PC Mag: AMD CEO denies smartphones plans, Slates/Tablets attack for 2012

[cbn]

http://www.pcmag.com/article2/0,2817,2380495,00.asp

AMD interim chief executive officer Thomas Seifert said Wednesday night that the company has no plans to enter the smartphone space at the present time, and the company was not developing a baseband chip.

And although some tablet makers are selecting AMD parts now, Seifert said that its 2012 APU roadmap addresses the tablet space.

"We have no intention of entering the smartphone space at this point of time," Seifert told the Goldman Sachs Internet and Technology Conference. "We are not in the process of developing [a] baseband [chip] and this is a clear statement. There are enough players in the is market that have a hard time earning money."

Seifert made his first appearance after taking over from chief executive Dirk Meyer, who was asked to leave in January. With two more leaving last week, Seifert was asked whether or not customers were skittish.

Seifert reiterated that he viewed his position as a temporary one, and that the company's Fusion processors were key to holding customers in the AMD camp. Seifert said that AMD had demonstrated its Llano APU at the conference, which it began in October of last year.

Seifert said that AMD would transition to Llano in the second quarter, and that it demonstrated a 4-core, 1.8-GHz chip that was fully capable of running Microsoft's DirectX 11 API. An AMD spokesman said that Seifert did not mean to imply that that was the speed that Llano would run at. "The architecture is vastly superior" to Intel's own microprocessors, which are code-named "Sandy Bridge," Seifert said.

Seifert also reiterated that the company was sampling its next-generation Bulldozer core now, and would ship for consumer PCs in early summer. Servers will receive the core in late summer, he said.

Although some saw Seifert as just an instrument for the board that ousted Meyer, Seifert said that one of the conditions he had in taking the role of temporary CEO was that "he had complete freedom to act".

"There is so much gross margin expansion potential for us in our business model this year, we cannot afford to lose time/speed," Seifert said, according to an official transcript of the event at AMD provided after the presentation. "And you can only act in this role if you are fully empowered and we have been able to make all the changes so far that need to be done. They are much more operational at this point in time, really making sure that we hit the business case, that we pick up pace where there's opportunity. But we've also started to put teams in place to look for opportunities beyond our current situation."

Seifert also addressed the tablet question, the issue that some said had forced Meyer out. Over the last couple of months, he said, usage models had shifted the market in favor of AMD, including 3D graphics and higher overall performance.

AMD's customers may prefer to wait until the second-generation APU products coming next year before AMD can address a significant amount of those form factors, Seifert said. But some tablet makers are buying AMD chips now, he added.

"So we see that there are parts of this form factor that move in our direction," Seifert said, according to the official transcript. "And we were really happy to see the Acer just in the beginning of the month launched the first Windows based tablet based on our low power Bobcat architecture. So we see that there are usage scenarios in this tablet segment that move towards our capability. At the same time, we are continuing to work hard on reducing our power consumption for our products in the low power segment and the same time being able to deliver and continue to deliver cutting edge graphics capability. So especially with the products--with the second generation APU products that are on the roadmap for next year, we feel confident that we can also address a significant amount of those form factors out in the tablet space."

He also said that a significant source of profits remained the server segment, and that it was important "not to be distracted from where those opportunities are".

Sounds like they expect a really good year for Fusion APU. (which is using up all their energy/time/resources). Makes sense.

But what about a future where ARM and/or docked smartphones become sufficient for most users? Will they still have time to innovate in that direction as well?

Lately the company put quite a bit of emphasis on GPGPU and OpenCL. Can they continue this same line of thinking into the ARM era?

 

[Idontcare]

In short, even I have no f'ing idea why they pushed Meyer out, I'm not doing anything different as acting CEO that Dirk was not intending to do as the then sitting CEO. So, go figure.

The 1.8GHz quip was funny too. No wonder they pushed the release back.

 

[cbn]

http://blogs.wsj.com/digits/2011/02/17/amds-short-term-chief-talks-transition-tablets/?mod=yahoo_hs

Thomas Seifert says he never sought to run Advanced Micro Devices. But while he’s at it, he’s not shy about gloating over the chip maker’s next big product–and giving clues about AMD’s aims in the mobile market.

Seifert, AMD’s chief financial officer, was elevated in January to interim chief executive following the sudden departure of Dirk Meyer. He said he immediately opted to take his name out of consideration for the permanent CEO job, because he wanted to avoid any indication that friction among AMD senior executives could have been a factor in Meyer’s resignation.

Source: Advanced Micro Devices
AMD’s Thomas Seifert

The management change “was a board decision,” Seifert said Wednesday, in an interview following his appearance at a Goldman Sachs technology conference in San Francisco. “I was as surprised as anybody else.”

The board, Seifert said, did not act because of anything the popular CEO had done or not done, but questions about AMD’s future. “Are we on the right trajectory? Do we set our objectives high enough?” Seifert said. In considering such questions, he said, directors concluded a different set of skills than Meyer’s were needed–prompting a CEO search that is now underway.

