[Ars] AMD confirms high-end Polaris GPU will be released in 2016

[AtenRa]

If this is supposed to mean multi-die, then you can throw all of this out of the window right now.

Because ???

 

[Headfoot]

Everyone is going to be missing real high end for consumers until 2017 or later. nVidia did it twice (680 -> 780, 980 -> 980Ti) and AMD once, arguably twice (7970 -> 290, 290 -> Fiji maybe?).

Why does everyone think for some reason they're going to abandon the "delay high end chip due to long process lifespan product planning?" They wont. The initial release will be low and and mid range, marketed as mid range and high end.

 

[tential]

I really can't see more than 5 cards with two chips if one is really small. If we were talking about a 200mm2 chip and a 300mm2 chip, then I could see 6 cards (enough for a full lineup without having to make a weak dual-GPU.

Where do you think the gap in the lineup will be if it's 5 cards? Just use the current name scheme or whatever to keep it easy.

 

[tential]

It shouldn't be as they've annouced 1.3dp and hdmi 2.0a support. Fury doesn't have either.....

If nvidia doesn't have the new standard of dp and amd does that sells me. But I think nvidia won't blunder on that.

 

[railven]

Everyone is going to be missing real high end for consumers until 2017 or later. nVidia did it twice (680 -> 780, 980 -> 980Ti) and AMD once, arguably twice (7970 -> 290, 290 -> Fiji maybe?).

Why does everyone think for some reason they're going to abandon the "delay high end chip due to long process lifespan product planning?" They wont. The initial release will be low and and mid range, marketed as mid range and high end.

Nvidia definitely did it twice, AMD - I wouldn't argue they did so.

NV held back their big chip and milked two "flaghsips" from that one line of chips. AMD never hinted they had Hawaii and later Fiji in the stack.

EDIT:
And I agree with your second paragraph. AMD caught whiff off all that money Nvidia did. And they are going to chase them.

HOWEVER, this will only work if AMD can deliver the same performance. IE to make it a 1:1 comparative
Pitcarin would have had to have the performance level of GK104. Even Hawaii got trounced by GM204.

 

[Techhog]

Everyone is going to be missing real high end for consumers until 2017 or later. nVidia did it twice (680 -> 780, 980 -> 980Ti) and AMD once, arguably twice (7970 -> 290, 290 -> Fiji maybe?).

Why does everyone think for some reason they're going to abandon the "delay high end chip due to long process lifespan product planning?" They wont. The initial release will be low and and mid range, marketed as mid range and high end.

I'd still argue that AMD hasn't truly done it yet. Tahiti doesn't seem like something conceived as a midrange chip. If you're comparing die sizes, don't forget about AMD's small die strategy.

This time, however, they don't have a choice. All we can hope for (well, those of us who aren't Nvidia\Intel fanboys anyway) is that they go for Nvidia's throat and aim for a $500 price point for the high-end

 

[Techhog]

Where do you think the gap in the lineup will be if it's 5 cards? Just use the current name scheme or whatever to keep it easy.

I'm honestly not sure, but it's unfortunately a key point in the midrange. Most likely $250-350.

 

[ShintaiDK]

It shouldn't be as they've annouced 1.3dp and hdmi 2.0a support. Fury doesn't have either.....

If that was true, we wouldn't see GCN 1.1 and 1.2 cards today. They only list it for new GPUs.

 

[MrTeal]

Nvidia definitely did it twice, AMD - I wouldn't argue they did so.

NV held back their big chip and milked two "flaghsips" from that one line of chips. AMD never hinted they had Hawaii and later Fiji in the stack.

EDIT:
And I agree with your second paragraph. AMD caught whiff off all that money Nvidia did. And they are going to chase them.

HOWEVER, this will only work if AMD can deliver the same performance. IE to make it a 1:1 comparative
Pitcarin would have had to have the performance level of GK104. Even Hawaii got trounced by GM204.

