saying it will do worse than Fiji. That's not what I want to hear. I mean I know you're being real, but it's not making me happy.

I'm getting bored

It would be really, really bad if it comes after Volta - going by NV's typical kinds of performance increase, realistically even the biggest Vega is liable to be behind the ~1070 equivalent in Volta's range. Basically no point releasing it at that point.

I slightly doubt if it will though. NV have got all their P102 based chips to release - early Spring being the earliest likely date at this point now Xmas is gone - and they'll want to sell those for at least 6 months before releasing Volta. More likely they'll just go Spring 2017/2018 to keep the range improvements evenly spread out.

I guess the timelines for the two architectures might get close to crossing over for data center style chips which might turn out to be vega's main target market anyway.

 

[Dave2150]

They've already shown it to be working. Now they just need to work on mass production and final board / cooler design.

I just hope they don't release sub par cooling this time, or heck, just drop the reference all together and let the partners do their own cooling a'la 390 vs 290 release.

Many online reviewers keep those throttling, hot, power hungry launch day reference cards years later, please don't release them like that again AMD!

Please read the last few posts, this has already been discussed.

Polaris was shown working in January 2016. It released on the 29th of June, almost 6 months later. Polaris is a small GPU, and uses ordinary GDDR5. (Relatively simple graphics card).

Vega was indeed shown working a few days ago, December 2016. It is logical that it will take longer than Polaris took to reach release, due to a massively more complicated manufacturing process and thermal solution (Vega = most likely double Polaris' TDP).

I hope you're right and I'm wrong, and that Vega launches in January or February. My XL2730Z 1440P 144Hz Freesync monitor is ready and waiting, though I can't see it arriving before June.

 

[Ajay]

Well Dave2150, if that's what happens - it does suck. Not sure why AMD went with HBM again, puts more cost sensitivity into the product.

 

[IEC]

There's already a perfect cooling solution. It's the Sapphire Nitro Fury air cooler...

 

[Bacon1]

There's already a perfect cooling solution. It's the Sapphire Nitro Fury air cooler...

It is amazing!

 

[tential]

If you don't give me a Fury X reference design at no premium after Fury X JUST came out, I'll be pretty annoyed. REALLY annoyed actually. On top of that though, I'm tired of the Fury X only 1 cooler thing. AMD... don't piss us off. Let AIBs also make coolers too.
I certainly wanted Fury X WC, but man, sometimes you also just want to be able to throw in 2 air cooled cards quickly without having to redo your WHOLE cooling airflow in your case.
I want options, not being FORCED into something. Really wish those fans hadn't had to be mounted horizontally, I had them ready for vertical placement

 

[tajoh111]

If you don't give me a Fury X reference design at no premium after Fury X JUST came out, I'll be pretty annoyed. REALLY annoyed actually. On top of that though, I'm tired of the Fury X only 1 cooler thing. AMD... don't piss us off. Let AIBs also make coolers too.
I certainly wanted Fury X WC, but man, sometimes you also just want to be able to throw in 2 air cooled cards quickly without having to redo your WHOLE cooling airflow in your case.
I want options, not being FORCED into something. Really wish those fans hadn't had to be mounted horizontally, I had them ready for vertical placement

Fury X came with no premium upon release. It was just 100 dollars more than a regular fury and considering the substantial amount of disabled shaders and the water cooler, it was underpriced relative to it's cost to make. Most AIB cards with a water cooler cost 100 dollars more than standards cards. In AMD case, you got a water and the full chip for 100 dollars more. In the past, AMD charged 100-150 extra to get the full chip alone. No upgraded cooler and on much smaller chip which meant the yields between the full chip and cut chip were more similar. Because there was only 100 dollar difference, unless the form factor doesn't allow it, the fury x is a no brainer over a fury.

There was never going to be an AIB fury X because it doesn't make sense from a marketing standpoint.

A so called none premium fury X would be around 599 and at that point it destroys regular fury sales. There needs to be enough difference between the cards that 1 card doesn't cannibalize the sales of another card. With the standard pricing of fury x, it already cannibalized the sales of fury. Fury x were initially sold out for a while, fury were easy to get. And considering the die size of fury, the yields on a fury x were much worse than a regular fury. Getting 100% perfect die on something that's 600mm2 is not easy. So a 50 dollar price difference is too small, even 100 dollar price difference is too small when you take in the water cooling. But for AMD, the design of the fury x was finalized and AMD probably didn't expect to have to price fury x at 650.

The problem came with Nvidia's gtx 980 ti which upon release was 650 dollars. Remember those rumors of a 800 dollar fury x, there was likely some truth to them. But the gtx 980 ti came in so it got a price cut before it was officially released. The Fury and the fury nano have realistic price point in terms of profitability due to the lack of competition. Fury X on the other hand was underpriced but AMD was forced to sell it at a lower price.

 

[railroadmaster]

AMD is gonna build a wall and Nvidia is going to pay for it!

 

[guskline]

Vega will be released when AMD is ready for it. It really appears that "all hands are on deck" for ZEN right now.

From a pure profitability stand point that makes sense and perhaps lots of "cents" to AMD.

To those who want a true high end competitor to GTX1070/1080/soon 1080 TI and perhaps even Titan X Pascal AMD has what? Fury X? Fury? Nano?

Tough sell but patience is required or do what I did. Slap 2 RX480s in CF and dream about what a Vega would be like expet dual 480s have a number of drawbacks (i.e. more power consumption, no HBM/HBM2 limited OCing, most likely different gpu chip etc)

I suspect that though Vega appears now in a demo, it is still in rough form with the cooling, software etc being tweeked.

The Christmas rush is over, and as I said above, AMD is probably making the right move on focusing on ZEN.

 

[flopper]

Vega will be released when AMD is ready for it. It really appears that "all hands are on deck" for ZEN right now.



I suspect that though Vega appears now in a demo, it is still in rough form with the cooling, software etc being tweeked.

Note Vega was used to be presented in a live setting so even in 4 weeks that means they got good stuff coming back from the design. no respin needed is a good sign

 

[Sweepr]

Fudzilla says (cheaper) GDDR5 Vega variants are in the works - could be the rumored small Vega

www.fudzilla.com/news/graphics/42422-vega-architecture-will-bring-multiple-cards

 

[Headfoot]

I would not be surprised at all if they did a similar thing with Vega they did with the 4870/4850. The SKUs were not all that far apart except for the clocks and memory tech. I wouldn't be surprised if they do the same here, like the 1070 vs 1080. I could even see a 3 card spread: little vega + GDDR5, cut down big vega + gddr5x, big vega + hbm2

 

[96Firebird]

From what I've read, you can't just interchange GDDR5 with HBM.

 

[zinfamous]

From what I've read, you can't just interchange GDDR5 with HBM.

The HBM "interface" is essentially part of the GPU architecture, no?