AMD Polaris architecture – GCN 4.0

[nashathedog]

Vega should be close to double the speed of Fiji. Surely 70-80% better at the least. Again, newer architecture, bigger die size to polaris and hbm 2 on top of that...

Let's hope so,
This all started when Nvidia released there premium brand graphics card, That was the Titan and it sat head and shoulders above anything either side had at the time without going dual chip. It stayed ahead for some time and even with the massive premium it had it was everything it was meant to be. A big success.

AMD attempted to follow suit and offer there own premium brand of graphics card but they failed and ended up giving us a card that does not sit that far from the 390x that's below it, They forced water-cooling on those that wanted the full chip and wasted an excellent air cooler on a cut down chip that sometimes struggles to beat the 390x in games.

They desperately need to get a win.

 

[Muz333]

AMD Nomenclature Analysis – Very Powerful ‘C99’ Polaris GPU and Vega 10 Test Vehicle Spotted on Zauba

Here I thought some of the posts on these forums involving AMD and Nvidia could be a bit over the top. Most of the commentators there are rabid!

 

[bru]

Your not wrong, we may poke and make fun at each other at times, but some other places I've read stuff that would make your hair stand on end. some people just take the faceless internet thing way way too far.

 

[humbug]

Don't forget that Maxwells power efficiency seemed so good because it was compared with 3 year old AMD tech. They got kicked in the balls by the cancellation of 20nm, so they have to use what they had.
Other than maxwell the two usually were close of PE, so i don't expect much difference between polaris and pascal either.

+1

If you compare a 290X to a GTX 980 then yeah Maxwell wins hands down, however... one has to remember that Hawaii has an IMC twice the size of Maxwell and 16 Memory IC's vs 8 on the 980 and Hawaii's compute is intact while its stripped out on Maxwell.

Also... Fury-X has about 10% higher power usage than a 980TI
While the Nano is about the same as a GTX 980 and beats it convincingly at 4K.

Latest market AMD architecture vs Nvidia in performance per watt is quite similar.

 

[FoxEye]

https://forums.overclockers.co.uk/showpost.php?p=29306706&postcount=34

Gibbo exlusively reveals Polaris not coming until "end of summer"

RIP June release?

 

[theMilitant]

^^Well, that's about as good a confirmation of a release date as any I've seen.

 

[CAT-THE-FIFTH]

https://forums.overclockers.co.uk/showpost.php?p=29306706&postcount=34

Gibbo exlusively reveals Polaris not coming until "end of summer"

RIP June release?

The problem is AMD is officially on record as saying before the back to school period - in the US the back to school period is between mid July and mid September.

According to the Met Office summer starts in the third week of June.

So I would say between June and July.

 

[ICDP]

https://forums.overclockers.co.uk/showpost.php?p=29306706&postcount=34

Gibbo exlusively reveals Polaris not coming until "end of summer"

RIP June release?

ROFL, just ROFL. Does anyone expect Gibbo to come and post "stop buying existing GPUs cus Polaris/Pascal is due soon".

Not only that his post is on an official OCUK promotional thread. Here have some rolling eyes

I remember Gibbo stating R9 Fury X would not be available anywhere for less than £600 on release when he put his pre-order deal up. Lo and behold Fury X is released and is available for MSRP apart from OCUK. Fair play to Gibbo he knows how to run a business and I respect him for it but he's not to be used as a valid source of GPU release info.

 

[FoxEye]

ROFL, just ROFL. Does anyone expect Gibbo to come and post "stop buying existing GPUs cus Polaris/Pascal is due soon".

Not only that his post is on an official OCUK promotional thread. Here have some rolling eyes

Tbh, I don't think he'd outright lie. People would remember and it wouldn't be good for business.

I think if Gibbo knew release was next week, he'd simply stay stum, rather than lying to say it was August some time.

So frankly either he's speculating like the rest of us, or it's going to be close to the truth. AMD did say "back to school" period, and that does imply July/August time.

 

[Ayahuasca]

I eagerly await DM's post on why Gibbo is wrong.

 

[CAT-THE-FIFTH]

I eagerly await DM's post on why Gibbo is wrong.

Apparently AMD are talking ******** too it seems. Not sure what DM has got to do with anything TBH!!

