[AT] AMD Reorganizes Business Units - no more high performance x86 cores?

[Fjodor2001]

http://www.anandtech.com/show/8174/amd-reorganizes-business-units-names-dr-lisa-su-as-coo

I'm not sure how to interpret that, but it could mean that AMD will exit the highest end performance x86 CPU core segment and instead focus only on performance/power optimized cores. I.e. no continuation of highest TDP Kaveri and Excavator models, and no competition for the highest TDP Intel desktop models from Broadwell/Skylake and onwards. Instead we'll see AMD products competing with Intel's U/Y mobile series, and perhaps low TDP desktop CPUs. Or perhaps they are only talking about the move from high TDP 8-core FX Bulldozer CPUs to APUs?

What's your take on this?

 

[jpiniero]

When I saw this, my first thought is that they are doing this to obscure/hide something from Wall Street.

 

[chrisjames61]

http://www.anandtech.com/show/8174/amd-reorganizes-business-units-names-dr-lisa-su-as-coo

I'm not sure how to interpret that, but it could mean that AMD will exit the highest end performance x86 CPU core segment and instead focus only on performance/power optimized cores. I.e. no continuation of highest TDP Kaveri and Excavator models, and no competition for the highest TDP Intel desktop models from Broadwell/Skylake and onwards. Instead we'll see AMD products competing with Intel's U/Y mobile series, and perhaps low TDP desktop CPUs. Or perhaps they are only talking about the move from high TDP 8-core FX Bulldozer CPUs to APUs?

What's your take on this?

My take is performance means nothing if it takes 220 watts to achieve it. Obviously the more cores approach is dead. I think its the right approach. It is the approach Intel is taking. They just started down the path 10 years ago. Really I think it is ARM that is forcing both Intel and AMD to follow this path.

 

[Insert_Nickname]

"Focus on power-performance optimized cores" can mean basically anything and nothing. I wouldn't read too much in it.

 

[USER8000]

The slide is from 2012.

 

[frozentundra123456]

Exactly. It is an old slide. Seems like she is just reinforcing and restating that strategy.

 

[witeken]

My take is performance means nothing if it takes 220 watts to achieve it. Obviously the more cores approach is dead. I think its the right approach. It is the approach Intel is taking. They just started down the path 10 years ago. Really I think it is ARM that is forcing both Intel and AMD to follow this path.

No, Intel didn't need ARM to figure this out.

 

[mrmt]

I'm not sure how to interpret that, but it could mean that AMD will exit the highest end performance x86 CPU core segment and instead focus only on performance/power optimized cores. I.e. no continuation of highest TDP Kaveri and Excavator models, and no competition for the highest TDP Intel desktop models from Broadwell/Skylake and onwards. Instead we'll see AMD products competing with Intel's U/Y mobile series, and perhaps low TDP desktop CPUs. Or perhaps they are only talking about the move from high TDP 8-core FX Bulldozer CPUs to APUs?

What's your take on this?

Not sure why Anand linked a 2012 slide when talking about a reorg happening in 2014. The writing was already in the wall in 2012 for the big core line and 2014 just fulfilled the worst expectations on the subject except that Kaveri and Excavator APU were/will be launched, more likely due to the WSA agreement than for any other significant business reason.

The message that AMD is delivering in conferences and Q&A was already aligned with the strategy they are implementing now, and the big cores aren't even mentioned here. All the talk goes for the upcoming ARM chips and for Jaguar/Puma, etc. The other interesting omissions on the Q&A are HSA and Microservers. We're yet to see a HSA SDK (but we already have CUDA with unified memory management for download...), apparently their Microserver expectations didn't materialize and Seamicro is currently valued by the credit rating agencies at less than $100MM than AMD paid for the business. Btw, Andrew "unmitigated failure" Feldman left AMD this week.

But I think the shift is much deeper than what Anand's article is showing. The division that Lisa Su will be leading is composed by CPU, GPU and professional GPUs. To group GPU and PGPU is what NVidia did about a year ago. To replicate that strategy would make sense only if AMD is willing to give a much stronger focus on their graphics division and pivot to compete mainly against Nvidia from competing mainly against Intel (when AMD had a divisional structure resembling a lot that of Intel).

That shift makes a lot of sense, since Nvidia will have no foundry advantage over AMD and has a R&D budget much closer to what AMD can muster and while Nvidia has a better financial structure, AMD has a much better CPU technology than Nvidia. The technology is already there and so are the resources, AMD just has to fix their "develop and they will come" approach and start to give serious consideration to software support and DevRel.

 

[tential]

Seems like a good strategy to me. Focus on biggest potential market and hope for the best.

They're able to compete in that market right now, with focusing only on that market they may be able to repeat their prior experience and snowball from that.

 

[frozentundra123456]

@mrmt: True. I especially agree that AMD needs to develop products that fit the market as it is now, not some pie in the sky scenario where moar cores/HSA rules everything. On top of that, their market share is so small it is really hard to drive adoption of new software. Mullins and Beema seem a step in the right direction, but we need more real world tests to see if performance and power usage hold up.

