[Videocardz rumour] Vega pushed forward to October

[Aristotelian]

If you're going to wait for something better, you'll be stuck waiting. Now, I'm not advocating you buying a particular brand/product, but if this post is any kind of reflection of your train of thought - you'll regret whatever you do, to some degree.

Just buy something. Anything. Enjoy it. And don't let the "well if I waited x-months I could have gotten this" because it's just going to keep repeating. I mean, even if get 1080 Ti - the 980 Ti itself was not a full GM200 chip, there could have always been a 980 Ultra. Just food for thought.

I don't recall this happening with graphics cards for a long time, and I'm talking from way back in the TNT2 Ultra days to the Geforce 256, then the DDR etc. Vega is meant to be the pure flagship of AMD's generation - when has it happened before that a few months after a fairly expensive mid-range card from Nvidia was released, a flagship from AMD that stomped it was released? Maybe I've been out of the loop for a long time but let's take an example:

I'll want to play Mass Effect Andromeda at 4k and HDR if possible - in early 2017 when the game is released. Nothing out yet right now (including monitors) fits that. But I was hoping to buy the PC now and upgrade my monitor around christmas when the Dell OLED 4k comes out. At this rate, I'll just shelve this damn build and wait for the damn flagships I guess. Bleh.

 

[moonbogg]

I don't recall this happening with graphics cards for a long time, and I'm talking from way back in the TNT2 Ultra days to the Geforce 256, then the DDR etc. Vega is meant to be the pure flagship of AMD's generation - when has it happened before that a few months after a fairly expensive mid-range card from Nvidia was released, a flagship from AMD that stomped it was released? Maybe I've been out of the loop for a long time but let's take an example:

I'll want to play Mass Effect Andromeda at 4k and HDR if possible - in early 2017 when the game is released. Nothing out yet right now (including monitors) fits that. But I was hoping to buy the PC now and upgrade my monitor around christmas when the Dell OLED 4k comes out. At this rate, I'll just shelve this damn build and wait for the damn flagships I guess. Bleh.

It started happening with Geforce 600 series. Mid range cards are supposed to come after the flagship so that the masses can buy a nice card for much less cost than what the enthusiasts spent on the big chip. But this is how things are now, so we simply have to change our mindset.
Now, I simply ignore mid range parts, period. The way I see it, you can either be a mid range buyer and focus on those releases, or a high end buyer and focus on those releases, OR you can buy them both one after the other non stop. However, you are still paying full blown flag ship price for the mid range cards, so you may as well ignore them and wait for the big chip. Throw your $700 per GPU at big pascal, not the mid range IMO. If you are spending $700 per GPU anyway it makes no sense to buy the mid range at such a price when we all know a big chip is coming. Nvidia hopes you buy them both for $700 year after year.
Vega coming early would be awesome. It would shake things up and give us a high end card sooner rather than later. I hope the rumor is true.

 

[railven]

I don't recall this happening with graphics cards for a long time, and I'm talking from way back in the TNT2 Ultra days to the Geforce 256, then the DDR etc. Vega is meant to be the pure flagship of AMD's generation - when has it happened before that a few months after a fairly expensive mid-range card from Nvidia was released, a flagship from AMD that stomped it was released? Maybe I've been out of the loop for a long time but let's take an example:

I'll want to play Mass Effect Andromeda at 4k and HDR if possible - in early 2017 when the game is released. Nothing out yet right now (including monitors) fits that. But I was hoping to buy the PC now and upgrade my monitor around christmas when the Dell OLED 4k comes out. At this rate, I'll just shelve this damn build and wait for the damn flagships I guess. Bleh.

Those days are long gone. Welcome to the new world order of a one-sided duoopoly.

NV has little reason to give us the big chip up front if they can keep the lead with their medium chip.

 

[Fallen Kell]

Hell yes! competition working the way it should be.

If it was competition working the way it should it would be AMD making this announcement, not random poster on random website forum.

Welp, no

There is a slide from eSilicon presentation for 14nm LPP HBM2 integration - test chip tepe out at March 2016 - Final Models availability at September 2016

//And there are some other ESes circling the world in Zauba with higher price then Polaris ones.

And that slide don't show anything about Vega being out in October/November. The only thing it shows is that HBM2 is slated to finally be AVAILABLE in September. And by that, it simply means they will finally start basic production. In no way does that mean we will have cards using it the following month. They would still need to do a final test with the production parts with the cards that use it, and then somehow stockpile at least 10-20k parts for use on their cards across all their OEM partners.

If that slide said HBM2 was available in July-August, I might agree with you that we could be seeing October release. If AMD pushes the release up early, it will be with just a couple hundred cards across the entire world... Good luck getting one unless you like spending 2-3x the retail price for it on ebay.

 

[Glo.]

Final Models means: qualification silicon.

