[TR] AMD reveals K12: New ARM and x86 cores are coming

[zalbard]

Source: TechReport

We now know that AMD intends to compete where it has done well historically - in desktops, servers, and laptops, as well as embedded systems and mobile devices - by fielding CPUs compatible with the two major instruction set architectures: ARM and x86.

[SinOfLiberty]

Skybridge vs Skylake showoff.

Off topic: Does anyone need even more BS info about Maxwell cards?
If yes: http://www.chiphell.com/thread-1028165-1-1.html

[AliG]

This is certainly an interesting, pleasant surprise. If AMD can take a solid jab at Intel, we'll finally start seeing price drops and innovation again

[paulbagz]

So the naming structure from the original maps will be the same but they've thrown everything out the window and are starting again?

-PB

[PatRaceTin]

I hope future AMD & Intel Desktop CPUs will have unlocked models.

[AMDforME]

AMD will continue unlocked.

[BeepBeep2]

^knows everything

...
"AMD: New High Performance non modular Core
Consumer: wat
AMD: such core much surprise wow very performance
Consumer: wat
AMD: lol jk 20% less IPC den Bulldozer runs 6 GHz 180w TDP
Consumer: trollface.jpg"

[vario]

^knows everything

...
"AMD: New High Performance non modular Core
Consumer: wat
AMD: such core much surprise wow very performance
Consumer: wat
AMD: lol jk 20% less IPC den Bulldozer runs 6 GHz 180w TDP
Consumer: trollface.jpg"

LOL

AMDforMe <- You cant possibly know that, the new cores arent going to drop before 2017 (high perf x86 ones), so theres like NO information on them, and most probably no one at amd knows too.I mean, there werent even any codenames, socket, memory configuration.Like NO information at all.I wouldnt even bet that theyre gonna come to fruition.
Anyhow, is there any info on the excavator parts coming in 2015 ?

[Lanek]

LOL

AMDforMe <- You cant possibly know that, the new cores arent going to drop before 2017 (high perf x86 ones), so theres like NO information on them, and most probably no one at amd knows too.I mean, there werent even any codenames, socket, memory configuration.Like NO information at all.I wouldnt even bet that theyre gonna come to fruition.
Anyhow, is there any info on the excavator parts coming in 2015 ?

Surely canceled, this new x86 cores are expected for 2016 ( possibly 2017, if the work is really not advanced )... so release current 2015 Excavator seems a vaste of time and energy. ( outside maybe on some APU ). ( thats what seems believe some article from TR, Anand etc ).

I will not say this is confirmed 100% myself... In reality, the presentation public by AMD was really for annonce: " what is coming next " : Skybridge, on the ARM front etc.. . Information are still vague anyway about new cores and mosts have been taken from the discussion official or not with Keller..

[FlanK3r]

Excavator is ready few months, so is buls*t cancel it. And its redisgned SR core, optimalized for lower power consumption.

[vario]

Surely canceled, this new x86 cores are expected for 2016 ( possibly 2017, if the work is really not advanced )... so release current 2015 Excavator seems a vaste of time and energy. ( outside maybe on some APU ). ( thats what seems believe some article from TR, Anand etc ).

I will not say this is confirmed 100% myself... In reality, the presentation public by AMD was really for annonce: " what is coming next " : Skybridge, on the ARM front etc.. . Information are still vague anyway about new cores and mosts have been taken from the discussion official or not with Keller..

I really doubt that its canceled, given its release time frame of start of the next year, its probably almost done in most aspects, its in every roadmap and so on, of course APU`s, amd doesnt do anything else now. Excavator was supposed to have some nice ipc gains like steamroller was supposed to ,SR have some nice gains ,however it was clocked lower than richland, but after a year they probably got manufacturing tech somewhat in control, so maybe it will be passable ...
Anyhow releasing excavator in 2015 wouldnt be a waste of energy as the new cores wont come sooner than end of 2016, they gonna have to have SOMETHING there.

[ilkkahy]

I think one of amd:s problem is that for convincing investors (or something) they start talking about unreleased products years and years before the launch. They should propably stop doing that so that competition would not have any products to response at the launch date (now competition has many years to think competing products for K12).

I might be wrong but i think the ATI side of the company has covered up their future products bit better and therefore many Radeons have been positive suprises in terms of performance.

[stuffme]

I think one of amd:s problem is that for convincing investors (or something) they start talking about unreleased products years and years before the launch. They should propably stop doing that so that competition would not have any products to response at the launch date (now competition has many years to think competing products for K12).

I might be wrong but i think the ATI side of the company has covered up their future products bit better and therefore many Radeons have been positive suprises in terms of performance.

That would be smart, but they can't do that. Otherwise all the investor are going to leave the company. By painting huge cloud castles for investor is only way to get some more money and time to work on those false hopes.