Pitcairn was also a 212mm² die that went into a $350 card. GK104 was a 294mm² die in a $500 card. Pitcairn really competed with the 221mm² GK106.

To be fair, Hawaii launched in Oct 2013, just before the 780Ti, and it was 10% or so slower than the 780Ti even with the crappy reference cooler. That was with a 438mm² die and $550 launch price vs a 561mm² die and $699 launch price. GM204 didn't launch until almost a year later. For the Kepler generation AMD did keep up, though they launched Hawaii much later than Titan and at a smaller die size.

It was later that AMD ran into trouble. nVidia launched GM204 in Sept 2014 at 398mm², while launched a cut down Tonga Pro at 359mm², but GCN1.2 was nowhere near the performance of Maxwell 2 at anything. Fury was actually a really good showing vs GM200 considering how badly Tonga went. nVidia got to do a four flagship launch because the gaming performance of GM204 was much better than GK110 despite the die size disadvantage. AMD wasn't really able to because the performance of Tonga left it well short of Hawaii.

 

[PPB]

If fiji wasnt such a bottlenecked design, it may very well fit with nano binning in the new gen of AMD gpus. Clock it to 900mhz, lower power consumption to 150w and get away with 980 performance with better perf/watt too.

But it is really bottlenecked in some areas and wasnt really designed with 1080p in mind, it really shines in higher resolutions.

I would totally buy such a card at the 350 pricepoint if it really existed. But then again, reality is another matter entirely

 

[Techhog]

Pitcairn was also a 212mm² die that went into a $350 card. GK104 was a 294mm² die in a $500 card. Pitcairn really competed with the 221mm² GK106.

To be fair, Hawaii launched in Oct 2013, just before the 780Ti, and it was 10% or so slower than the 780Ti even with the crappy reference cooler. That was with a 438mm² die and $550 launch price vs a 561mm² die and $699 launch price. GM204 didn't launch until almost a year later. For the Kepler generation AMD did keep up, though they launched Hawaii much later than Titan and at a smaller die size.

It was later that AMD ran into trouble. nVidia launched GM204 in Sept 2014 at 398mm², while launched a cut down Tonga Pro at 359mm², but GCN1.2 was nowhere near the performance of Maxwell 2 at anything. Fury was actually a really good showing vs GM200 considering how badly Tonga went. nVidia got to do a four flagship launch because the gaming performance of GM204 was much better than GK110 despite the die size disadvantage. AMD wasn't really able to because the performance of Tonga left it well short of Hawaii.

Again, AMD was using a small die strategy. Fiji and Hawaii are the two biggest dies they've ever put out by a significant margin.

 

[railven]

Pitcairn was also a 212mm² die that went into a $350 card. GK104 was a 294mm² die in a $500 card. Pitcairn really competed with the 221mm² GK106.

I'm aware of that, which is more so what I mean. Because GK104 wasn't position against Pitcairn, if it were it would have been a worse blood bath.

If AMD is going to follow suit with NV in terms of stretching out the chip lines, ie using Polaris 11 to father two "flagships" AMD has to make sure their middle chip is going toe-toe with NV's middle chip. Because all we saw was GK104 humiliating Tahiti at launch.

To be fair, Hawaii launched in Oct 2013, just before the 780Ti, and it was 10% or so slower than the 780Ti even with the crappy reference cooler. That was with a 438mm² die and $550 launch price vs a 561mm² die and $699 launch price. GM204 didn't launch until almost a year later. For the Kepler generation AMD did keep up, though they launched Hawaii much later than Titan and at a smaller die size.

Yes, and Hawaii was a completely new chip. One that was put against a cut down version of NVs first chip. And it ended up losing when NV just released the full version of the chip they had in waiting.

