KitGuru KitGuru.net – Tech News | Hardware News | Hardware Reviews | IOS | Mobile | Gaming | Graphics Cards
AMD has been making some moves these past few months, starting with forming the Radeon Technologies Group and more recently the company announced plans to move away from the Catalyst Control Centre in favour of ‘Radeon Software', which will feature a total overhaul for the driver. Now, AMD's Radeon Chief, Raja Koduri has promised two new GCN graphics cards for 2016.
Speaking with Patrick Moorhead at Forbes, the Radeon head explained that the new GPU driver software will play a key role in business strategy over the next few years, this will include more regular and stable updates.
While the driver team at the Radeon Technologies Group works away on that, the architecture team will be coming up with new hardware designs, which means we should be getting two brand new graphics cards next year, based on either the 14nm or 16nm fabrication process. Key focusses will include power consumption and higher performance.
Other areas of focus over the next couple of years will include virtual reality, education, professional graphics and even medicine. It looks like AMD is looking to make a push on all fronts possible. This, combined with the scheduled launch of Zen CPUs should make 2016 quite interesting.
Discuss on our Facebook page, HERE.
KitGuru Says: So we can expect two brand new graphics cards from AMD next year, though we don't know exactly when those will launch. Nvidia also has new architecture planned for next year, in the form of its Pascal GPUs, so it will be interesting to see how things turn out in 2016. Are any of you waiting until next year to pick up a new GPU?
I’ve got the upgrade itch but don’t feel like buying anything $600+ for a mere few fps.
One GPU for highend and one for everything else??? That is only thing that makes sense to me
They can with 2 GPU designs with variation in CU counts and therefore die-size. They are going to get dies that won’t make the cut for full chip with all CUs active those will be sold as the lower grade models, that way they can have slightly larger family of models to sell to the market at launch. Think the flagship enthusiast model being 500-590mm^2 while the performance one being 350-400mm^2. If they bin each GPU two get atleast 2 different tiers of models they can have 4 products for the market and completely phase out Pitcarin, Tahiti and Hawaii. If the R9 380X materializes, it might end up re-branded as the new low end for 2016 along Fiji Series being the new performance model joined by the new performance models from the new family of GPUs that would sit right above Fiji.
Definitely there will be some re-brands, either with full Tonga (minus two mem controllers) or the Fiji chips is my guess.
Well i agree almost with everything you say except that tonga will be a rebranded. I do not think that costs will justify such move. Next year every chip will be a 14/16 nm FinFet. So i think that they will introduce two versions for next gen. Maybe one from TSMC and one from GloFo to offer complete new lineup top to bottom. Also, lets not forget that they promise major focus on efficiency with next year products.
Because of costs is precisely why the entire stack wont be moving to 14/16 FF. The 28nm process will be still leveraged for the low end as the cost to develop chips in the 14/16nm FF it still too high. It is best reserved for the performance and enthusiast markets where margins can cover the expenses. Cutthroat low end market needs mature and cheaper process this why I suspect 28nm Tonga will be re-branded.
We don’t need to have insider insight to predict re-brands are definitely on the cards for 2016 lineup, I am willing to bet on it. Given AMD’s cash situation for the next one year I would be really surprised if they changed their entire line up to 14/16nm FF unless the yields are very bad and they need to salvage many dies and turn them into many lower models to avoid trashing them to the cruncher and salvage elements (write downs)
You are absolutely right, their focus is about major efficiency gains but am pretty sure AMD are referring to the enthusiast and performance markets, this new rumor about two GPU designs sort of gives more credence to that idea.
Ok. You have a point. But that topic could be drawn in many directions. I think that AMD for a full success needs to make a complete new lineup. I own their R9 270X in Gigabyte envelop and its great but having one i do realize that they need some fresh things. Well, that is my opinion atleast. One of major steps is allready taken, proper drivers for their hardware, announcement of Crimson drivers and etc. There was rumors for a complete lineup few months back. Next year will offer a major step in performance and i do think that AMD and Nvidia will take that oportunity.
What you say is exactly what anyone at AMD would really like to do but their finances are forcing them to choose where they put their money to get the bigger return. Once the profitability improves for a few quarters they can surely go all FF for everything.
