Official Fury X specs

[AnandThenMan]

I thought the reason is thermal microbumps that could be damaged by overclocking and overvolting memory.
Breaking the microbump could fry a HBM quite easy.

I've never heard about this, where did you get this info?

 

[Erenhardt]

I've never heard about this, where did you get this info?

There was a rumor way back when that stacking dies is problematic due to thermal expansion of different materials and such...

http://semiaccurate.com/2012/09/06/die-stacking-has-promise-and-problems/

Running modules out of spec can do harm to the stack.
HBM works quite different to GDDR5.
http://www.hotchips.org/wp-content/...Bandwidth-Kim-Hynix-Hot Chips HBM 2014 v7.pdf
slide #14 and a couple of next slides.

HBM can read data, write data and refresh different memory dies at the same time. GDDR5 rely on a full stream from all modules at the same time to get enough bandwidth.
If in single HBM stack some dies are idle, and some are stressed there will be a temperature gradient between them. The more overclocked memory, the greater the gradient. Stressing a stack with too much tension from thermal expansion can damage the stack.

The last thing amd and hynix want is a mass failure rate of new, most innovative memory system of the decade.

 

[LTC8K6]

There was a rumor way back when that stacking dies is problematic due to thermal expansion of different materials and such...

http://semiaccurate.com/2012/09/06/die-stacking-has-promise-and-problems/

You are going to get pummeled for linking to those clowns. :whiste:

 

[3DVagabond]

I don't know if anyone has shown this pic yet?

Are they just talking about the interposer being Made in Korea?

 

[Silverforce11]

I don't know if anyone has shown this pic yet?

Are they just talking about the interposer being Made in Korea?

Usually that implies the total package. It would be extremely strange if the GPU is made by TSMC, the HBM made by Hynix, then the package is assembled on the interposer/stacked in Korea.

Korea implies Samsung, but considering GloFo is on 14nm due to Samsung, not sure if there's collaboration, does GF have a fab in Korea?

 

[96Firebird]

Korea implies SK Hynix...

The GPU die is probably fabbed at TSMC, shipped to Hynix where the package is put together with the interposer and HBM.

 

[UaVaj]

Official specs:

looks like going QUAD FIRE will "require" a custom water loop.

why not an air cooled option?

Bashing is not allowed in the sub-forums. This area is for technical discussion without negativity.
Moderator Subyman

 

[sam_816]

looks like going QUAD FIRE will "require" a custom water loop.

another STUPID AMD decision. at 275w tdp. why not an air cooled option?

I think there will be an air cooled version called 'fury' with all the basic specs same as fury-x.. But with fans instead of clc..

 

[railven]

I think there will be an air cooled version called 'fury' with all the basic specs same as fury-x.. But with fans instead of clc..

I was under the impression Fury was a cut down Fury X, thus the $100 diff.

Perhaps AIB will roll out a Fury X air cooled version.

 

[flopper]

I was under the impression Fury was a cut down Fury X, thus the $100 diff.

Perhaps AIB will roll out a Fury X air cooled version.

or the watercooler added a $100

 

[railven]

or the watercooler added a $100

Very possible, but if that were true - why no specs for Fury? Even AT shows "will depend on yields"

http://anandtech.com/show/9387/amd-radeon-300-series

Wait and see.

 

[railven]

On a side note, anyone wondering why the "R9" portion of the name stayed?

Radeon Fury rolls off the tongue better than Radeon R9 Fury.

 

[Erenhardt]

looks like going QUAD FIRE will "require" a custom water loop.

another STUPID AMD decision. at 275w tdp. why not an air cooled option?

Why would it require?
There will be this:

QuadFire = 2 cards = 2 Rads = lots of $$$

 

[sam_816]

On a side note, anyone wondering why the "R9" portion of the name stayed?





Radeon Fury rolls off the tongue better than Radeon R9 Fury.

I don't think people are even addressing it as Radeon furyx... Every one is addressing it as fury-x.

Also, if I remember the presentation correctly all they said was that fury is aircooled , fury x is with clc, dual fury is also with clc(but only used in that prototype PC) but will be in market end of fall. All of them are same card basically but dual will be 2 of these on same pcb. Nano is also (supposedly) same but little was talked about its specs.

 

[MagickMan]

Having 4GB of RAM bugs me. Not so much WRT present titles, though a few do already get close to using that much (I play @4k), but because of many others slated for release in the next year. To me this seems like a rush job to respond to the 980Ti/Titan X, instead of waiting for the 8GB variants to be available in large quantities.

