AMD Announcement: ARMv8 Opterons In 2014

[Idontcare]

If I am to buy an ARM server, I can ask prices to AMD and, let's say, generic manufacturer XYZ that also sells vanilla A5x based SoCs. If my design isn't too complex to the point of Freedom Fabric not being the deciding factor but a nice plus, as there isn't any differentiation between the two chips, I can just make the two race to the bottom.

AMD may win the order but at razor-thin margins. This model could work for Seamicro alone, but it does not work for a company that designs its own chips.

This is AMD's challenge. They are approaching the traditional ARM customer (who is use to commodity-like pricing for a commodity-like product) and asking them to be willing pay a price premium for the AMD solution on the basis of the solution containing a proprietary network fabric.

That is a tough sell!

My prediction is that freedom fabric is going to end up being the new iGPU - you give it away for free as an incentive to the customer to buy the otherwise non-noteworthy CPU.

It will be a tough future for AMD, no doubt. If they make any kind of headway in terms of creating a market for 64bit server ARM then any number of much more enriched ARM players will decide they should pursue it too.

Not too mention Oracle, IBM, and Intel who will create their own ARM server competitors if they see ARM servers taking much in the way of marketshare from their own bigger iron products.

 

[Arachnotronic]

This is AMD's challenge. They are approaching the traditional ARM customer (who is use to commodity-like pricing for a commodity-like product) and asking them to be willing pay a price premium for the AMD solution on the basis of the solution containing a proprietary network fabric.

That is a tough sell!

My prediction is that freedom fabric is going to end up being the new iGPU - you give it away for free as an incentive to the customer to buy the otherwise non-noteworthy CPU.

It will be a tough future for AMD, no doubt. If they make any kind of headway in terms of creating a market for 64bit server ARM then any number of much more enriched ARM players will decide they should pursue it too.

Not too mention Oracle, IBM, and Intel who will create their own ARM server competitors if they see ARM servers taking much in the way of marketshare from their own bigger iron products.

Intel is already aggressively targeting the ARM micro-server space with its upcoming "Avoton".

 

[Cerb]

My prediction is that freedom fabric is going to end up being the new iGPU - you give it away for free as an incentive to the customer to buy the otherwise non-noteworthy CPU.

But, why don't they give it a noteworthy CPU, too?

 

[Ferzerp]

But, why don't they give it a noteworthy CPU, too?

Because apparently the proper response to not having the funding to compete in you core market means that you should spread yourself even more thin? Beats me.

Actually, they are pretty obviously hoping that this market that doesn't exist will become big (but where is the demand? The journalist arm fetish will not sell servers). The kinds of companies that may look in to this would just go full custom (google, facebook, etc) and AMD would not get any sales anyway. As I've said before, it's a solution desperately seeking a problem.

 

[blckgrffn]

I am with the doubters on this one. We see AMD being used in custom caching servers for folks like Facebook right now because they can cheaply address much more memory than the equivalent Intel SKU.

Why doesn't AMD focus on building a next gen bobcat-like CPU (Jaguar? I am losing track of these things...) and server chipset that could continue to maintain/widen that one significant advantage? Lots and lots of memory per u without a fancy interconnect, etc. Could they not bolt four quad/tri/dual cores of Jaguar together via MCM and go to ~512GB of memory capacity? It seems like with a highly customized server layout this would be feasible. Maybe not *fast*, but feasible. And these are very much custom servers.

Built to facilitate server applications where the pure performance of the platform is much less important than other factors

Going with ARM to do this just seems like a good way to be resource strapped on your other projects.

Pick something and do it well enough to survive, dammit.

 

[Idontcare]

But, why don't they give it a noteworthy CPU, too?

R&D money and IP space.

You need money to fund an IC development team - doesn't matter whether that team is designing an ARM-based chip or an x86-based chip.

And AMD is axing 15% of their staff because they don't have the money needed to even develop their x86 chips...adding in the cost of developing their own custom ARM chip is just not in the cards for them at the moment.

