Why Amazon is Not Interested in TI’s Mobile Processor Group

Image Credit: iFixit Some reports have came out that Amazon is interested in TI’s mobile processor division. I find this extremely difficult to believe. I personally, think TI’s move to shift focus from mobile APs (application processors) and more into embedded chipsets is a fascinating market development. However, I think it is a serious stretch to connect the dots that Amazon would use their extremely valuable cash to acquire something they don’t really need given their business model.

If Amazon was in the for profit hardware business then I can see how a case could potentially be made for purchasing TI’s OMAP mobile AP group. However they are not in the for profit hardware business and are rather in the hardware as a service business. They make practically nothing on the hardware and within the business model they are entrenched in, it would take an incredibly long time if ever to recoup their investment in a semiconductor group.

Furthermore, TI licenses and ships the ARM core but does not have an architectural license to customize or alter the chipset design like Qualcomm, NVIDIA, Apple, and Samsung do. If the logic was that Amazon was going to use proprietary semiconductor assets to help them further differentiate their hardware, I’m not sure acquiring TI’s OMAP group would do this. In fact if this was the logic then it would be just as easy, and probably cheaper, for Amazon to acquire an ARM architecture license and simply hire a team of qualified SOC engineers.

However, because Amazon is building a hardware as a service business, it seems unlikely that making a large investment around hardware makes sense. Companies that are in the hardware as a service model are generally better served simply negotiating and buying components rather than making them.

Several reports also mentioned Amazon’s intent to get into the smartphone market and speculated that buying this group from TI could help this initiative. I continue to remain skeptical that Amazon will make a smartphone. I simply can’t see how it fits with their business model. Amazon is a retailer and any argument as to why a retailer should make a smartphone would be null in light of an argument that those same reasons can be accomplished with an app running on any platform. The many reasons why Amazon (a retailer) made a tablet does not translate into why they should make a smartphone.

Of course we can’t rule anything out in this industry, especially considering I would have never guessed an advertising company would have gotten into the smartphone hardware business. Or could I?

HSA Foundation: for Show or for Real?

I recently spent a few days at AMD’s Fusion Developer Summit in Seattle, Washington.  Among many of the announcements was one to introduce the HSA Foundation, an organization  currently including AMD, ARM,  Imagination, MediaTek, and Texas Instruments.  The HSA Foundation was announced to “make it easy to program for parallel computing.”  That sounds a bit like an oxymoron as parallel programming has been the realm of “ninja programmers” according to Adobe’s Chief Software Architect, Tom Malloy at AMD’s event.  Given today’s parallel programming challenge, lots of work needs to be done to make this happen, and in the case of the companies above, it comes in the form of a foundation.  I spent over 20 years planning, developing, and marketing products and when you first hear the word “foundation” or “consortium” it conjures up visions of very long and bureaucratic meetings where little gets done and there is a lot of infighting.  The fact is, some foundations are like that but some are extremely effective   like the Linux Foundation. So which path will the HSA Foundation go down?  Let’s drill in.

The Parallel/GPU Challenge

The first thing I must point out is that if CPUs and GPUs keep increasing compute performance at their current pace, the GPU will continue to maintain a raw compute performance advantage over the CPU, so it is very important that the theoretical performance is turned into a real advantage.  The first thing we must do is distinguish is between serial and parallel processing.  Don’t take these as absolutes, as both CPUs and GPUs can both run serially and in parallel.  Generally speaking, CPUs do a better job on serial, out of order code, and GPUs do a better job on parallel, in-order code.   I know there are 100’s of dependencies but work with me here.  This is why GPUs do so much better on games and CPUs do so well on things like pattern matching. The reality is, few tasks just use the CPU and few just use the GPU; both are required to work together and at the same level to get the parallel processing gains.  By working at the same level I mean getting the same access to memory, unlike today where the CPU really dictates who gets what and when.  A related problem today is that coding for the GPU is very difficult, given the state of the languages and tools.  The other challenge is the numbers of programmers who can write GPU versus CPU code.  According to IDC, over 10M CPU coders exist compared to 100K GPU coders.  Adobe calls GPU coders  “ninja” developers because it is just so difficult, even with tools like OpenCL and CUDA given they are such low level languages.  That’s OK for markets like HPC (high performance computing) and workstations, but not for making tablet, phone and PC applications that could use development environments such as the Android SDK or even Apple’s XCode.  Net-net there are many challenges for a typical programmer to code an GPU-accelerated app for a phone, tablet, or a PC.