It was later that AMD ran into trouble. nVidia launched GM204 in Sept 2014 at 398mm², while launched a cut down Tonga Pro at 359mm², but GCN1.2 was nowhere near the performance of Maxwell 2 at anything. Fury was actually a really good showing vs GM200 considering how badly Tonga went. nVidia got to do a four flagship launch because the gaming performance of GM204 was much better than GK110 despite the die size disadvantage. AMD wasn't really able to because the performance of Tonga left it well short of Hawaii.

AMD had been running into trouble since 28nm started. Again, their top chip was no longer competing with NV's top chip. NV didn't even have to bring out their top chip until AMD designed a new chip. When NV designed a new chip, it just repeated itself. And now AMD is on their third 28nm design and is still losing to NV's second design.

 

[MrTeal]

Again, AMD was using a small die strategy. Fiji and Hawaii are the two biggest dies they've ever put out by a significant margin.

Fiji was. Hawaii was only 4% larger than R600, and 80nm was a lot newer than 28nm was at the time. Of course R600 was also kind of a disaster.

 

[Techhog]

Because ???

Because multi-die interposers are not happening this year (or anytime soon, but I'll stick to talking about this year). Polaris 10 is a GDDR5 part, so it doesn't work well with this theory.

 

[Techhog]

Fiji was. Hawaii was only 4% larger than R600, and 80nm was a lot newer than 28nm was at the time. Of course R600 was also kind of a disaster.

Good point, though I believe that strategy switch was actually from HD 4000 and on up until Hawaii and Fiji were released.

 

[Arachnotronic]

Good point, though I believe that strategy switch was actually from HD 4000 and on up until Hawaii and Fiji were released.

I hate the naming of GPUs in this industry. For a minute I thought you were talking about Intel HD graphics 4000 and was confused, but then remembered that AMD did HD 4000 long ago.

 

[MrTeal]

I hate the naming of GPUs in this industry. For a minute I thought you were talking about Intel HD graphics 4000 and was confused, but then remembered that AMD did HD 4000 long ago.

Oh, you're going to be pissed when AMD launches the Polaris 11 GPU in the GTX 490X and GTX 490 products.

 

[Techhog]

I hate the naming of GPUs in this industry. For a minute I thought you were talking about Intel HD graphics 4000 and was confused, but then remembered that AMD did HD 4000 long ago.

Hahaha. I knew I should have said R700... unless that would have confused someone else with the 7000 series...

 

[AtenRa]

Because multi-die interposers are not happening this year (or anytime soon, but I'll stick to talking about this year). Polaris 10 is a GDDR5 part, so it doesn't work well with this theory.

I never talked about multi-die interposers, i said dual polaris which can just be a dual chip solution like we had so many the last years.

 

[railven]

Oh, you're going to be pissed when AMD launches the Polaris 11 GPU in the GTX 490X and GTX 490 products.

Haha, nice! :thumbsup:

 

[Arachnotronic]

Oh, you're going to be pissed when AMD launches the Polaris 11 GPU in the GTX 490X and GTX 490 products.

I know

I owned a couple of GTX 470s back in the day. Strange to think there is a fairly good chance that I will own "400-series" GPUs once again

 

[railven]

What NV goes the X route too.

GTX X80 instead of GTX 1080.

Woof. That's going to be even worse!

 

[MrTeal]

What NV goes the X route too.

GTX X80 instead of GTX 1080.

Woof. That's going to be even worse!

Somewhere, there's someone holding onto this card and just waiting to sell it on eBay for $500.

 

[Techhog]

I never talked about multi-die interposers, i said dual polaris which can just be a dual chip solution like we had so many the last years.

Okay. That said, I don't think that a mid-range dual GPU is a good idea at all.

 

[zlejedi]

LOL the fury x didnt and doesnt suck.priced wrongly perhaps ?

It uses same die area and similar transistor budget coupled with super new and efficient memory technology to barely match in specific circumstances performance of classic GDDR5 construction (which we might add is harvested die).

How is that not an engineering failure ?