AMD Polaris architecture – GCN 4.0

[Admiral Huddy]

This is the Polaris thread, please keep it that.

^

This

Stop replying to their posts and they might go away. Makes cleaning up a little easier too.

 

[Mtom]

1080 nothing before the end of the year, 1070 maybe polaris 10 if AMD manages better than expected, otherwise rumor have polaris 10 about 20% slower for 100$ cheaper, which is still a good bang for the buck.

Except if they really release Vega in October which should tramp the 1080...ofc this is just a rumor yet.
It would be interesting to see if Nvidia could bring out the Ti the same time as well, or not.

 

[Steampunk]

Both teams have bungled it massively.

A shrink + new arch should be a huge uplift. Instead we get weak ish.

You can't really say that until we've seen Vega. You're complaining that the mid-range parts are not powerful, when the high-end parts are still months off.

 

[flopper]

Except if they really release Vega in October which should tramp the 1080...ofc this is just a rumor yet.
It would be interesting to see if Nvidia could bring out the Ti the same time as well, or not.

AMD changed their release, instead of following suit what people has as far been expected they went with a low end mobile low power adressing the mid range market much more and then later (sep/oct) will pull a high end entusiast card and it will cost a bit as it will be big.

 

[Seanspeed]

AMD changed their release, instead of following suit what people has as far been expected they went with a low end mobile low power adressing the mid range market much more and then later (sep/oct) will pull a high end entusiast card and it will cost a bit as it will be big.

Yea, we dont know any of this man. You're just choosing to believe one rumor over another here.

 

[flopper]

Yea, we dont know any of this man. You're just choosing to believe one rumor over another here.

I am sorry?
It been stated by Roy (TAM) which is an AMD representative so how is that a rumour?
dont you read stuff on the net?
or google?

I dont believe as I am not religious.

 

[Seanspeed]

I am sorry?
It been stated by Roy (TAM) which is an AMD representative so how is that a rumour?
dont you read stuff on the net?
or google?

I dont believe as I am not religious.

I've been trying to keep up with everything, but maybe I missed something.

Do you have a link?

 

[Rannoch]

I am looking forward to seeing what AMD are brining to the table. I hope they produce something that can compete with the 980ti at a much lower cost.

 

[AlamoX]

Sorry about that, I am trying to discuss Polaris but some people continuously taking the thread off topic.

that link still might be relevant to Polaris though, my inclination is the AMD are also trying to increase clock speed at the cost of maximizing IPC. but it sinot clar at this stage with so many vague rumurs. We've seen clocks speed form 800Mhz to 1.6GHz, if the latter is true then i expect IPC is slaos going to be somewhat lower.

It is liekly no coincidence if both AMD and nvidia choose a higher clock, lower IPC architecture. Could be related to the new fabrication process, or how they see DX12 games panning out in the future.

personally i prefere if AMD released polaris 10 with the least TDP possible, instead of pushing the clock to increase performance which i think wont be enough anyway, so if they are going to get like 20% performance for 50TPD more, they rather keep it at lower clock with 100 TDP and let ppl overclock it

 

[flopper]

I've been trying to keep up with everything, but maybe I missed something.

Do you have a link?

http://arstechnica.com/gadgets/2016/04/amd-polaris-will-be-a-mainstream-gpu/

"The reason Polaris is a big deal, is because I believe we will be able to grow that TAM [total addressable market] significantly," said Taylor. "I don't think Nvidia is going to do anything to increase the TAM, because according to everything we've seen around Pascal, it's a high-end part. I don't know what the price is gonna be, but let's say it's as low as £500/$600 and as high as £800/$1000. That price range is not going to expand the TAM for VR. We're going on the record right now to say Polaris will expand the TAM. Full stop."

I also like this what Koduri said,
"When we set out to design this GPU (Polaris), we set a completely different goal than for the usual way the PC road maps go," explained Koduri at the time. "Those are driven by 'the benchmark score this year is X. Next year we need to target 20 percent better at this cost and this power.' We decided to do something exciting with this GPU. Let's spike it so we can accomplish something we hadn't accomplished before."