If you step back and look big picture here, AMD is priced for a takeover. Their financials are screaming out "please, dear god, someone with deeper pockets swoop in and save us from ourselves!".

Only they don't have much to offer in terms of reusable IP (the x86 license goes away if AMD is acquired), nor do they have revenue projections that inspire buyers to buy (even if x86 were negotiable with Intel after the acquisition - who wants to compete with Intel on Intel's terms?), but they do have cashflow and debt issues that must be dealt with by whoever would dig into their bank accounts and buy them out.

So Rory's challenge right now, big picture looking at stock price and financials forecasts, is to somehow build enough of a non-x86 based AMD within the shell of the existing x86-based AMD such that they are viewed as a viable takeover target for anyone who is not interested in the x86-side of AMD.

They can't really do that by attempting to be a Calxeda, anyone who is looking to buy into the 64bit ARM server market will just buy Calxeda and not have to deal with any large debts on the books. So AMD won't pursue Calxeda, they will try and leverage what they have that others can't buy elsewhere - Radeon and the integrated GPU technology.

Even then, I don't know, it is hard for me to drink enough beer to get away from the cynical reality of where AMD is headed (and has been headed for years as many people have correctly projected since the days when Phenom was nothing but overhype and underdelivery) and see a silver lining in AMD's cloud.

AMD lacks everything that everyone else in the ARM realm has going for them - momentum, entrenched products, money, employee morale, investor confidence, etc. Pretty much all they got for them at the moment is a plan, a power-point strategy, and a CEO who isn't afraid to keep cutting out employees until they reach a $1.3B break-even point

 

[Cerb]

R&D money and IP space.

You need money to fund an IC development team - doesn't matter whether that team is designing an ARM-based chip or an x86-based chip.

But, why try to jump into a den of wolves, instead of using CPUs they already have?

 

[mrmt]

If you step back and look big picture here, AMD is priced for a takeover. Their financials are screaming out "please, dear god, someone with deeper pockets swoop in and save us from ourselves!"

I think not take over, but liquidation. AMD x86 business is beyond salvation, and their GPU business is in a downward trend that so far gives no sign of stabilizing.

They cannot even become a second source of x86 processors, as they are fabless now and nobody with a bit of sanity would want to manufacture and sell AMD designs by himself.

 

[Ajay]

If you step back and look big picture here, AMD is priced for a takeover. Their financials are screaming out "please, dear god, someone with deeper pockets swoop in and save us from ourselves!".

Only they don't have much to offer in terms of reusable IP (the x86 license goes away if AMD is acquired), nor do they have revenue projections that inspire buyers to buy (even if x86 were negotiable with Intel after the acquisition - who wants to compete with Intel on Intel's terms?), but they do have cashflow and debt issues that must be dealt with by whoever would dig into their bank accounts and buy them out.

And there is Intel's clever poison pill. Anyone who buys AMD has to re-negotiate with Intel for an x86 license (actually, Intel is required to negotiate by the agreement, but any company who could actually threaten them would find the going rough).

What a mess! I think ShintaiDK was right about the direction of AMD being the same as Via... very sad indeed

 

[podspi]

It is clear who needs to buy AMD, Via! Sure, AMD would lose the x86 license, but who cares as Via would have an extra one!

 

[ViRGE]

But, why try to jump into a den of wolves, instead of using CPUs they already have?

Grass is greener on the other side? Bandwagoning?

Obviously what AMD is doing right now (competing with Intel) isn't working well enough, so they have to try something else. ARM has momentum, so it's a good bandwagon for them to jump on.

 

[Cerb]

Grass is greener on the other side? Bandwagoning?

Obviously what AMD is doing right now (competing with Intel) isn't working well enough, so they have to try something else. ARM has momentum, so it's a good bandwagon for them to jump on.

Yeah, but it's a bandwagon of general sameness. Freedom Fabric may allow for a nice dense box, but with plain ARM cores, it'll be the same kind of box as everyone else but Intel has, leaving Intel as the only real standout, with Valleyview-based server CPUs.

 

[Ferzerp]

ARM has momentum

ARM has 0 momentum or presence in the space their trying to use ARM.

