AMD tiptoes past sparring rivals in embedded CPU foray

There's more to life than PCs and servers

The new management team at Advanced Micro Devices is looking everywhere, including under the couch cushions, to find some money so it can afford to explore the embedded systems market again. The chip biz hopes rivals Intel and the ARM collective are too distracted to notice the foray as they fight over each others' territories in PCs, servers and mobile devices.

With the move, AMD is pitting itself against the embedded Power processor designs from each of IBM and Freescale Semiconductor; the eponymous RISC chips of MIPS Technologies; the embedded x64 products from Intel; and a myriad other niche products based on a variety of processor architectures.

AMD cites statistics from VDC Research Group saying that the chip firm's new Embedded Solutions Group can now chase an embedded processor market that is expected to hit $6bn in sales in 2012 and to grow at a rate ranging between 12 per cent and 15 per cent in the next two years.

Advanced Micro Devices plans to pitch both the embedded versions of its Fusion Accelerated Processing Unit (APU) designs, which marry a CPU with a GPU on a single package, as well as discrete graphics and low-powered Opteron processors to such customers, who are looking for both performance and low power.

AMD was way out in front of the market with its Opteron Highly Efficient (HE) and Extremely Efficient (EE) parts, which debuted soon after the original Opteron processors as low voltage and therefore low power options that were well suited to embedded systems with modest processing requirements.

The chip maker also geared down the clocks to make them even cooler for certain SKUs. The Opteron HE chips originally ran at 55 watts and the EE chips at 30 watts back when the first ones came out in February 2004 and ran in 95 watt thermal envelopes.

The problem with the HE and EE parts back then is that they were etched in 130 nanometer processes and you simply could not cram a lot of cores and cache onto the chips like you can today, so ramping down the clock speed had a very dramatic effect on the oomph of the chip. Today, with a 32 nanometer process shrink, you can put two or four cores on the chip and run it at a reasonable clock speed with lots of L1 and L2 cache and get a low-power package that can nonetheless still do some work.

In the wake of last year's restructuring, AMD made no secret that it was interested in the embedded market, saying in November that it would put out the "Zurich" Opteron 3200 chip for single-socket machines - which includes both microservers for hyperscale data centers as well as other kinds of embedded systems used in the telco, military, and electronics areas.

Watts at stake in AMD embedded adventure

The Opteron 3200 came out in late March with four-core variants burning at 45 watts and an eight-core weighing in at 65 watts. AMD also has two Opteron 4200 HE part with six or eight cores that consumes 65 watts and an Opteron 4200 EE part with eight cores that sucks 35 watts. What the company has not yet done is pump out Opteron 3200 parts that run at the lower HE or EE voltages, so this is the obvious next deep-bin-sorting move for the chip maker.

It would also make sense for AMD to create server variants of its "Bobcat" family of laptop processors, something the AMD before Rory Read dismissed as silly. Greg Hoepper, corporate vice president of design engineering who managed the Bobcat core design, told El Reg in August 2010, ahead of the launch of chips based on Bobcat cores, that you could put an "enormous number of these on a single die". However, AMD balked at the suggestion that perhaps, for the sake of multithreaded server workloads, this might be a good idea.

Maybe not so any more, with Calxeda, Marvell, Applied Micro, and others gearing up to do ARM variants for servers.

AMD does peddle Bobcat chips to embedded systems makers, including the G-Series APUs first launched in January 2011 and positioned against Intel's Atom chips and the R-Series, which has beefier graphics and goes up against Intel's embedded Core i3, i5, and i7 variants.

There are also embedded variants of Athlon, Turion, Sempron, and Geode LX processors for all kinds of devices.

The new Embedded Solutions Group will be tucked up underneath AMD's Global Business Units organization, which was created late last year when Read cleaned house and hired Lisa Su - a semiconductor researcher and manager who did stints at Freescale, IBM Microelectronics, and Texas Instruments - to rationalize and organize the company's client, commercial, graphics, professional graphics and game console products.