Bold text added

 

[Yamahahahahaha]

AMD are going for the high(er) revenue mid market with the Polaris launch.

If they can bring a card to the market which has +10%~ performance over the GTX970, which has been a roaring success for Nvidia, with more memory, lower power, tiny die size and at a lower price point then they're on to a winner. If they can produce a ton of them for the Christmas period then they'll be in a much healthier position market-wise. Throw in any VR-specific optimisations then they're really going to be the darlings of the VR manufacturers.

The GTX1070 will offer 15-20% over the 480X but I'd expect the price premium to be significantly more than that.

Vega will be their heavy hitter in 2017 and I would expect that to trade blows with the GTX1080 at least, perhaps the GTX1080 Ti or whatever Nvidia brings out.

 

[flopper]

AMD are going for the high(er) revenue mid market with the Polaris launch.

If they can bring a card to the market which has +10%~ performance over the GTX970, which has been a roaring success for Nvidia, with more memory, lower power, tiny die size and at a lower price point then they're on to a winner. If they can produce a ton of them for the Christmas period then they'll be in a much healthier position market-wise. Throw in any VR-specific optimisations then they're really going to be the darlings of the VR manufacturers.

The GTX1070 will offer 15-20% over the 480X but I'd expect the price premium to be significantly more than that.

Vega will be their heavy hitter in 2017 and I would expect that to trade blows with the GTX1080 at least, perhaps the GTX1080 Ti or whatever Nvidia brings out.

People talk about fast cars and so on but when they buy a car they dont buy that fast car due to its expensive both in cost and maintance. They buy a smaller car, not as fast not designed to be the fastest but that fullfill what you want to do with it. cruising for chicks and boys or some other worthwhile stuff.

Koduri said it well, AMD changed strategy with the design for Polaris.
we find out soon what he meant and designed.

 

[Ayahuasca]

cruising for chicks and boys or some other worthwhile stuff.

I thought I'd seen all the car analogies in this section till this one LMFAO

 

[Seanspeed]

http://arstechnica.com/gadgets/2016/04/amd-polaris-will-be-a-mainstream-gpu/

"The reason Polaris is a big deal, is because I believe we will be able to grow that TAM [total addressable market] significantly," said Taylor. "I don't think Nvidia is going to do anything to increase the TAM, because according to everything we've seen around Pascal, it's a high-end part. I don't know what the price is gonna be, but let's say it's as low as £500/$600 and as high as £800/$1000. That price range is not going to expand the TAM for VR. We're going on the record right now to say Polaris will expand the TAM. Full stop."

I also like this what Koduri said,
"When we set out to design this GPU (Polaris), we set a completely different goal than for the usual way the PC road maps go," explained Koduri at the time. "Those are driven by 'the benchmark score this year is X. Next year we need to target 20 percent better at this cost and this power.' We decided to do something exciting with this GPU. Let's spike it so we can accomplish something we hadn't accomplished before."

Bold text added

I'm not seeing anything here where they say they are aiming for the low end mobile stuff nor where they say they are releasing their 'big' card in Sep/Oct.

I think I get what you were saying now, but there's still some here who predict Polaris 10 is going to hit 980Ti levels of performance, which would put their strategy mostly in line with Nvidia's basically.

I also think AMD have said very plainly that Vega is a 2017 release. Only rumors have said it will come earlier than that.

If they can bring a card to the market which has +10%~ performance over the GTX970, which has been a roaring success for Nvidia, with more memory, lower power, tiny die size and at a lower price point then they're on to a winner. If they can produce a ton of them for the Christmas period then they'll be in a much healthier position market-wise. Throw in any VR-specific optimisations then they're really going to be the darlings of the VR manufacturers.

Keep in mind Nvidia are also releasing GP106 which will be aimed at a similar market.

But yes, I think that for people still on 1080p, which is the vast majority of gamers, performance like the 1070/1080 offers is almost overkill. So there's a good market for bringing in all those who never upgraded to a 970/390 and beyond and also to bring more people up to VR-ready capability.

