• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Polaris architecture – GCN 4.0

I hope that is a joke. If AMD make something with Titan-x performance id expect them to be charging Titan-x prices...

The 290X beat the original Titan, it wasn't £800.

With 14nm being less than half the size per transistor and with 3D transistors less than half the power, plus an uplift in architectural performance Titan-X performance shouldn't be hard.

A 35 watt Polaris is as fast as a 90 watt GTX 950 / 120 Watt R9 270 so i don't see why not.
 
Last edited:
Question for the room...

If the new mainstream/mid-range card was £200, and the same performance as the current £200 card (380), but used only 50% of the power... would that be a success? Would people here approve of such a mid-range card?

You see, I'm wondering if both companies might choose to deliberately keep the same mid-range performance year on year, to "encourage" people to move up the price range, if they want more performance.

They could still legitimately say "our new mid-range cards are the best ever, better features, less power", even if they only performed about 5% better in terms of FPS.

TPU has the launch price for each card they review, plus the price at that time on the cards that are on the market. Have a look from 5870 until now how AMD (and nVIDIA of course), has risen the prices all the way to Fury X. :)

They are already doing it for quite some time.
 
I hope that is a joke. If AMD make something with Titan-x performance id expect them to be charging Titan-x prices...

Do you know why the 970 has been so successful.

Launched on the 18th September 2014, I bought my MSI970 on the 21st sept 2014 for £281.99. This was a second tier card that went toe to toe with the fastest of the prievious generation (780ti) and here we are 16 months later and the same card is in fact slightly dearer £289.99.

Here is a link to the prices of 780ti's in April of 2014, before they came down in price due to the new 900 series coming out.

https://web.archive.org/web/20140409085947/http://www.overclockers.co.uk/productlist.php?catid=1914&groupid=701&sortby=priceAsc&subid=1402

Yes isn't it nice to see the old style shop again ;), anyway this just goes to show that just because a new card performs just as fast at the top tier, it doesn't mean it has to be priced similar to it.

Bottom line if NVidia or AMD for that matter want to sell loads of cards, take a leaf out of the 970's book and price it sensibly.

Of course this is ignoring the whole incorrect specs issue, for which I hope somebody got fired, as it should have never been allowed to happen and the way they handled it was despicable.
 
i have a bad feeling that both nVidia and AMD are a little afraid something like 28nm is going to happen again and the first batch on 14/16nm(2016) will not have the expected performance increase.
They will start to spoon feed us like intel does right now. Hope i'm wrong.
 
i have a bad feeling that both nVidia and AMD are a little afraid something like 28nm is going to happen again and the first batch on 14/16nm(2016) will not have the expected performance increase.
They will start to spoon feed us like intel does right now. Hope i'm wrong.

That's exactly what will happen.

NV have been selling midrange as high end for 3 gens now already.
 
i have a bad feeling that both nVidia and AMD are a little afraid something like 28nm is going to happen again and the first batch on 14/16nm(2016) will not have the expected performance increase.
They will start to spoon feed us like intel does right now. Hope i'm wrong.

Hum, not sure how you can even reason that way.
By doubling the transistors in the same die space you have by default an increase, theory state double, and while true a further die shrink might not happen directly there is already development for 10, 7nm and 5nm.

What your afriad off did happen and happens without a die shrink no matter how you dice it. So the question would be if they then suddenly would decrease performance on their first cards, well that dont make sense in any world.

The notion a bad feeling comes from star wars and here is the thing, Han solo did see his ship again. The dark side always loose in spite of being powerful and corrupt and ruling the galaxy might make you think twice about that dark side.
 
Hum, not sure how you can even reason that way.
By doubling the transistors in the same die space you have by default an increase, theory state double, and while true a further die shrink might not happen directly there is already development for 10, 7nm and 5nm.

What your afriad off did happen and happens without a die shrink no matter how you dice it. So the question would be if they then suddenly would decrease performance on their first cards, well that dont make sense in any world.

The notion a bad feeling comes from star wars and here is the thing, Han solo did see his ship again. The dark side always loose in spite of being powerful and corrupt and ruling the galaxy might make you think twice about that dark side.

words fail me ;)

Started off so well then just went off on some weird star wars tangent hahah
 
Hum, not sure how you can even reason that way.
By doubling the transistors in the same die space you have by default an increase, theory state double, and while true a further die shrink might not happen directly there is already development for 10, 7nm and 5nm.

What your afriad off did happen and happens without a die shrink no matter how you dice it. So the question would be if they then suddenly would decrease performance on their first cards, well that dont make sense in any world.

