KitGuru KitGuru.net – Tech News | Hardware News | Hardware Reviews | IOS | Mobile | Gaming | Graphics Cards
A picture of an unknown hybrid cooling system for an unknown Radeon graphics card has been published. The cooler is allegedly designed by Asetek and is expected to be used on AMD’s forthcoming flagship single-chip graphics adapter, which might be called Radeon R9 390X.
The cooling system, which resembles that of AMD Radeon R9 295X2, features a place for a fan as well as openings to connect a liquid loop. The cooler has the “Radeon” trademark imprinted on it. The cooler seems to be made of aluminum, or a similar metal. The new liquid cooling system is designed to cool-down a single-chip graphics card, which is why the place for the fan is located on a side of the board.
VideoCardz web-site claims that the cooling system is designed for AMD’s forthcoming single-chip Radeon R9 390X graphics card, which is based on the code-named “Maui”, “Fiji” or “Iceland.” The new GPU is expected to be based on the third incarnation of the graphics core next (GCN) architecture (GCN 1.2).
Earlier this year it was reported that AMD had taped out a graphics processing unit with 500mm² die size. The chip is projected to be made using 28nm HPM [high-performance mobile, 28HPM] process technology at Taiwan Semiconductor Manufacturing Co.
Liquid cooling system should allow AMD to set maximum clock-rates on its new Radeon 390X graphics cards. However, not all of AMD’s customers will like the new solution for a variety of reasons.
It is completely unclear when AMD plans to unveil its new-gen single-chip flagship graphics solution. It is logical to assume AMD to release the Radeon R9 390X sometimes in October or November, but actual roadmap of AMD remains completely unclear. Theoretically, the recent slash of the AMD Radeon R9 295X2’s price could indicate that a new single-GPU flagship is near.
AMD and Asetek did not comment on the news-story.
Discuss on our Facebook page, HERE.
KitGuru Says: It is not completely clear why AMD decided to use a hybrid liquid-cooling solution for a graphics board powered by a GPU made using 28nm HPM process technology. Thermals of such chips should not be too high unless the developer plans to either give a headroom for overclockers or just increase the default frequency to ensure that its new flagship will be the fastest in the industry.
” It is not completely clear why AMD decided to use a hybrid liquid-cooling solution”
It’s an AMD GPU, you can bet it will be hot and the reference air coolers just cannot deal with it anymore.
Maybe when 20nm drops, we’ll see if AMD can drop the temps enough for air cooling on flagships.
we’ll see ?? You live under a rock or what? 290x windforce or dCU2 or tri-frozor are all liquid and are not flagships or what?? Geez you halfwhits… 295×2 is a dual GPU board it needs liquid cooling to efficiently move all that heat (and it is the ONLY AMD goard with liquid cooling as reference).
Because amd made a bad reference cooler you all go crazy… geez… remember fermi?
I’m not saying any of the custom coolers are bad just the AMD reference version which does sod all for cooling. Yup Fermi was terrible for heat too but that was made at 40nm not 28nm, and at least it wasn’t designed to run at ~90C (unlike the 290X) which is a stupid temperature goal to have in mind.
I’ll say with a bit of confidence that AMD will need single GPU water coolers for their reference designs for the 390X as it’s unlikely they’ll be at 20nm to increase efficiency and they’ll need to pour even more power to even have a chance of competing with Maxwell. Not a fanboy remark, just a statement on how well Maxwell will perform.
Maxwell is on 28nm hp*m* process short for the best 28nm process there is. Tonga is 20% more efficient per core with 10 core size increase over hawaii that is a net 10% more on the same node same process. Now if you look at the new fire pro cards that are at 225w or lower providing best in class at least performance you can see what binning and good mature process can do. Now AMD is moving to HPM too so the advantages you see with tonga will be amplified by the much better process (read energy efficiency) if they will need a waterblock to cool the 390 down expect it to run at 1.1-1.2 ghz and be a behemot of 550-600 mm^2… if that is the case 390 would not just be faster then anything nvidia can throw at it it would most likelly ridicule it in every possible way imaginable… BUT for this to happen AMD would need to do what nvidia did with titan and amd should be smarter then that. Since the prices for both 290x AND 295×2 have been lowered my feeling is the new waterblock might actually be for a 295×2 update or even 395×2(which i kinda doubt since 295×2 is unrivaled as is) or most likelly an enthusiast version of 390x with higher binning and overclocking headroom. If the go with regular 390x with wayerblock need that would mean either continuing to use hp process or that they f-ed up something really badly. This last hipothesis i kinda don’t belive as a tonga core with 290x core count would be just 470mm^2 using hpm process they could deliver it at ~200w which would not require any type of waterblock… and still deliver 20% more performance then the 290x. Since the 980 will not deliver more then 780ti or be ballpark performance as per nvidia statement and leaked benches… a 390x with 20% more power then 290x would sufice to counter using hpm would just nullify any tipe of efficiency maxwell has.
So i’d say that waterblock must be hiding something special, either hp process or a behemoth of a chip or insane stock frequencies or SOMETHING. A 390x on hpm even at 600mm^2 which woul devastate a 980… so no point in building one…
Fermi still haunts me. My GTX 480 fried bacon. Literally. I’m so glad I switched to AMD back then! On Kepler now, GTX 770 4GB! Will switch back when AMD’s next flagship is out!