One of the big questions–about both AMD and arch-rival Intel–is how effectively the two chip makers can diversify beyond computers. In the biggest potential target, the cellphone market, chips based on designs from ARM Holdings dominate; the x86 variety, which is sold by both Intel and AMD, has made no headway. ARM technology is also at the heart of Apple’s iPad and looks an early favorite in other tablets.

Seifert, a former Qimonda chief operating officer who joined AMD in 2009, stressed that it has no current plan to go after smartphones, unlike Intel. One reason is that future processors for those devices are likely to be included on the same piece of silicon with baseband processors that manage cellular communications. Intel has purchased baseband-chip capabilities from Infineon; AMD doesn’t have baseband technology.

Tablets are likely to be a different story. AMD sees an interesting opportunity in that market “at the right point in time,” Seifert said.

AMD will have to lower the power consumption of its chips to play a big role in tablets. But it already has one advantage–tablets are likely to be used for handling video and other multimedia chores, where AMD’s graphics technology excels.

But what if hardware makers continue to prefer ARM-based microprocessors, which run popular software such as Google’s Android? Might AMD license and sell that technology, too? Seifert doesn’t rule it out, saying the company has to be flexible. There is so much change in the market “you cannot be dogmatic about it,” he said.

The big short-term opportunity for AMD, Seifert says, will come late in the second quarter with the somewhat-belated arrival of a powerful chip for PCs dubbed Llano that adds graphics and other circuitry along with processors. AMD gave demonstrations Wednesday of a prototype computer powered by the new chip performing a set of chores alongside a laptop PC running Intel’s latest chips, code-named Sandy Bridge.

Intel for several years has enjoyed advantages over AMD chips in performance and power consumption. In the demonstration, however, loading the competing systems with multiple jobs that exploit graphics circuitry caused images to run much more smoothly on the Llano system than Intel’s technology–and it drew 10 watts to 30 watts less power.

An Intel spokesman did not immediately respond to requests for comment.

Because of a recent glitch with a support chip, Intel’s broad rollout of Sandy Bridge systems has been delayed. With its shipments now expected to reach full volume by April, Intel’s products will still arrive in systems months ahead of Llano. But the glitch cuts into Intel’s head start, Seifert said. “It’s an exciting time,” he added.

Two interesting points:

1. AMD states one reason they will not go into smartphones has to do with baseband being likely integrated into the SOC.

So how many companies make baseband chips? Is there anyway they could partner with one of them in such a way that is not a conflict of interest?

2. AMD is flexible in terms of ISA.

 

[busydude]

The 1.8GHz quip was funny too. No wonder they pushed the release back.

LOL seriously, but in your experience.. have you seen a huge change in performance going from ES to final launch chip?

1.8GHz sounds terribad.

 

[itsmydamnation]

LOL seriously, but in your experience.. have you seen a huge change in performance going from ES to final launch chip?

1.8GHz sounds terribad.

and if its a 25watt TDP including GPU part? statement without context is meaningless.

 

[Arkadrel]

the dual core 1.6Ghz botcat + GPU is terribad too then Busydude? Thats at 18watts...

What if the 1.8ghz llano is a small dual core, like the bobcat, ment to go into laptops (low TPD), and it comes with a GPU thats 3x times as powerfull as the one in the bobcats?

Like Itsmydamnation said, without context its meaningless.

 

[Idontcare]

LOL seriously, but in your experience.. have you seen a huge change in performance going from ES to final launch chip?

Yes, of course, that's one of the big reasons why ES is ES and not retail already.

1.8GHz sounds terribad.

That wasn't what I meant. The quip was funny because the way it was authored in the OP quote block they made it sound like Seifert said something that was terribly embarrassing.

An AMD spokesman said that Seifert did not mean to imply that that was the speed that Llano would run at.

You guys can make excuses if you like, but it is obvious that the AMD spokesperson concluded that some damage control was needed after Seifert let the 1.8Hz comment slip out.

Note the spokesperson does not say "yeah, but seriously folks on a performance/watt basis this IS the bomb! don't let this seemingly low demo clockspeed raise any red flags, all is well, all is well".

For me the context is perfectly clear, AMD compared it to Sandy Bridge more than once. The spokesperson obviously felt compelled to qualify Seifert's statement because of the implied context and that clarification was to specifically comment on the clockspeeds, the same context which I was aiming to address with my post.

 

[busydude]

That wasn't what I meant.

I was not trying to put words in your mouth, sorry if it came off that way. That statement I made was my reaction to that 1.8 GHz quip... its totally unrelated to the first statement I made.

the dual core 1.6Ghz botcat + GPU is terribad too then Busydude? Thats at 18watts...