That rumor about all new lineup can technically be achieved by having 28nm GloFo designs, porting from TSMC takes about 4 months. Most of the extra effort is at post-silicon characterization. Grapevine suggests atleast 1 such product is coming from GloFo on 28nm there by utilizing more of AMD’s purchased volume on the wafers anyway from GloFo.
Could be. We will see what will show up next year. I do not say you’re not right. But if you are, then i hope that atleast a midrange cards will be new, along with highend. If not, i’ll simply buy Nvidia, my 270X starts ageing badly.
Aging badly? That GPU is over 4 years old and aged much better than its direct competitor, the GTX 660. At this time the GTX 660 can’t even play half of the games you can actually play on that 270X chip.
I did not mean in that way. In a certain titles that i like there’s a evident lack of performance. I’m not trying to talk bad about my card, i have only positive words for it, but it is time to replace it.
A 290X/390X or GTX 970/GTX 980 would give you a nice performance boost. If you are planning to stick with a GPU for a long time, Radeons are the best, if you plan to switch to a newer GPU every year or and a half, then nVidia.
Well you have a valid point, but i live i Croatia and here price for 970 is around 450 USD, for a 980 600 atleast so that i no go for me. I can only go for something like 380X. Maybe a 970, and that if i manage to sell my card for reasonable money. And that i do not think it will happen. I’ll sell it, for “peanuts” 😀
The 380X should be around 40% faster and since it is based on a newer GCN 1.2 architecture.
I think that will be even faster. That is more than 50 percent bigger core count and only slightly lower clock than mine. Also, like you mentioned it is GCN1.2. That should keep me satisfied around a year cause i do think that this year GPUs will not be a match to a next year ones. By then i’ll maybe save some money to exchange it. Or i simply wait another year, but that is really hard considering performance.
How long are they going to ride GCN? They’ve been using that arch for years.
Well both sides will be getting better at culling and gelding off pieces form each wafer, I think the days of getting just a XT/Pro designation might be waning.
I see one chip will be at 380-390 pricing that easily bumps into GTX980 territory with GDDR5. I could see 3 Sku’s from that, a $200/300/400. Would like that to see that from GloFo and be the first to show.
The other (TSMC) using HBM that goes well above Fiji performance, while down to 390X price points, probably at least 3 Sku’s from that. Like a Nano at $450, then $500, and $600. However that’s not true High-End Enthusiast part that will go against a GP100 that chip, it will be early 2017 about the same timing a Big Pascal can make to the Gaming side.
While yes other 28nm parts for all intents and purpose round out the quickly drying-up of $80-150 price points, basically the same as what happen to the $60-120 price points that ceased being in 2012 for Gaming. I think there will be another chip for that $80-150 added but it will come later. Right now AMD see’s winning back market share in the $200-500 range as where they have to focus on.
Make it simple — upgrade when there is a game you are dying to max out (your favourite franchise) or when the game(s) you run no longer runs to your satisfaction. Until then, spend your money on roller-coasters/sports, travel, girlfriend/women, alcohol, nice dinners with friends/family 😉 That’s what I do. I used to buy flagship GPUs every generation and then I was like why am I doing this when most PC games are console ports? Nah, I got better shit to spend my money on in life.
As long as they keep improving it, adding more functional units (shaders, TMUs, ROPs), enhancing the geometry pipeline, what difference does it make? They used VLIW architecture from 2006 until 2011, or 5 years. HD4000/5000/6000 were awesome despite the same underlying architecture with incremental changes. Chances are GCN will be around until 2020, since it was designed with a longer life-span then VLIW per Eric Demers. It’s far more scaleable – already has DX12 Asynchronous Compute – and can easily be adopted to work with HBM1/2. There is a high chance that next generation PS5/XB2 consoles will also have GCN, just newer generations such as 2.0 or 3.0.
l just g○t paid` $6784` work!ng off my Iapt○p last 7 days. lf You think that’s CooI, my d!vorced friend with twin t○ddIers, is making over $10k every month & convinced me To try. l can’t beIieve how easy it was once l tried it out. Earning potentiaI with this is endIess!…..
To See What I Do Check my Pr0FlLe`………….. 49
Well sid! Therese is NO good reson to buy a new GPU to get a few better frames (If you dont want to brag about it) until you rally need one. For me Fallout4 was one reson to uppgradera. All my games hade run fine on my old AMD 6790 until Fallout4.