12nm AMD Polaris Up and Coming - Pretty Soon Already

Hey... If it ain't broke...

 

I hope they can decrease the power draw as well.

 

all they gotta do is say they now have ray tracing and could double the price.


(this comment may or may not be serious.)

 

I guess next year we will get Polaris 40, which is the exact same thing, only manufactured using TSMC's 7nm technology. Then in 2020 we will get Polaris 50, which is once more the exact same thing, but with enhanced 7nm+ process. With good luck, in 2021 or 2022 we will finally get Polaris 60 with TSMC 5nm process!

 

AAAAAAAHAAHAHAHAHAHAAAAA............ I love it.

Power to performance makes the cards they're trying to refresh kind of a joke. Maybe a %5 increase in performance with nothing else to bring to the table.

I have been vying for a reason to buy AMD again since the X1800xt days and their first foray into crossfire with that sorry dongle hanging off the back of your cards inputs....

 

It's a step in the right direction, but if you already have an RX480 graphics card, then there's no reason to upgrade.

 

I wrote it before, and I'll repeat it. RX-580 is held back severely by memory bandwidth and latency.
RX-580 with 1350MHz can gain good 15% performance just from serious memory tuning. How much more it could gain on ~1500MHz.

Memory limitation of stock cards is so bad that above 1400MHz there are nearly no gains. So, AMD can gain those 15% just by use of better memory. I think that investment to newer memory technology would deliver great benefit as GDDR5 can go only so far.

 

Some really good deals om RX-580 cards out there right now.

Hoping to see some stepped improvements going into the 7 nm cards with these. Maybe not much though.

 

Too bad nobody sent a note to AMD about this. Apparently they haven't ever heard of GDDR5X, let alone GDDR6. They still think GDDR5 is the best one out there, and the only step upwards is the troublesome HBM.

 

No wonder Nvidia launched the 2080 Ti along with the 2080 this time, they must have shat themselves!

 

They did... with Pascal.

 

Am I missing something? It's 15% on top of a 580.. that puts it against a 1070Ti at best?

 

why the snark? oh yeah...
this is a slam dunk. not every segment of the market has to be bleeding edge. for the gamer on a budget this is a godsend along along with that R5 cpu. for the shareholders it's an "atta-boy", this is a proven design on a proven smaller node. there's also the fact of completing the contract with GloFlo.
now you get seamless 1080p, good 1440p, free-sync, lower power/heat, and a marked performance bump that everyone can see.
most of all it's cheap to make, cheap to buy, and very profitable.

 

Lets make a nice steak on it no frying pan needed xD

 

Apple releases the Iphone again and again, and people buy them anyway.
Intel has done that for the past 7 years with Sandy Bridge and people always bought them.
AMD spent millions developing Polaris, why not use existing working designs on a new fab process?
The performance will go up, or if you want less power draw, you can undervolt and have a green RX500 series.
I didn't need a new GPU when the RX400 landed, the RX500 I couldn't get because of mining, I'll gladly buy a RX670 at MSRP.

GDDR5X is a bit more expensive compared to GDDR5. And GDDR6 needs a new controller, probably being developed for Navi. I don't know if they tested GDDR5X or if it would make sense.

 

Just because others have done something doesn't make it okay. We all do already know, for a fact, the reason for all this is AMD's lack of R&D money (in graphics) and the focus on console demands with what little they do have. That's why Koduri left AMD, because the AMD financial leadership told him to forget about developing new tech and just concentrate on relaunching the old one and hyping it. And so that's what we get.

 

No, that's not why Koduri left.
AMD has been remarkably successful despite whatever opinions gamers may have.
the success of AMD can be partially measured by the headhunting of their competition, in this case Intel re-examining the GPU market.
and guess what else, the new technology has been developed.
AMD has been living at the cutting edge and decided to wait for the 7nm process, of which they're invested into with TSMC.

physics alone will make Polaris extremely compelling at the low end. physics and new architecture will make Navi extremely enticing at mid-market. we will have to wait another six months for the truly high-end AMD, but the performance of Navi will open many, many eyes especially at the price.

 

Moment Polaris was done, brightest people went to work on Navi. Koduri was left with limited resources to make last GPU in GCN line (Vega), and he not only did not meet schedule. He apparently did not admit that new changes are not working. He was solely responsible for every single feature announced for Vega which is disabled even year after launch. He could have asked for few more months, it would not harm anything... quite contrary, those features had good ideas behind.

Koduri failed to deliver fully working Vega, wasted transistors, ... But hey, Good news, moment Vega was done, rest of team went to work on Navi too. And in meanwhile they had some help from Zen team for something "very very"... So, no AMD did not drop GPU business ... and division got quite more cash/talent than you think.