Is K12 still happening?

[NTMBK]

Keller is gone, the CEO behind the ARM push is gone, Seattle is MIA, AMD haven't mentioned K12 in some time now... Has it been canned?

My pet theory- initial tests showed it to be only negligibly quicker than Zen.

 

[Justinbaileyman]

I wanna know what are we left with to hold us over till zen comes to light.. Surely it can't be Am3+ FX or a10 Apu's. Is there a refresh going to happen? Not looking good for AMD at all.

 

[VirtualLarry]

I wanna know what are we left with to hold us over till zen comes to light.. Surely it can't be Am3+ FX or a10 Apu's. Is there a refresh going to happen? Not looking good for AMD at all.

Did you miss Carrizo for mobile, and the Godaveri refresh for desktop?

 

[Justinbaileyman]

No I know all about them but I hardly doubt Carrizo and Godaveri will hold us till 2017 when Zen is being released. At leased this release date is all the buzz now.

 

[AtenRa]

There is a rumor for 14nm Carrizo APU for the AM4 socket desktop in 2016.

 

[AtenRa]

Keller is gone, the CEO behind the ARM push is gone, Seattle is MIA, AMD haven't mentioned K12 in some time now... Has it been canned?

My pet theory- initial tests showed it to be only negligibly quicker than Zen.

I believe it has more to do with the software and the ecosystem than the hardware. Thats why they push the x86 hardware first because the software and ecosystem is already there.

 

[Justinbaileyman]

I hope that rumor is true AMD needs to stop sitting on there hands and throw us a fricken bone here. Its really pissing me off how they "AMD" get us all excited for a release of a product and then the next week they cancels said product.So is k12 really Canceled or is this all speculation?

 

[SOFTengCOMPelec]

The AMD Opteron (up to) 8 core, A1100, Arm Cortex A-57 (Seattle) looked like a very interesting product, quite some time ago (pre-K12 Arm).

Basically the A1100 Seattle, kept on being delayed and delayed, and gradually/quietly disappearing, from public perception.

I'm not sure if it ever appeared in the end ?
Apart from a (in my opinion over-priced), quad core $3,000 pre-release developers edition. Which I also have not seen anywhere (that I can remember), or if I had, it was very brief.

I'm half expecting the same thing to potentially happen with Amd Arm K12, so I'm trying to NOT get excited about it. Until there is definitive news of it existing as a buyable product and/or reviews of pre-release samples come out, or something.

I guess Zen (, +, ++) are the biggies.

I'm NOT even sure, if the K12 would ever end up on the home server market.
E.g. Xeons are readily availble for purchase, but I'm not sure if K12 would have ever been so widely available.

If it does eventually come out, is available from home user orientated stockists (amazon, scan.co.uk etc), and is powerful, reasonable cost etc. It could still be very interesting.
But I will NOT hold my breath, waiting for K12 Arm devices to appear.

 

[LTC8K6]

Yeah, this may have been the last sighting of A1100:

http://www.techradar.com/us/news/co...s-a1100-64-bit-opteron-developer-kits-1259802

AMD launches A1100 64-bit Opteron developer kits

 

[SOFTengCOMPelec]

http://www.techradar.com/us/news/co...s-a1100-64-bit-opteron-developer-kits-1259802

AMD launches A1100 64-bit Opteron developer kits

That's a $3000, for a quad core cpu, which is probably way slower than a quad Xeon, at a fraction of the cost.

For $3000, one could go for a dual Xeon, super computer.

What I meant was, $50 .. $200++ Arm Cortex A57, motherboards including the cpu.

At $3000 I will still have to pass/decline.

 

[dark zero]

There is a rumor for 14nm Carrizo APU for the AM4 socket desktop in 2016.

Actually it exists and is called Bristol Ridge... But only at 28 nm. But rumors says that there is a chance that they enter at 14nm, but limited up to 2.7Ghz due the process.

 

[AtenRa]

That's a $3000, for a quad core cpu, which is probably way slower than a quad Xeon, at a fraction of the cost.

For $3000, one could go for a dual Xeon, super computer.

What I meant was, $50 .. $200++ Arm Cortex A57, motherboards including the cpu.

At $3000 I will still have to pass/decline.

I dont believe K12 is meant for Home use, it is primarily made for the Cloud and Data Center.

I will guess, you will only be able to purchase the metal directly from HP, Dell etc.

 

[SOFTengCOMPelec]

I dont believe K12 is meant for Home use, it is primarily made for the Cloud and Data Center.

I will guess, you will only be able to purchase the metal directly from HP, Dell etc.

I agree. That is exactly what I was hearing about the Opteron A1100 Seattle stuff.

The presentations they would make. Were done in conferences, where big cloud (and similar) server computing people, were at.

I like the idea of being able to buy reasonable cost microservers for home use.

As a comparison. Just about the entire Xeon range (excluding Xeon D's), are freely available, even to home users. So I can make my own micro servers, from Xeons if I want to.

I still think there is a potential market, for e.g. 8 core Arm motherboards, with a degree of computing power (for microserver like activities), and yet they are lowish cost and very low power consumption as well.