The GTX1070 will offer 15-20% over the 480X but I'd expect the price premium to be significantly more than that.

P10 needs to be a good bit more than 10% better than a 970 if it wants to get within 15-20% of a 1070. 1070 is going to be a big leap over a 970. Probably around 50% better or so.

I do think Nvidia will still attract more attention and thus sales/marketshare if P10 doesn't compete, though. After all the wait to move on from 28nm, many are expecting to see progress, not just cheaper prices. And that would be a shame.

 

[flopper]

I'm not seeing anything here where they say they are aiming for the low end mobile stuff nor where they say they are releasing their 'big' card in Sep/Oct.

I think I get what you were saying now, but there's still some here who predict Polaris 10 is going to hit 980Ti levels of performance, which would put their strategy mostly in line with Nvidia's basically.

I also think AMD have said very plainly that Vega is a 2017 release. Only rumors have said it will come earlier than that.


Keep in mind Nvidia are also releasing GP106 which will be aimed at a similar market.

But yes, I think that for people still on 1080p, which is the vast majority of gamers, performance like the 1070/1080 offers is almost overkill. So there's a good market for bringing in all those who never upgraded to a 970/390 and beyond and also to bring more people up to VR-ready capability.

I just go with what is stated, can we have a 980ti performance from the Polaris? Doubtful but we dont know. I suspect with OC it will be but question is base level performance.
AMD choose a different strategy, how that will pan out and work for them we dont know due to none has before they just did that.

The statements made from Roy and Kudari tells me they did things differently and Kudari is happy with the card and what they designed it for.
his statement "suprising" indicate just that.
How that design and target is with Polaris and Vega is atm, not known and how it stacks vs whatever market cards that exists today and whatever nvidia made.

 

[CAT-THE-FIFTH]

It would be nice if AMD can get Fury X/GTX980TI rivalling performance out of the highest end Polaris 10 SKU as they might have a chance in catching the GTX1070 and if the GTX1080 price is anything to go for,Nvidia might have given them some leeway unless the GTX1070 is like £300 at launch.

Still seeing it has more R9 390X level though.

 

[Lokken86]

I would be disappointed with 390x level tbh. They may as well just do another rebrand if it were the case.

 

[Rannoch]

Come on AMD let's see your cards

 

[D.P.]

I just go with what is stated, can we have a 980ti performance from the Polaris? Doubtful but we dont know. I suspect with OC it will be but question is base level performance.
AMD choose a different strategy, how that will pan out and work for them we dont know due to none has before they just did that.

The statements made from Roy and Kudari tells me they did things differently and Kudari is happy with the card and what they designed it for.
his statement "suprising" indicate just that.
How that design and target is with Polaris and Vega is atm, not known and how it stacks vs whatever market cards that exists today and whatever nvidia made.

The Once I think I agree with you


The fact is Nvidia had nearly 5% if the actual installed user base with a single SKU - the 970. That is not 5% market share, that is 5% of Steam users.The market share must have been something ridiculous like 20-25% (someone feel free to do some math). It means Nvidia probably sold more 970 GPUs than every single GPU AMD sold from top to bottom (AMD dropped to 18% market share).


When working out what the heck they can do next to win back customer, taking a long hard look at the 970 and replication that success could double AMD's market share. Nvidia beat AMD at their own game: price-performance and a perfectly balanced price point. The cards was also cheap to make, made use of salvaged 980 cores.



AMD could do really well if they have a card close to the 1070 for significantly less money. Of course it will be disappointing for high end users who basically get the choice of an expesnive 1080 or look for a 980Ti/FuryX cheap.

 

[Mauller]

P10 needs to be a good bit more than 10% better than a 970 if it wants to get within 15-20% of a 1070. 1070 is going to be a big leap over a 970. Probably around 50% better or so.

I do think Nvidia will still attract more attention and thus sales/marketshare if P10 doesn't compete, though. After all the wait to move on from 28nm, many are expecting to see progress, not just cheaper prices. And that would be a shame.

What makes you think the P10 would be slower than a 390x? i mean seriously? the 970 is already slower than the 390x by up to 10%