Tbh, I don't think he'd outright lie. People would remember and it wouldn't be good for business.

I think if Gibbo knew release was next week, he'd simply stay stum, rather than lying to say it was August some time.

So frankly either he's speculating like the rest of us, or it's going to be close to the truth. AMD did say "back to school" period, and that does imply July/August time.

I don't know why people are debating what AMD has specifically said.

The problem is AMD is officially on record as saying before the back to school period - in the US the back to school period is between mid July and mid September.

According to the Met Office summer starts in the third week of June.

So I would say between June and July.

It is mentioned here:

http://www.computerbase.de/2016-01/amd-ceo-lisa-su-interview/

The target is mid-2016, particularly in time for the major back-to-school season in the united states. Polaris should be available not just in the form of desktop solutions, but also notebook graphics chips.

That is from the CEO of AMD and if people actually used Google,it does not take a rocket scientist to find out the rough release period,which is late June or July.

September would be past the targetted period and would also make the release Q3 2016 too,and would not be the "middle" of the year.

Even August would be Q3 2016 too and pushing the boundary of the middle of the year,and it would miss a decent part of the "back to school" period.

OFC,like the HD7970 asnd HD7950 it could very well be more a limited launch too,but late June or July looks a good time period and that would be in time for the "back to school" period.

 

[bru]

Yup exactly that, it has been gone over in a few threads, probably even this one a couple of times. School term starts on the 9th of September I believe, so anytime before that.
I expect they will be shown at Computex with availability 2-8 weeks after, therefore easily arriving before the US start of school term.

 

[TheMorningStar]

I imagine the best way to call a release would be to watch for multiple brands and versions to start going out of stock, leaving mostly ordinary cards.

 

[tavtavtav]

Risk mitigation. They want to get the smaller chips working on a new process before they try it with bigger chips. It's better for yield and thus profitability. Starting first with a big chip has bitten both AMD and Nvidia before, because it's much harder to do.

Oh I understand this, I was more talking along the lines of releasing hawaii size polaris (if It is true polaris 10 is only 232mm^2) and fiji size polaris before Vega in order to extend the life of the node.

 

[Orangey]

Pretty sure you'll see laptops with Polaris in BTS period, desktops could well be Sept/Oct though.

No idea how they will sell them being pretty much shut out of OEM support though. Oh, Apple (those skinflints) will throw them some business but I expect you might have a hard time finding the exact config you want from the brand you want.

 

[Mtom]

I don't get why you guys argue on that. It seems Gibbo just said what was on the news...new cards coming at summer.
Maybe they will say the release date on computex. We'll see.

 

[FoxEye]

I don't get why you guys argue on that. It seems Gibbo just said what was on the news...new cards coming at summer.
Maybe they will say the release date on computex. We'll see.

Could be reading too much into his post, but he did say "end of summer". Not "coming summer".

End of summer is much more specific.

 

[D.P.]

Apparently AMD are talking ******** too it seems. Not sure what DM has got to do with anything TBH!!





I don't know why people are debating what AMD has specifically said.



It is mentioned here:

http://www.computerbase.de/2016-01/amd-ceo-lisa-su-interview/



That is from the CEO of AMD and if people actually used Google,it does not take a rocket scientist to find out the rough release period,which is late June or July.

September would be past the targetted period and would also make the release Q3 2016 too,and would not be the "middle" of the year.

Even August would be Q3 2016 too and pushing the boundary of the middle of the year,and it would miss a decent part of the "back to school" period.

OFC,like the HD7970 asnd HD7950 it could very well be more a limited launch too,but late June or July looks a good time period and that would be in time for the "back to school" period.

Back to school in the US is August to early September, June is still at school in the previous school year!I know, because I happen to live there.

First Day of School last year was September 9th, last day is June 28th. Back to school is the last couple of weeks in august first couple in September.

 

[FoxEye]

Back to school in the US is August to early September, June is still at school in the previous school year!I know, because I happen to live there.

Heh, I though it was a bit mean to be sending kids back to school in mid-July

 

[D.P.]

Could be reading too much into his post, but he did say "end of summer". Not "coming summer".

End of summer is much more specific.

End of summer is the "back to school period" which is what AMD talked about, Gibbo is purely re-iterating AMD's own information.