[vario]

I think one of amd:s problem is that for convincing investors (or something) they start talking about unreleased products years and years before the launch. They should propably stop doing that so that competition would not have any products to response at the launch date (now competition has many years to think competing products for K12).

I might be wrong but i think the ATI side of the company has covered up their future products bit better and therefore many Radeons have been positive suprises in terms of performance.

That isnt a problem for AMD, for years and years their talk and final product have not that many things in common :-) , so in essence its best deceptive tactic :P.
Anyhow, even if they would deliver , intel is MUCH more forthcoming with their products and has been, in cpu market you need stability in order to plan for the future, large OEMs must know whats coming so they can build around it.
I pretty much know whats coming from intel in 12 months ,process node, cores, caches, chipsets .We all know whats coming, not that theres something interesting coming, but we know it isnt.And further down the line we even know some things whats coming down the line in like 2 years.Again, not anything really interesting ;-).If amd doesnt do something, in 2017 were gonna be buying yet another intel 4 core, with yet bigger unused igp (for the desktop part), on yet another chipset with a 1 cpu generation lifetime.On ddr4 though, thats gonna be probably somewhat more pricey.
As for AMD, pretty much ANYTHING more powerful than what they already have would be a surprise. So no need to hide it really.
It just dawned on me, until around 2017 most powerful amd cpu is gonna be am3+`s piledriver fx ...

[zalbard]

It just dawned on me, until around 2017 most powerful amd cpu is gonna be am3+`s piledriver fx ...

These new chips should be released (hopefully) before 2017.



Not to mention Excavator coming next year.

[vario]

These new chips should be released (hopefully) before 2017.



Not to mention Excavator coming next year.

Excavator will be a 4 core, the chips that come in 2016 are the arm ones, and the new x86 is supposed to come after that :/

[Lanek]

Excavator will be a 4 core, the chips that come in 2016 are the arm ones, and the new x86 is supposed to come after that :/

APU 4cores for the excavator, like steamreolle ( if it is released, thats really not sure ), it will not goes to high end cores ... as for time of release of the ARM / x86 K12,,,, for what i have understand this is the invert, first the x86 and then the ARM ( we can allready call it ARM-AMD core, this one will have nothing to do with the traditionnal ARM desigh we see today )

[haylui]

Fusion taken to another level...fuse anything together...

[vario]

APU 4cores for the excavator, like steamreolle ( if it is released, thats really not sure ), it will not goes to high end cores ... as for time of release of the ARM / x86 K12,,,, for what i have understand this is the invert, first the x86 and then the ARM ( we can allready call it ARM-AMD core, this one will have nothing to do with the traditionnal ARM desigh we see today )

Yeah,no more cpus just apus, thats why im saying 2012`s Piledriver FX is gonna be most powerfull amd cpu for a LOONG time, no SR or EX in 8 core variants.I havent seent any mention of the new x86 core being here faster than the K12.There was talk about new iterations of puma earlier.And as K12 was detailed in more detail than the "new x86 from ground up" i would think that it is more advanced in development.From what i read most optimistic interpretation that theyre coming together sometime in 2016 :-/

[naokaji]

Fusion taken to another level...fuse anything together...

Makes perfect sense.

First we got increased clockspeed, then performance gains from that path hit the nasty part of the diminishing returns curve.
Then we got an increased number of identical cores, then performance gains from that path hit the nasty part of the diminishing returns curve.
Next we will get a whole bunch of different cores slapped together.

[duron]

not much divulge also at xbit

http://www.xbitlabs.com/news/cpu/dis...15_Report.html

[FlanK3r]

While nothing is clear now, a good news is that AMD has not given up high-end microprocessors for servers and clients, the Opteron and the FX-series.

Nice!

[askimo]

While nothing is clear now, a good news is that AMD has not given up high-end microprocessors for servers and clients, the Opteron and the FX-series.

Nice!

Totally agree with you. Hopefully they are committed and is not another marketing show

[AbortRetryFail?]

Fusion taken to another level...fuse anything together...

Yeah ... SoCs are a waste of money

AMD has made no secret over the last 4-5 years of the ability of HSA to 'bolt-on' tech to what is now the UNB.

ARM is a good step ... SPARC or PowerPC cores will be even better.

[lutjens]

I wonder how a hybrid chip will really perform for each instruction set. Will one instruction set run better than the other, or will the presence of the other drag performance of the other down? If they can pull it off (which would be surprising), I'll be truly amazed. It would probably need some kind of special hybrid OS as well, something that runs x86 as well as ARM code, and I wonder how willing Microsoft would be to design that, rather than force a machine with a K12 to choose between the OSes for each instruction set. i can just imagine the drivers for such a hybrid OS...would be interesting to see how they'd do it.

So many hurdles not even counting those in the CPU itself and so many unanswered questions...