No, I'm not an AMD basher, I very much want them to succeed with this launch, as gamers we need this to maintain competition, this is simply the one thing about it that irks me.

 

[kapulek]

Are they just talking about the interposer being Made in Korea?

http://www.bitsandchips.it/52-english-news/5722-gpu-fiji-made-in-south-korea-at-amkor

 

[ajc9988]

Having 4GB of RAM bugs me. Not so much WRT present titles, though a few do already get close to using that much (I play @4k), but because of many others slated for release in the next year. To me this seems like a rush job to respond to the 980Ti/Titan X, instead of waiting for the 8GB variants to be available in large quantities.

No, I'm not an AMD basher, I very much want them to succeed with this launch, as gamers we need this to maintain competition, this is simply the one thing about it that irks me.

You're uninformed then. The reason for adding 6+GB is to add more bandwidth while the ram largely isn't fully utilized. Check the current benchmarks without optimized drivers. It has the same or better performance than 980Ti and sometimes Titan X. Once directx 12 and openCL 2.x are utilized, because it uses IP developed by AMD, AMD cards will pull away significantly over Nvidia, even with, as you say, only 4GB memory. Now maybe the 8GB fury X variant greatly outperforms, I don't know. But if you truly look at the numbers, the three month driver optimization curve AMD usually has for greatly improving performance, and the above new APIs, it only makes sense to buy AMD!!!

Edit: To speak to your "rush job" comment, AMD was going to release info on the card back in March until the Titan X was released right before they were going to talk, that way to steal AMDs thunder. Then, they announced E3 for the event in May. There was nothing rushed about it. Further, fears from Nvidia caused it to sacrifice Titan X sales because they thought performance of Fury X would be great enough to drown both Titan X and 980Ti sales. It forced a reduction of $200 on the rumored release price to match the 980Ti, potentially costing millions in margin on the card. It wasn't rushed and this is Nvidia fanboy rhetoric!!!

 

[krumme]

You're uninformed then. The reason for adding 6+GB is to add more bandwidth while the ram largely isn't fully utilized. Check the current benchmarks without optimized drivers. It has the same or better performance than 980Ti and sometimes Titan X. Once directx 12 and openCL 2.x are utilized, because it uses IP developed by AMD, AMD cards will pull away significantly over Nvidia, even with, as you say, only 4GB memory. Now maybe the 8GB fury X variant greatly outperforms, I don't know. But if you truly look at the numbers, the three month driver optimization curve AMD usually has for greatly improving performance, and the above new APIs, it only makes sense to buy AMD!!!

Edit: To speak to your "rush job" comment, AMD was going to release info on the card back in March until the Titan X was released right before they were going to talk, that way to steal AMDs thunder. Then, they announced E3 for the event in May. There was nothing rushed about it. Further, fears from Nvidia caused it to sacrifice Titan X sales because they thought performance of Fury X would be great enough to drown both Titan X and 980Ti sales. It forced a reduction of $200 on the rumored release price to match the 980Ti, potentially costing millions in margin on the card. It wasn't rushed and this is Nvidia fanboy rhetoric!!!

Its for that reason i am buying fury. I am pretty sure dx12 and optized driver for hbm will make this card crazy fast in half a year in new dx12 games.

 

[Glo.]

First, there have to be games optimized for DirectX, or any other low-level API.

Of course, DX12, Vulkan and Metal even in current games will at least reduce the overhead, but I think where Fury will really fly is when games will be made from gound up for DX12.


However, Im am not an expert and can not see whole picture.

 

[Lalilulelo]

With all this talk about AMD and better optimization , I'm leaning towards a 144hz Freesync monitor than that Asus one that got released.

 

[thehotsung8701A]

What the different between Fury and Fury X?

 

[Silverforce11]

What the different between Fury and Fury X?

Currently X = water cooled ($649) vs air cooled ($549). We don't have info to really say beyond that.

 

[itsmydamnation]

The X :thumbsup:

im here all week

 

[Silverforce11]

http://www.bitsandchips.it/52-english-news/5722-gpu-fiji-made-in-south-korea-at-amkor

There's quite a few papers (google it, will find pdfs) & articles on AMKOR collaborating with GloFo for HBM stacking too. I would be very surprise if GPU is TSMC made, due to the increased density and the rumored efficiency gains.

 

[DiogoDX]

There's quite a few papers (google it, will find pdfs) & articles on AMKOR collaborating with GloFo for HBM stacking too. I would be very surprise if GPU is TSMC made, due to the increased density and the rumored efficiency gains.

TSMC by anandtech.


http://anandtech.com/show/9387/amd-radeon-300-series