End User Problem/Opportunity

Without the need to solve an end user or business problem, any foundation is dead in the water.  Today NVIDIA  is using CUDA (C, C++, C#,), OpenCL, and OpenACC and AMD supports OpenCL to solve the most complex industrial workloads in existence.  As an example, NVIDIA simulated at their GTC developer conference what the galaxy would look like 3.8B years in the future.  Intel is using MIC, or Many Integrated Cores to tackle these huge tasks.  These technologies are for high-performance computing, not for phones, tablets or PCs. The HSA Foundation is focused on solving the next generation problems and uncovering opportunities in areas like the natural user interface with a multi-modal voice, touch and gesture inputs, bio-metric recognition for multi-modal security, augmented reality and managing all of the visual content at work and at home.  ARM also talked on-stage and in the Q&A about the power-savings they believed they could attain from a shared memory, parallel compute architecture, which surprised me.  Considering ARM powers almost 100% of today’s smartphones and tablets around the world, I want to highlight what they said.  Programming for these levels of apps at low power and enabling 100’s of thousands of programmers ultimately requires very simple tools which don’t exist today to create these apps.

The HSA Foundation Solution

The HSA Foundation goal, as stated above, was to “make it easy to program for parallel computing.” What does this mean?  The HSA Foundation will agree on hardware and software standards.  That’s unique in that most initiatives are just focused on the hardware or the software.  The goal of the foundation is to literally bend the hardware to fit the software.  On the hardware side this first means agreement on the hardware architectural definition of the shared memory architecture between CPU and GPU.  This is required for the CPU and GPU to be at the same level and not be restricted by buses today like PCI Express.  The second version of that memory specification can be found here.  The software architecture spec and the programmer reference manual are still in the working group.  Ultimately, simple development environments like the Google Android SDK, Apple’s XCode and Microsoft’s Visual Studio would need to holistically support this to get the support of the more mainstream, non-ninja programmer.  This will be a multi-year effort and will need to be measured on a quarterly basis to really see the progress the foundation is making.

Foundations are Tricky

The HSA Foundation will encounter issues every other foundation encounters at one time in its life.  First is the challenge of founding members changing their minds or getting goal-misaligned.  This happens a lot where someone who joins stops buying into the premise of the group or staunchly believes it isn’t valuable anymore.  Typically that member stops contributing but could even become a drag on the initiative and needs to be voted off.  The good news is that today, AMD, ARM, TI, MediaTek and Imagination have a need as they all need to accelerate parallel processing.  The founding members need to make this work for their future businesses to be as successful as they would like. Second challenge is the foundation is missing key players in GPUs.  NVIDIA is the discrete GPU PC and GPU-compute market share leader, Intel is the PC integrated GPU market share leader, and Qualcomm is the smartphone GPU market share leader.  How far can the HSA Foundation get without them?  This will ultimately be up to guys like Microsoft, Google and Apple with their development environments.  One wild-card here is SOC companies with standard ARM licenses.  To get agreement on a shared memory architecture, the CPU portion of ARM SOC would need to be HSA-compliant too, which means that every standard ARM license derived product would be HSA-compliant.  If you had an ARM architecture license like Qualcomm has then it wouldn’t need to be HSA-compliant.  The third challenge is speed.  Committees are guaranteed to be slower than a partnership between two companies and obviously slower than one company.  I will be looking for quarterly updates on specifications, standards and tools.

For Show or for Real?

The HSA Foundation is definitely for real and formed to make a real difference.  The hardware is planned to be literally bent to fit the software, and that’s unique.  The founding members have a business and technical need, solving the problem means solving huge end user and business problems so there is demand, and the problem will be difficult to solve without many companies agreeing on an approach.  I believe over time, the foundation will need to get partial or full support from Intel, NVIDIA, and/or Qualcomm to make this initiative as successful as it will need to be to accelerate the benefits of parallel processing on the GPU.

 

 

Gaming AMD’s 2012 Strategy

AMD intends to pursue “growth opportunities” in low-powered devices, emerging markets and Internet-based businesses.

There’s an awful lot of mis-guided analysis wafting about regarding AMD’s new strategic direction, which the company says it will make public in February. This piece is to help you (and me) sort through the facts and the opportunities. I last took a look at AMD’s strategies earlier this year, available here.