Currently one would have to use something like a Raspberry PI 2, to do that. But it has very limited HDD/SSD connection capabilities, little memory and 4 rather than 8 cores. Also it is a fair bit slower (900 MHz), than could be currently achieved (2 GHz ++).

I would like a microserver motherboard at about $100, with Sata/USB etc, 8 core, 2 GHz ++, Cortex A57 or even A72's, which ideally can take ECC, if we want to. But maybe the $100 is still not achievable, I'm not sure if it is yet.

 

[AtenRa]

I agree. That is exactly what I was hearing about the Opteron A1100 Seattle stuff.

The presentations they would make. Were done in conferences, where big cloud (and similar) server computing people, were at.

I like the idea of being able to buy reasonable cost microservers for home use.

As a comparison. Just about the entire Xeon range (excluding Xeon D's), are freely available, even to home users. So I can make my own micro servers, from Xeons if I want to.

Perhaps you will be able to do the same with the ZEN, i dont see why you want the ARM based K12 for home use.

 

[Soulkeeper]

I've always thought a reasonably high clocked arm based 8 or 16 core socketed cpu available for desktop/workstation atx boards would be fun to play with. Reality just won't let it happen tho. In my mind AMD focusing on x86 desktop/workstation/server is the way to go. I think the whole arm push was mostly coming from their BOD and management. They thought arm was gonna take over the world at every level.

 

[SOFTengCOMPelec]

Perhaps you will be able to do the same with the ZEN, i dont see why you want the ARM based K12 for home use.

Definitely. I'm very hopeful that Zen will indeed do this.

Unfortunately (despite me being a partial fan of Arm stuff, stemming from the UK). The relative lack of software for Arm devices, is an important issue. But not a show stopper.

If Zen is available with many core options, including at least 8. Is NOT a huge power hog (like FX was. N.B. I'm typing this message on an Amd FX computer. The 95 W of the FX-6300, is NOT that bad really, anyway).
If they can get the price right as well. I would be very interested in it.

The old, original Opterons that Amd use to make and sell, were NOT used much at home. So K12's going the same route, makes sense, I guess.

 

[erunion]

My pet theory- initial tests showed it to be only negligibly quicker than Zen.

I'm seriously skeptical about it being due to any engineering reason. Business/economics seem more likely.

My theory is that K12 is cancelled due to external and internal reasons.

First, demand for ARM server products has not materialized as expected.

Lots of people were predicting ARM could grab as much as 20% server market share in a few years. Today they aren't on track to reach even close to that number, so the market AMD would be selling K12 into would be much smaller than they anticipated.
Seattle sampling then disappearing suggests a lack of interest from buyers.

Second, AMDs financial situation has forced them to reprioritize funding.

AMDs recent reorganization of GPUs suggests less money for CPU development.

 

[Dresdenboy]

Thanks to the Austin Statesman we got confirmation of some Zen related names. With this it's easier to distinguish between x86 and ARM related patents. Interestingly, some people appear in both camps, hinting at some common work/reused components. And I got the impression, that they planned to have easily interchangeable or hybrid designs. But that needs more throurough checks.

I won't rely to much on old news and PR stuff and would try to find out, what they're really doing instead.

 

[Arachnotronic]

I'll bet that K12 will never see the light of day. It was a dumb idea to begin with. There is no market that AMD can go after with K12 that it can't service with Zen.

 

[Phynaz]

I'll bet that K12 will never see the light of day. It was a dumb idea to begin with. There is no market that AMD can go after with K12 that it can't service with Zen.

This.

Even Intel gave up trying to create a unified platform.

 

[jpiniero]

I'll bet that K12 will never see the light of day. It was a dumb idea to begin with. There is no market that AMD can go after with K12 that it can't service with Zen.

The early rumors said that K12 was wider and more power efficient (due to using ARM versus x86) than Zen albeit not able to reach high clock speeds. It was more for SeaMicro and microservers. Zen's obviously a bad fit for that.

 

[Dresdenboy]

The early rumors said that K12 was wider and more power efficient (due to using ARM versus x86) than Zen albeit not able to reach high clock speeds. It was more for SeaMicro and microservers. Zen's obviously a bad fit for that.

The Cat cores were stopped on their paws and some of the "Zen team" were Cat core developers before. So there is no real new low power contender in sight. Zen might fit that role a bit like XV.

But a beefed up ARM core (they worked on many aspects of the uarch according to patents) could have fit that hybrid role of low power + ARM ISA compatibility (for any devices or OSes requiring that or providing advantages then).

 

[DrMrLordX]

I wanna know what are we left with to hold us over till zen comes to light.. Surely it can't be Am3+ FX or a10 Apu's. Is there a refresh going to happen? Not looking good for AMD at all.

The 7890k is about it, unless you're in the market for a laptop. 35W cTDP Carrizo should be available in North America in October. Should've been available in July, though. Fooey.

Actually it exists and is called Bristol Ridge... But only at 28 nm. But rumors says that there is a chance that they enter at 14nm, but limited up to 2.7Ghz due the process.

Nobody knows when to expect Bristol Ridge. It would be wise for AMD to use it to launch AM4 if they can't get Zen on the market in 2016. I don't know where you got the 2.7 GHz number, though.