The notion a bad feeling comes from star wars and here is the thing, Han solo did see his ship again. The dark side always loose in spite of being powerful and corrupt and ruling the galaxy might make you think twice about that dark side.

when we were at 28nm there was developement already for 22/20 and the finfet 14/16, so your point doesn't really stand.
Maybe i didn't express my thoughts properly, let me try again.
Even though they are working on 10nm, that doesn't mean it'll be ready for production next year, more so when we won't see 14/16 products at least another Quarter (probably more 2).
What i tried to say is that AMD and nVidia might get scared with what happened on 28nm and first launch GEN1 of GPU's which WILL get some performance, but it will not their best (10-20% instead of 30%).
Them being scared is a normal thing, probably more AMD than nVidia, you can't expect to have next manufacturing nodes ramping up as quickly as they did couple of years ago.
Also 7nm and what's beyond, already is ANOTHER matter.
 
Hum, not sure how you can even reason that way.
By doubling the transistors in the same die space you have by default an increase, theory state double, and while true a further die shrink might not happen directly there is already development for 10, 7nm and 5nm.

What your afriad off did happen and happens without a die shrink no matter how you dice it. So the question would be if they then suddenly would decrease performance on their first cards, well that dont make sense in any world.

The notion a bad feeling comes from star wars and here is the thing, Han solo did see his ship again. The dark side always loose in spite of being powerful and corrupt and ruling the galaxy might make you think twice about that dark side.

Is that Yoda-talk? That's either brilliant or terrifying.
 
The notion a bad feeling comes from star wars and here is the thing, Han solo did see his ship again. The dark side always loose in spite of being powerful and corrupt and ruling the galaxy might make you think twice about that dark side.

Seriously.......you're my favourite poster on this forum. May the force be with you and AMD :D
 
Is that Yoda-talk? That's either brilliant or terrifying.

Yoda takes a moment, his breath shallows out, his eyes goes down, he smiles slightly and then the spirit goes up, his eyes opens up and he says "the power of the force is an ally not one to own as absolut power corrupts" he then makes a short pause, smiles even wider and says, being dead and glowing is absolutly awesome!


On topic,
seems the new card Polaris brighter than the sun....has a redesigned GCN that is a big step up from previous iterations of GCN add the die shrink and the improved control gates of power we have with the HDR 1.3 displayport enabled 144hz screens a whole new year of fun. That AMD have it up and running and demos the card shows the drivers are already there stable enough to do demo inspire me with confidence as they have 6 months to improve current state.

Yoda approves.
 
August... is such a long time away :/ Although technically "before back to school" could be tomorrow, you don't say something like "within five years" if you mean "next week".

So we're set for several months of nothing but Tiny Polaris, much like the nV did with the 750.
 
August... is such a long time away :/ Although technically "before back to school" could be tomorrow, you don't say something like "within five years" if you mean "next week".

So we're set for several months of nothing but Tiny Polaris, much like the nV did with the 750.

it is a new process, large chips will take a bit longer to tweak production for. Considering this was only used for tiny Mobile chips before it is bloody good going so far.
 
August... is such a long time away :/ Although technically "before back to school" could be tomorrow, you don't say something like "within five years" if you mean "next week".

So we're set for several months of nothing but Tiny Polaris, much like the nV did with the 750.


You physically can't sell or have millions of laptops/desktops ready to be sold in July/August if the cards didn't ship 2 months before that.

This was a statement basically to investors confirming their products will be in back to school computers. If Apple say the Icrap 27 will be available for back to school, that is a device and gets sold as is. I'd expect launch within the back to school window. When someone says a discrete gpu will be available, you know that it has to be out a couple of months before to make it into products.

AMD's statement implies very strongly that Polaris 10 and 11 will be launched by June at the latest. I wouldn't be surprised to see one if not both the Polaris GPUs out significantly before that. Mid march for ickle maybe, April/May for medium.
 
Soooo what timeframe and price do we reckon the top tier AMD offerings will be? im going with around £600 for the top tier AMD card, not sure on release though, as were not sure which 2 polaris GPU's are coming first right? they said Enthusiast? but that could still be mid range imho...

I'll say pre Xmas for the top end AMD cards and £600, i reckon November.
 
Soooo what timeframe and price do we reckon the top tier AMD offerings will be? im going with around £600 for the top tier AMD card, not sure on release though, as were not sure which 2 polaris GPU's are coming first right? they said Enthusiast? but that could still be mid range imho...

I'll say pre Xmas for the top end AMD cards and £600, i reckon November.

All bets are off atm. I still don't think we will see a top-to-bottom new range. I think there will have to be at least one or two rebrands in there. Biggish Polaris might make 3 cards and tiny Polaris 2, but in theory there is a large gap between them.

AMD did say (in the Forbes interview) that "some of their cards were getting old and needed replacing", but they didn't say /all/ of them did.

I'm not sure if Fiji with it's HBM1 (not used for anything else) will be rebranded, or if the 390 will have a final outing.

I also think £500 to £600 for the top tier card. I don't think any of the Biggish Polaris cards will be <£350 - if there are two cut-down versions - or £400 if there is only one cut-down version.

The most expensive Tiny Polaris card might well be <£200, so there's a big gap (in my mind) between the two. What could fill it?
 
Back
Top Bottom