 

[caswow]

http://www.aida64.com/downloads/OGJlZDU5MDM=

Version: 5.70.3843 beta (May 11, 2016)
Release notes:
preliminary GPU information for AMD Polaris 10 (Ellesmere)
preliminary GPU information for AMD Polaris 11 (Baffin)
preliminary GPU information for AMD Vega 10 (Greenland)
preliminary GPU information for nVIDIA GeForce GTX 1080 (GP104)
improved memory type detection for AMD GPUs
Intel Processor Number detection for Pentium D1519
Intel Processor Number detection for Xeon D-1529, D-1539, D-1559
extended sensor support for AquaStream Ultimate
fixed: motherboard specific sensor info for MSI MS-7881, 7882, 7883, 7885 (too low Vcore)

No GP106 but Vega10? Something smells fishy right guy?

 

[Kenmitch]

http://www.aida64.com/downloads/OGJlZDU5MDM=



No GP106 but Vega10? Something smells fishy right guy?

No GTX 1070 listed. Was it delayed?

 

[Glo.]

BY the looks of things the support is for full dies. There is nothing more needed, if GTX 1070 is cut down part from GP104.

improved memory type detection for AMD GPUs

That is much more important, my friends.

 

[Arachnotronic]

http://www.aida64.com/downloads/OGJlZDU5MDM=



No GP106 but Vega10? Something smells fishy right guy?

AIDA64 added preliminary support for Cannonlake, Goldmont, and Skylake-E/EN/EP/EX processors mid-last year. It will be more than a year before the first of these products makes it to market.

http://www.aida64.com/news/pr/aida64-v530-press-release

I wouldn't read into that for launch timing guidance one way or another.

 

[MrTeal]

BY the looks of things the support is for full dies. There is nothing more needed, if GTX 1070 is cut down part from GP104.
That is much more important, my friends.

The improved memory type detection might not be for the new GPUs, but resolving many of the ridiculous bugs AIDA has detecting current GPUs. It kept telling me my 290s had GDDR3.

 

[maddie]

If it was competition working the way it should it would be AMD making this announcement, not random poster on random website forum.

Originally Posted by Krteq
Welp, no

There is a slide from eSilicon presentation for 14nm LPP HBM2 integration - test chip tepe out at March 2016 - Final Models availability at September 2016

This image has been resized. Click this bar to view the full image. The original image is sized 800x409.

//And there are some other ESes circling the world in Zauba with higher price then Polaris ones.

And that slide don't show anything about Vega being out in October/November. The only thing it shows is that HBM2 is slated to finally be AVAILABLE in September. And by that, it simply means they will finally start basic production. In no way does that mean we will have cards using it the following month. They would still need to do a final test with the production parts with the cards that use it, and then somehow stockpile at least 10-20k parts for use on their cards across all their OEM partners.

If that slide said HBM2 was available in July-August, I might agree with you that we could be seeing October release. If AMD pushes the release up early, it will be with just a couple hundred cards across the entire world... Good luck getting one unless you like spending 2-3x the retail price for it on ebay.

That slide is for the PHY not HBM2. It is the base layer of a HBM stack.

You can use the 28nm parts with Vega. That is one of the advantages of transposers. Mix and match separate processes and nodes on the same transposer.

There is nothing preventing a company from using a 28nm PHY with the DDR dies to assemble a stack and then using these stacks with a 14nm GPU. They can upgrade the PHY when 14nm becomes available.


Trick question. Which PHY do you think GP100 is using?

 

[Ranulf]

This is becoming a difficult time for people like me (and I think there are a few) that want to build a nice PC this year. Sooner rather than later. If I went 1080 Tri SLI watercooled and a top Vega with 16 HBM2 came out where say, 1 Vega was as fast as 2 1080s (roughly), I'd be an idiot buying the 1080s in June with such a performance increase coming soon after. And yet, for the past while, it's always (in my mind) "The 980Ti is a great card, but Broadwell-E isn't out. Wait for Broadwell-E. The 1080s will be out when the 6950X is, so get those and watercool them, get a new monitor - but there's no HDR 4k monitors available yet. And worse, the 1080s will be stomped either by a 1080Ti or Vega, or whatever the name of Nvidia's HBM 2 monster Titan will be, so...wait".

And I sit here on a 2600k overclocked with 580s in SLI playing at a peasant 1440p as I have been since 2010 or so, and in modern games I dial it down to 1080p :'(

Please, to the gods that run these companies, put out the blistering performance parts now! Didn't any of them want to catch those of us building new broadwell-E systems with top end parts that will not make us wonder why we dropped thousands of dollars on mid range parts NOW? Please!

Meh, its been this way forever. Sometimes you win, sometimes you lose in the price/perf game.

 

[Sweepr]

Amazing how a single forum post from a guy on 3DCenter is the source of this "news." We'll see if user "RIP" on 3DCenter is right or not soon enough.