 

[mrmt]

Mullins and Beema seem a step in the right direction, but we need more real world tests to see if performance and power usage hold up.

Marketing is all about bringing attention to your strong points and diverting it away attention from your weak points. If AMD is showing performance and not showing power consumption, I think you already got your answer.

 

[chrisjames61]

No, Intel didn't need ARM to figure this out.

Ummm, Paul Otellini was shown the door for not figuring it out sooner. They have zero presence in cell phones and near zero in tablets. Explain that?

 

[mrmt]

Ummm, Paul Otellini was shown the door for not figuring it out sooner. They have zero presence in cell phones and near zero in tablets. Explain that?

Otelini failed to hook up with Apple and failed to aim correctly in the power/performance curve to get into the mobile market, but he did see a lot of value in the mobile market. It was an execution problem, not direction problem.

 

[witeken]

Ummm, Paul Otellini was shown the door for not figuring it out sooner. They have zero presence in cell phones and near zero in tablets. Explain that?

The ASP was too low and they were too ignorant.

 

[el etro]

Is not no more x86 big cores, but rather focusing on smaller TDP cores for now.

 

[pablo87]

WAG: Lisa was probably offered CEO position somewhere else so they had to promote her to COO with promise to CEO soon...which is very good news for AMD.

 

[pablo87]

Otelini failed to hook up with Apple and failed to aim correctly in the power/performance curve to get into the mobile market, but he did see a lot of value in the mobile market. It was an execution problem, not direction problem.

I disagree. Otellini had StrongArm and basically killed it - lacked vision. Pushed for P4 too IIRC (it's only Israel mobile team saved the day) still hung on to P4 way too long.

 

[mrmt]

I disagree. Otellini had StrongArm and basically killed it - lacked vision. Pushed for P4 too IIRC (it's only Israel mobile team saved the day) still hung on to P4 way too long.

P4 was developed and marketed under Barret. Otelini took the ceo post in 2005.

 

[witeken]

Otelini failed to hook up with Apple and failed to aim correctly in the power/performance curve to get into the mobile market, but he did see a lot of value in the mobile market. It was an execution problem, not direction problem.

Here's Otellini's comment on this:

Otellini said Intel passed on the opportunity to supply Apple because the economics did not make sense at the time given the forecast product cost and expected volume. He told The Atlantic, "The thing you have to remember is that this was before the iPhone was introduced and no one knew what the iPhone would do... At the end of the day, there was a chip that they were interested in that they wanted to pay a certain price for and not a nickel more and that price was below our forecasted cost. I couldn't see it.

Intel had to see it, in my opinion. They knew where the market was going.

 

[Roland00Address]

Quick question not trying to do a serious threadjack but just see if this will help the reorganization of AMD.

Supposedly Intel lost the appeal this week about them doing rebates with OEMs in the pentium 4 days, and thus they owe AMD 1 billion Euros. Is there anybody higher can Intel appeal to or is AMD going to be coming into some more money in the near future? I ask for AMD would dramatically benefit from an incoming (even if one time) revenue.

 

[mrmt]

Intel had to see it, in my opinion. They knew where the market was going.

In hindsight it was the Iphone that started the mobile revolution, because from there phones became true content consumption platforms, but I think it was the Ipad that nailed it. Most people wouldn't stop to to spend times on their laptops or desktops to stay in front of their phones, but they would with their tablets. Intel indeed had the vision of mobile and miniaturization before Otelini, that would be a carefully controlled process by Intel and Microsoft, where the big, clunky notebooks of the early 2000's would shrink into Ultrabooks and only them tablets, a process timely matched with Intel's own roadmap for node shrinks.

The Ipad changed everything. They scaled up their phone ecosystem, leveraged on ARM, not on x86 and generated an entire group of devices that competed against Intel-powered devices. All that in a moment when Microsoft is extremely disoriented, chasing its own tail trying to close down the Windows ecosystem to themselves and destroying the platform in the process.

Steve Jobs once said that Intel was a very slow company, and here they proved themselves again. Jobs wanted to storm the market, he wanted the devices NOW on the market, not Intel vision that would only allow for an Ipad years after the actual launch date. They knew where the market was going, but they didn't know how fast the market was going.

 

[mrmt]

Supposedly Intel lost the appeal this week about them doing rebates with OEMs in the pentium 4 days, and thus they owe AMD 1 billion Euros. Is there anybody higher can Intel appeal to or is AMD going to be coming into some more money in the near future? I ask for AMD would dramatically benefit from an incoming (even if one time) revenue.

The fine won't go to AMD's pocket, it goes straight to the EU coffers.

 

[witeken]

The fine won't go to AMD's pocket, it goes straight to the EU coffers.

And it's already been paid 5 years ago.

 

[NostaSeronx]

Or perhaps they are only talking about the move from high TDP 8-core FX Bulldozer CPUs to APUs?

Did people miss the memo about the 28-nm SOI 16-core MPU?

 

[inf64]

Did people miss the memo about the 28-nm SOI 16-core MPU?

There is no such thing on any roadmap, sorry :/.