 

[Idontcare]

ARM has 0 momentum or presence in the space their trying to use ARM.

I suspect ViRGE is referring to ARM in general, as CEO if you are going to tie the future of your company to something - and your choices are ARM or x86 - then the choice is kinda made for you if you have a BoD that is like AMD's.

They look at the TAM CAGR of ARM versus that of x86 and they see momentum in ARM, and they see a humongous barrier to re-entry for tapping into the profitable x86 TAM...Intel has made sure of that with their relentless node shrink cadence.

At least with ARM you are competing with companies who don't have a process node advantage, everyone uses the same foundry or two, so you can more easily delineate how your company added value to the product via the management and direction given to the design teams.

(look at TI vs. Qualcomm or Broadcom)

 

[Ferzerp]

Well, I keep seeing a lot of really bad assumptions that arm has *any* server presence at all. I guess it being my backyard, I am more up on it, but it boggles my mind just how widespread a few buzzword happy articles on the web can mislead people.

 

[Idontcare]

Well, I keep seeing a lot of really bad assumptions that arm has *any* server presence at all. I guess it being my backyard, I am more up on it, but it boggles my mind just how widespread a few buzzword happy articles on the web can mislead people.

That's how mindshare starts. The engine behind this is huge, lots of money to be made on wallstreet as rumors are made, hype is built, etc.

Look at Calxeda and the investor capital they have raised, that doesn't happen without investor confidence and growing mindshare.

But yeah, until stuff actually shows up in the channel, its all BitBoys Oy!

 

[Ferzerp]

But their hype machine won't create sales. Just look at Cisco doing the same tactics with their UCS servers (which actually exist and are compatible with the software the bulk of the market is using).

They've managed a whole 1.5% of the x86 server market with those things. (And that's with sneaking them in the back door with PBX deployments as well).

Success in the server space is built on the merits of the product, not a marketing machine.

 

[mrmt]

Well, I keep seeing a lot of really bad assumptions that arm has *any* server presence at all. I guess it being my backyard, I am more up on it, but it boggles my mind just how widespread a few buzzword happy articles on the web can mislead people.

You have a point here, ARM has almost zero presence in the server market. ARM bread and butter isn't servers, but mobile chips, so the backbone of their investments won't be on servers, but on mobile.

But ARM is trying to enter on servers, some folks are putting money there, and once ARM has a foothold on the server market, more money will be channeled to servers, which improves the position... you know the drill.

Intel is moving exactly to make sure that ARM won't gather any momentum.

 

[MaxPayne63]

Obviously what AMD is doing right now (competing with Intel) isn't working well enough, so they have to try something else. ARM has momentum, so it's a good bandwagon for them to jump on.

I think at this point their end-game no longer involves keeping their company alive. x86 is completely worthless when they fold, so now it's just about generating enough other value (real or imagined) so that the insiders get a nice payday when it's over. Or maybe I'm just too cynical.

 

[AtenRa]

I will just leave those here

 

[Ferzerp]

AMD marketing says AMD is moving in the right direction. Gee. What insight. You've totally educated us with your marketing slides. I'm going to preorder some AMD ARM servers now.

 

[AtenRa]

AMD marketing says AMD is moving in the right direction. Gee. What insight. You've totally educated us with your marketing slides. I'm going to preorder some AMD ARM servers now.

First picture = IDC Server Forecast

I wasn't expecting anything more from you in an AMD thread.

 

[Haserath]

Why is it going to take until 2014 for them to release this?

 

[Ferzerp]

First picture = IDC Server Forecast

I wasn't expecting anything more from you in an AMD thread.

The IDC forecast is valid, but then you post AMD marketing spin on it. The message is worthless. AMD is the future because AMD says so? How long have they been saying so? Right now, the question is how long will they survive, not will they become dominant anywhere.

You really believe that because AMD says so that they are well positioned?

 

[ShintaiDK]

Why is it going to take until 2014 for them to release this?

No 64bit ARM CPUs before. Might even be 2015