According to another forum post there is something coming from AMD in October, and it's not Vega.

 

[Maverick177]

If it's not Vega then it has to be Polaris, which is highly unlikely.

 

[guskline]

This is becoming a difficult time for people like me (and I think there are a few) that want to build a nice PC this year. Sooner rather than later. If I went 1080 Tri SLI watercooled and a top Vega with 16 HBM2 came out where say, 1 Vega was as fast as 2 1080s (roughly), I'd be an idiot buying the 1080s in June with such a performance increase coming soon after. And yet, for the past while, it's always (in my mind) "The 980Ti is a great card, but Broadwell-E isn't out. Wait for Broadwell-E. The 1080s will be out when the 6950X is, so get those and watercool them, get a new monitor - but there's no HDR 4k monitors available yet. And worse, the 1080s will be stomped either by a 1080Ti or Vega, or whatever the name of Nvidia's HBM 2 monster Titan will be, so...wait".

And I sit here on a 2600k overclocked with 580s in SLI playing at a peasant 1440p as I have been since 2010 or so, and in modern games I dial it down to 1080p :'(

Please, to the gods that run these companies, put out the blistering performance parts now! Didn't any of them want to catch those of us building new broadwell-E systems with top end parts that will not make us wonder why we dropped thousands of dollars on mid range parts NOW? Please!

When Broadwell E 10 core is released, make your move. Buy a single 1080 until the big boys drop and switch out.:thumbsup:

 

[Sweepr]

So no threads based on forum posts for now. Will this be closed as well?

http://forums.anandtech.com/showpost.php?p=38218684&postcount=31

 

[Rvenger]

So no threads based on forum posts for now. Will this be closed as well?

http://forums.anandtech.com/showpost.php?p=38218684&postcount=31

If you have a problem with the moderation take it to the Moderator discussions forum. Not here.

This was a Videocardz link, not a forum post. We are not here to dispel every rumor that gets posted by someone. If it gets out of hand, I will start locking the threads.

-Rvenger

 

[Sweepr]

This was a Videocardz link, not a forum post.

Videocardz's source here is a 3DCenter article based on forum posts.

www.3dcenter.org/news/amd-zieht-den-vega-launch-angeblich-auf-oktober-2016-vor

 

[Vesku]

And that slide don't show anything about Vega being out in October/November. The only thing it shows is that HBM2 is slated to finally be AVAILABLE in September. And by that, it simply means they will finally start basic production. In no way does that mean we will have cards using it the following month. They would still need to do a final test with the production parts with the cards that use it, and then somehow stockpile at least 10-20k parts for use on their cards across all their OEM partners.

If that slide said HBM2 was available in July-August, I might agree with you that we could be seeing October release. If AMD pushes the release up early, it will be with just a couple hundred cards across the entire world... Good luck getting one unless you like spending 2-3x the retail price for it on ebay.

GDDR5X is in a similar state now and Nvidia is committing to launching GDDR5X using 1080 cards in by the end of May.

 

[C@mM!]

That slide is for the PHY not HBM2. It is the base layer of a HBM stack.

You can use the 28nm parts with Vega. That is one of the advantages of transposers. Mix and match separate processes and nodes on the same transposer.

There is nothing preventing a company from using a 28nm PHY with the DDR dies to assemble a stack and then using these stacks with a 14nm GPU. They can upgrade the PHY when 14nm becomes available.


Trick question. Which PHY do you think GP100 is using?

Also, eSillicon uses TSMC for manufacture. From what I understand, AMD is using Samsung fabs on an already established 14nm process for its HBM2 supply. This eSilicone article probably affects NVidia's supply of HBM2 though, hence why there is nothing in the wild about big Pascal.

 

[atticus14]

Those days are long gone. Welcome to the new world order of a one-sided duoopoly.

NV has little reason to give us the big chip up front if they can keep the lead with their medium chip.

I think they just want to sell to the big spenders twice. If a 1080 can beat a 980ti, even by a little, some of those buyers will make the switch no matter the cost, then they'll line up again for the real high end in another 6 months.

 

[ShintaiDK]

Videocardz's source here is a 3DCenter article based on forum posts.

www.3dcenter.org/news/amd-zieht-den-vega-launch-angeblich-auf-oktober-2016-vor

Yep, its all based on a single random forum post without any information at all.

 

[Tuna-Fish]

The only thing it shows is that HBM2 is slated to finally be AVAILABLE in September.

That's not what the slide means at all. The slide is about when a tested PHY is available. That is a piece of IP that goes on the GPU die.

 

[Tuna-Fish]

That slide is for the PHY not HBM2. It is the base layer of a HBM stack.

Not the base layer. This part goes on the GPU/CPU die.

 

[maddie]

Not the base layer. This part goes on the GPU/CPU die.

Actually, there is a PHY in the base layer also.

It's possible AMD is using this IP in their GPUs also. Don't know.