Starting With the Facts

  • AMD is a fabless semiconductor company since 2009. The company depends on GlobalFoundries and soon Taiwan Semiconductor to actually fabricate its chips;
  • In its latest quarter, AMD had net income of about $100 million on $1.7 billion in revenue. Subsequently, the company announced a restructuring that seeks to cut costs by $118 million in 2012, largely through a reduction in force of about ten percent;
  • AMD has about a 20% market share in the PC market, which Intel says is growing north of 20% this year, largely in emerging markets;
  • AMD’s products compete most successfully against rival Intel in the low- to mid-range PC categories, but 2011 PC processors have underwhelmed reviewers, especially in performance as compared to comparable Intel products;
  • AMD has less than a 10% market share in the server market of about 250,000 units, which grew 7.6% last quarter according to Gartner Group;
  • AMD’s graphics division competes with nVidia in the discrete graphics chip business, which is growing in profitable commercial applications like high-performance supercomputing and declining in the core PC business as Intel’s integrated graphics is now “good enough” for mainstream buyers;
  • AMD has no significant expertise in phone and tablet chip design, especially the multi-function “systems on a chip (SOCs)” that make up all of today’s hot sellers.

What Will AMD CEO Rory Read’s Strategy Be?

I have no insider information and no crystal ball. But my eyebrows were seriously raised this morning in perplexity to see several headlines such as “AMD to give up competing with Intel on X86“, which led to “AMD struggling to reinvent itself” in the hometown Mercury News. I will stipulate that AMD is indeed struggling to reinvent itself, as the public process has taken most of 2011. The board of directors itself seems unclear on direction. That said, here is my score card on reinvention opportunities in descending order of attractiveness:

  1. Servers —  For not much more work than a desktop high-end Bulldozer microprocessor, AMD makes Opteron 6100 server processors. Hundreds or thousands more revenue dollars per chip at correspondingly higher margins. AMD has a tiny market share, but keeps a foot in the door at the major server OEMs. The company has been late and underdelivered to its OEMs recently. But the problem is execution, not computer science.
  2. Desktop and Notebook PCs — AMD is in this market and the volumes are huge. AMD needs volume to amortize its R&D and fab preparation costs for each generation of products. Twenty percent of a 400 million chip 2011 market is 80 million units! While faster, more competitive chips would help gain market share from Intel, AMD has to execute profitably in the PC space to survive. I see no role for AMD that does not include PCs — unless we are talking about a much smaller, specialized AMD.
  3. Graphics Processors (GPUs) — ATI products are neck-and-neck with nVidia in the discrete graphics card space. But nVidia has done a great job of late creating a high-performance computing market that consumes tens of thousands of commercial-grade (e.g., high price) graphics cards. Intel is about to jump into the HPC space with Knight’s Corner, a many-X86-core chip. Meanwhile, AMD needs the graphics talent onboard to drive innovation in its Fusion processors that marry a processor and graphics on one chip. So, I don’t see an AMD without a graphics component, but neither do I see huge profit pools either.
  4. Getting Out of the X86 Business — If you’re reading along and thinking you might short AMD stock, this is the reason not to: the only legally sanctioned software-compatible competition to X86 inventor Intel. If AMD decides to get out of making X86 chips, it better have a sound strategy in mind and the ability to execute. But be assured that the investment bankers and hedge funds would be flailing elbows to buy the piece of AMD that allows them to mint, er, process X86 chips. So, I describe this option as “sell off the family jewels”, and am not enthralled with the prospects for success in using those funds to generate $6.8 billion in profitable revenue or better to replace today’s X86 business.
  5. Entering the ARM Smartphone and Tablet Market— A sure path to Chapter 11. Remember, AMD no longer makes the chips it designs, so it lacks any fab margin to use elsewhere in the business. It starts against well-experienced ARM processor designers including Apple, Qualcomm, Samsung, and TI … and even nVidia. Most ARM licensees take an off-the-shelf design from ARM that is tweaked and married to input-output to create an SOC design, that then competes for space at one of the handful of global fab companies. AMD has absolutely no special sauce to win in the ARM SOC kitchen.To win, AMD would have to execute flawlessly in its maiden start (see execution problems above), gain credibility, nail down 100+ design wins for its second generation, and outrace the largest and most experienced companies in the digital consumer products arena. Oh, and don’t forget volume, profitability, and especially cash flow. It can’t be done. Or if it can be done, the risks are at heart-attack levels.

“AMD intends to pursue “growth opportunities” in low-powered devices, emerging markets and Internet-based businesses.” One way to read that ambiguous sentence by AMD is a strategy that includes:

  • Tablets and netbooks running X86 Windows 8;
  • Emerging geographic markets, chasing Intel for the next billion Internet users in places like Brazil, China, and even Africa. Here, AMD’s traditional value play resonates;
  • Internet-based businesses such as lots of profitable servers in the cloud. Tier 4 datacenters for Amazon, Apple, Facebook, Google, and Microsoft are a small but off-the-charts growing market.

So, let’s get together in February and see how the strategy chips fall. Or post a comment on your game plan for AMD.