Bobcat met or even exceeded every metric that AMD promised to deliver.. it is nothing short of an engineering marvel in my book.

Llano, just like BD, was promised to run at relatively high clocks. I would be surprised if AMD increased the IPC of Llano when compared to Phenom II. Just think about it.. its just a die shrunk Phenom II running at 1.8 GHz.

As IDC pointed out.. its nothing but damage control. The statement "Architecture is incredibly superior".. sounds cheesy to say the least. How can a die shrunk stars core be architecturally superior to sandy bridge?

 

[Idontcare]

I was not trying to put words in your mouth, sorry if it came off that way. That statement I made was my reaction to that 1.8 GHz quip... its totally unrelated to the first statement I made.

Not at all bud, it doesn't and I didn't. I was terse and that left my post open to interpretation, that's my bad all the way.

That said, AMD makes miracles happen that no other company save Intel can accomplish. They only look bad because they stand next to Intel. Compared to the rest of the industry AMD is a giant and always has been.

Bobcat met or even exceeded every metric that AMD promised to deliver.. it is nothing short of an engineering marvel in my book.

It really is.

To be manufactured on a foundry process no less, without the benefit of HK/MG or SOI, and yet manage to deliver those kinds of clocks and that sort of power consumption AND that level of performance...well it would be cliche for me to say it is "the stuff of legends" or that it is epic, but in the professional circles with which I associate it really is all that.

The problem for AMD, as a business, is that so too was the Alpha 21164...and yet DEC sank regardless the engineering prowess of the talent they had accumulated and assembled.

They need to really drive this thing into volume shipments and marketshare wins to justify having developed it in the first place.

 

[Arkadrel]

Llano, just like BD, was promised to run at relatively high clocks. I would be surprised if AMD increased the IPC of Llano when compared to Phenom II. Just think about it.. its just a die shrunk Phenom II running at 1.8 GHz.

What if the Llano comes in differnt versions? slower dual core versions that are ment for laptops? that are faster than the botcat cores, and have 3 times as much gpu power?

could mean=
1.8Ghz dual core Llano for Laptops (say around 35watts).
3.0Ghz quad core Llano for desktop pcs (say around 60watts or something).

Would that still make Llano a disappointment?

How fast are the sandybridge CPUs that go into laptops (2ghz+ ?)? what TPD do those have? How much more than a Llano will it cost?

 

[cbn]

What if the Llano comes in differnt versions? slower dual core versions that are ment for laptops? that are faster than the botcat cores, and have 3 times as much gpu power?

I think I read somewhere that Llano would come on two wafers. One lower power for laptop and one high power for desktop. (It would be great to be able to confirm if this is true or not in the future.)

 

[IntelUser2000]

The only reason they aren't going in smartphones in the near future is because they won't have a product that'll fit in one. Plus these kind of articles always mean only 1-2 years into the future. Doesn't mean they won't ever do it.

 

[sandorski]

The Smartphone market is exceedingly competitive. Almost every other month or so someone releases a new superior CPU/GPU. Really doesn't seem like something AMD should rush into.

 

[cbn]

The Smartphone market is exceedingly competitive. Almost every other month or so someone releases a new superior CPU/GPU. Really doesn't seem like something AMD should rush into.

Agreed.

I'm just a little concerned about Tablets dying off (through software) and people just hooking a phone up to a larger screen ("Tab dock?") or something like the Atrix 4G lap dock?

If something like that begins to happen, what kind of technology could AMD bring to the table its competitors don't have?

Maybe Crossfire (or SFR) between an AMD phone and larger device with a GPU inside it?

P.S. Maybe AMD can start off at the bottom of the market? (instead of the crowded top). I made a suggestion here of Cortex A5 vs the weaker existing ARM11 Application processor technology in this thread.

 

[ElFenix]

why would llano be for relatively high clocks? it's a die-shrunk stars core with a halfway decent gpu tacked on.

 

[Arkadrel]

why would llano be for relatively high clocks? it's a die-shrunk stars core with a halfway decent gpu tacked on.

Why would llano be for high clocks? it wont... its a mainstream OEM product, ment to bring cheap cpu/gpu to the masses, with GPGPU being at the center.

There will be quad core Llanos, that run stock 3Ghz+, that are ment for the desktops.

And then there will probably 2-4 cores, that run ~1.8Ghz and are ment for Laptops.

 

[Mopetar]

So how many companies make baseband chips? Is there anyway they could partner with one of them in such a way that is not a conflict of interest?

Qualcomm is, but they also make their own line of ARM SoCs so they're probably more interested in building their own all-in-one solution. Another large manufacturer is Texas Instruments, but they also have their own ARM SoC.

Otherwise, Broadcom, Infineon, MediaTek, and ST-Ericsson are all in the baseband business. As far as I know, none of these companies are making their own ARM chips, or at least not any mass market designs so it might be possible to pair up or look into merging with one of these companies.