AMD has hired Arun Iyengar, a vice president at field programmable gate array (FPGA) maker Altera, to be general manager of the Embedded Solutions Group. While at Altera, Iyengar ran its military, industrial, computing (MIC) division, and before that he was in charge of the company's communications unit. Iyengar got his BS in engineering from Bangalore University and his MBA from Carnegie Mellon University.

It seems unlikely that Iyengar's hire is a precursor for AMD to get into the FPGA racket, and it also seems unlikely that AMD would venture out into the ARM chip space, either. But should AMD make either move, it certainly would not be boring.

Altera has a market capitalization of $10.8bn right now and had $2.1bn in sales last year. That makes Altera considerably more valuable than AMD, at least as far as Wall Street is concerned, with its $4.1bn market cap against its $6.57bn in sales in 2011.

FPGA maker Xilinx is similarly too expensive for AMD to buy, with a market cap of $8.67bn against $2.24bn in revenues last year. If anything, Altera or Xilinx could eat AMD if they went wild.

As for ARM chips, like Intel, AMD has been perfectly content to stay with the x64 instruction set and not wander out into ARM RISC. This, one could argue, is precisely why AMD should get in before the market for PC and server chips powered by ARM's architecture gets too crowded. But there is no guarantee at all that ARM will take off in either arena, or hold its hegemony in smartphone and other mobile devices. Anything is possible when a very rich Intel focuses. ®

Broader topics

Narrower topics

Other stories you might like

  • Intel is running rings around AMD and Arm at the edge
    What will it take to loosen the x86 giant's edge stranglehold?

    Analysis Supermicro launched a wave of edge appliances using Intel's newly refreshed Xeon-D processors last week. The launch itself was nothing to write home about, but a thought occurred: with all the hype surrounding the outer reaches of computing that we call the edge, you'd think there would be more competition from chipmakers in this arena.

    So where are all the AMD and Arm-based edge appliances?

    A glance through the catalogs of the major OEMs – Dell, HPE, Lenovo, Inspur, Supermicro – returned plenty of results for AMD servers, but few, if any, validated for edge deployments. In fact, Supermicro was the only one of the five vendors that even offered an AMD-based edge appliance – which used an ageing Epyc processor. Hardly a great showing from AMD. Meanwhile, just one appliance from Inspur used an Arm-based chip from Nvidia.

    Continue reading
  • AMD’s AI strategy comes into view with Xilinx, GPU, software plans
    Chip designer hopes to have broad inference and training coverage from the edge to the cloud

    Analysis After re-establishing itself in the datacenter over the past few years, AMD is now hoping to become a big player in the AI compute space with an expanded portfolio of chips that cover everything from the edge to the cloud.

    It's quite an ambitious goal, given Nvidia's dominance in the space with its GPUs and the CUDA programming model, plus the increasing competition from Intel and several other companies.

    But as executives laid out during AMD's Financial Analyst Day 2022 event last week, the resurgent chip designer believes it has the right silicon and software coming into place to pursue the wider AI space.

    Continue reading
  • AMD touts big datacenter, AI ambitions in CPU-GPU roadmap
    Epyc future ahead, along with Instinct, Ryzen, Radeon and custom chip push

    After taking serious CPU market share from Intel over the last few years, AMD has revealed larger ambitions in AI, datacenters and other areas with an expanded roadmap of CPUs, GPUs and other kinds of chips for the near future.

    These ambitions were laid out at AMD's Financial Analyst Day 2022 event on Thursday, where it signaled intentions to become a tougher competitor for Intel, Nvidia and other chip companies with a renewed focus on building better and faster chips for servers and other devices, becoming a bigger player in AI, enabling applications with improved software, and making more custom silicon.  

    "These are where we think we can win in terms of differentiation," AMD CEO Lisa Su said in opening remarks at the event. "It's about compute technology leadership. It's about expanding datacenter leadership. It's about expanding our AI footprint. It's expanding our software capability. And then it's really bringing together a broader custom solutions effort because we think this is a growth area going forward."

    Continue reading

Biting the hand that feeds IT © 1998–2022