AMD Radeon 400 series is based on Polaris microarchitecture
Click here to post a comment for AMD Radeon 400 series is based on Polaris microarchitecture on our message forum
Kloet075
Photoshopped?
If this is a quick photo, then tell me why the "for" and the word "every" are not in line of context? it seems photo-shopped, could be so you could read it, but this is certainly not the original. :question:
somemadcaaant
Yeah that image just looks all kinds of wrong.
Backstabak
That image is probably just edited with what would be otherwise not visible. You can even see that "n" in inspiration is also shorter than it should be.
PinchedNerve
I've been thinking about getting an Asus R9 390 Strix. Does anyone have an idea about how long it will be, before the 400 series is released?
Undying
TBA 2016, just like Pascal.
R9 390 is a very good card. If you need a card now no need to wait something we still dont know nothing about.
DeskStar
Maaaaaaaan I love the sound of new hardware on the horizon....!!
Either company could release their big dogs next and I think I'll scoop up at least one from each company this go around. Going to be nice rocking something new in a year or so....
schmidtbag
I have the R9 290 (basically the same thing) and for 1080p gaming it's probably the best-valued GPU you can get at the moment, with or without the bonus games the GPU may come with. I too am interested to see where the 400 series goes but I don't think you'd regret getting a 290/390.
Just something to keep in mind though - I don't know if you like doing BIOS mods but many of these GPUs have their BIOS totally locked-down. Also, seeing as you're in Flordia, I hope your AC works during the summer, because these GPUs exhaust a lot of heat.
xIcarus
Didn't Nvidia say a while back that Pascal would be released in Q1? What happened to that?
PinchedNerve
AC works well. 🙂 The Asus Strix looks to OC very well, should I want to do that, so I wouldn't be looking to mod the bios. Running a Dell U2715H 2560x1440
Denial
Nvidia never gave a time-frame other than 2016. But it's usually about 9-10 months from tape out to a chip launch. Which would put the launch around April. But this isn't a normal chip launch. It's the first Finfet GPU, it's the first HBM GPU for Nvidia, it's a new architecture and they are most likely going to use a mezzanine connector to a PCI bridge and ship the chip in modules to OEMs. All of these things will definitely complicate the time table for the launch.
TDurden
Why would AMD want to scrape GCN? GCN is well known for AMD software engineers, early tests show perform well on DX12, is same architecture used in consoles and 14nm should make it power efficient too. Plus its cheaper to improve on GCN then create completely new arch.
Probably Polaris is just a name for new GCN version 🙂 why should nVidia be the only one with fancy names 🙂
theoneofgod
As said, Polaris is most likely GCN 1.3/2.0
Dazz
GCN 2.0 i believe it will be, since recently AMD said GCN wasn't going anywhere. It works well enough. I want to upgrade my R9 290 but don't feel compelled to as it playing everything i own @ 2560x1080p without breaking a sweat.
Reddoguk
Everything in the near future is light based. I think this is just a reference to it's power though.
As in Polaris is 2.5 times faster than GCN. That i doubt is totally true. Not 2.5x faster than their fastest GPU. 2.5x faster than GCN 1.1 or even 1.0.
I've been reading up on these light based processors of the near future, were they just replace copper traces in the PCB with fiber instead.
Since we are getting close to the end of Moores Law(with silicon based tech) they are looking for ideas for a new way to use current silicon but with better/faster latency and throughput using light based internal connections.
I very much doubt AMD will receive this new tech with Polaris but the light based CPU/GPUs will start making an appearance in 2016 that i have no doubt about.
It's very clever tech because they can stretch out Silicon usage for longer while they look for alternatives.
I've read that they might not even bother aiming for 8nm and 10nm will be a huge risk as well. So they are looking into ways of using 12/14nm for much longer and that means they need to find new ways of using 12/14nm.
These light based processors have already been given the green light so expect half Silicon, half Quantum real soon. 🙂
Lavcat
Brightest visible star in the sky?
https://en.wikipedia.org/wiki/List_of_brightest_stars
PrMinisterGR
As I posted in the AMD subforum, I believe it means 2.5x more energy efficient (watt/transistors packed), and the 169bce reference probably means 16.9 billion transistors for the top part.
Tree Dude
Silicon is an element, quantum is a type of computing. You would compare binary computing to quantum computing. Quantum is not becoming mainstream any time soon as we still don't fully understand how to use it outside of encryption algorithms.
Graphene, carbon nano tubes, black phosphorus and a few others are being looked at to replace silicon. There is no clear winner yet so silicon will stay for a while longer.
Reddoguk
Controlling electrons using silicon is quantum physics, i never said it was going to be quantum computing.
Directing atoms to travel down tiny fiber optics to remove some of the resistance copper has will double or even triple the current speed of CPU's.
It's like putting a tiny internet web inside the silicon and removing need of copper traces. The bandwidth would improve immensely and caching would be near instant. This would also make silicon last god knows how long because they can just keep improving 10 nm forever.
Xendance
I'm having trouble understanding what you mean.
BColt
Im guessing the 2.5 more effective, combined with another slide that says HDR 1080 better than 4k... means the cards will be 5% more powerfull that this gen, but a bit more efficient. So no mid-range 4k gaming soon 🙂
Also my 6th sense says the 14nm parts and HBM2 will probably be reserved for the top parts only. Mid range maybe HBM1 at best, or the new GDDR5. Also most likely 28nm wont go away this year. 470 ( or even 480) might be 28nm aswell?
