EE Daily News: December 2011

I first heard of the Spin Clean Record Cleaner when Michael Fremer reviewed it for Sterophile Magazine, in February 2010. The device also was named as one one of the runners up for Stereophile's Accessory Product of the Year in 2010.

As a frugal audiophile, I have used the DiscWasher D4 record cleaning brush and fluid since I first purchased my Yamaha YP-701 turntable (still working great after ~35 years!). The more sophisticated LP vacuum systems are just too expensive and bulky for me. But the Spin Clean sells for just $79.99, so I finally picked one as a Christmas present to myself a few weeks ago. I've acquired some LPs from the 50s and 60s in the last few years, and I was anxious to see if a more thorough cleaning could make an improvement in their sound.

The Mk II consists of a plastic container that you fill with distilled water, then add just a few drops of the supplied fluid to the cleaning pads. You then drop the two wheels into the appropriate slot for the size record you are cleaning, and you're ready to start cleaning. Spin Clean recommends that you plan a "cleaning session" for 25-50 records at a time. That's more than I need to do, and with such a small amount of cleaning fluid needed the supplied 4 ounce bottle should last me for years.

With the cleaner all setup, you slide a record into place and the wheels hold it so that the label is just above the cleaning pads and fluid. Turn the record three times clockwise (it is held snugly between the pads), then reverse and turn the record in the bath three times counter-clockwise. Lift the record out and let some of the excess fluid drip back into the basin before you lay it down to dry. I used paper towels to protect the records from my work surface, though the manufacturer doesn't provide any recommendations in this regard. I didn't want to introduce any new particles after washing the records, and this seemed to work fine.

I found that the supplied cleaning cloths do a good job of drying a record, but they are lightweight and get wet after a few cleanings. I couldn't see doing as many as 25 records without having several more of these on hand.

After cleaning approximately a dozen records, I was ready to hear some results. Half of the LPs I cleaned were given to me by my uncle, who was a big jazz aficionado. The other (long-abandoned) group was rescued from my mother-in-law's apartment in Belgium, where they had been stored in boxes for ~40 years. The second group had visible dust and fingerprints that were completely removed in the cleaning.

The results? I was blown away! The Shelly Manne 234 album was recorded in 1962, and it sounded as good as new. This disk was probably fairly well taken care by my uncle, but he never had what you would call an audiophile setup. This one LP is just about enough to justify my purchase. A more recent pressing will cost you $62 on Amazon. What a thrill it was to hear this and imagine my uncle tapping his foot to the beat almost fifty years ago!

Following on Quantenna Communications' recent claim to be first to market with a 802.11ac chipset and reference design, add Broadcom and Redpine Signals to the companies planning to rollout chipsets for the next-generation IEEE WiFi standard. Though the standard will not complete the IEEE 802.11 Working Group ratification process until December 2013, Broadcom and Redpine have both announced plans to have chipsets available by the 2nd half of 2012.

In a pre-Consumer Electronics Show (CES) briefing on December 8th, Broadcom's VP for the Mobile and Wireless Group - Rahul Patel talked about his companies plan to develop "several types of .11ac chipsets" targeting Consumer Electronics (CE) devices. Patel focused on shifts in video consumption in the "post-PC" era, citing statistics from the October, 2011 Sandvine Broadband Report. Sandvine found that real-time on-demand entertainment now accounts for 60% of North America’s fixed access network traffic during the evening’s peak period. Broadcom's target for their new wireless chipsets is the "post-PC era" shift, which Sandvine identified, in the type of devices consumers are using to consume their digital entertainment:

By volume, 55% of Real-Time Entertainment traffic is destined for the television (either directly to a smart TV or via an intermediary like a game console or set-top device), a mobile device or tablet. The remaining 45% is being delivered to desktop and laptop computers; even then, some will be forwarded to a television. (source: Sandvine - "Global Internet Phenomena Report", Fall 2011)

The digital living room is now becoming a network of wireless connected devices, with game consoles, smartphones and tablet PCs increasingly used to deliver video to the TV, typically via an HDMI cable. Broadcom sees the emerging 802.11ac standard, which will shift use entirely to the 5GHz band, as the enabler to deliver faster throughput and broader coverage to CE, while supporting simultaneous connections for a larger number of devices. The new standard allows for wider channels, up from 40Mhz in IEEE 802.11n to 80MHz and 160MHz, with doubling of support for up to eight Multiple-Input/Multiple-Output (MIMO) streams, and an increase from 64-Quadrature Amplitude Modulation (QAM) to 256-QAM. Broadcom says that they expect to begin shipping 802.11ac products in the second half of 2012.

Ten-year old Redpine Signals has announced that they are ready to take 802.11ac to mobile devices, with their claim to have the "industry’s first IEEE 802.11ac technology offering for integration into smartphone application processors". Venkat Mattela, Chairman and Chief Executive Officer of Redpine, says that the company's focus from the beginning has been on low power techniques for WiFi chipsets. In 2005 Redpine licensed their 802.11b/g technology to PortalPlayer, a company that NVIDIA acquired in 2007, which had been providing SoCs to Apple for the iPod. Mattela says that two years later, Infineon also licensed Redpine's low power technology.

In March of this year Redpine completed development of their 802.11ac technology, and for the last nine months, according to Mattela, they have been working with "a big mobile company" for integration into their smartphone platform. Mattela says that he only licenses to companies that ship 300M-400M devices per year, and his 3rd licensing deal is "happening anytime". Redpine also manufactures their own chips and modules, the latter focused on industrial market segments.

Although Redpine designs the complete RF/Analog Front-End (AFE) for their WiFi subsystems, they have licensed just the analog and digital baseband functions, as a combination of hard and soft IP, to companies who have their own RF designs and Power Amplifiers (PAs). Mattela says that licensees gain assurance from access to the soft IP, but due to complexity of the WiFi subsystem it is always his company's team that develops the complete macro for integration into a customer's SoC. Rather than use an ARM or MIPS processor in their WiFi subsystem, Redpine developed their own "ThreadArch" processor for Layer-2 WiFi processing. Mattela says that this, and their low-power Orthogonal Frequency-Division Multiplexing (OFDM) engine, are key differentiating factors that contribute to lower power consumption by requiring fewer CPU cycles .

For mobile handset applications, Mattela says that initially the industry thought that only single-stream WiFi implementations would be required, but the need for higher data rates has driven the use of MIMO with dual antennas. He sees this being the case for 802.11ac as well, with handsets employing up to 80MHz channels on the 5GHz carrier. The standard allows up to 160MHz channels, but this causes other problems, says Mattela, and is not necessary for mobile devices. Along with the Gigabit data rates of 802.11ac, Mattela sees the capability for multi-user MIMO, which will enable spreading the bandwidth over a larger number of users, as being the most critical factor that will drive adoption of the new standard. This will alleviate the problem of hotspots with high user density, common to demos at places like CES, where connectivity can be locked out by a single user consuming all of the available channel bandwidth.

The higher data rates of 802.11ac could be expected to result in higher power consumption, but Redpine used the profile of their previous generation 802.11n design, with <1mW consumption while connected to an access point, as a reference for the new design. Mattela says that they have been able to achieve the same power profile, through changes to the analog front end and other engineering innovations, so that users will not experience any penalty when the new devices come online in the 2nd half of next year.

Related Article:

Quantenna claims 1st chipset for future IEEE802.11ac Gigabit WiFi standard

Texas Instruments (TI) has announced enhancements to their KeyStone DSP architecture to support emerging cloud-based Radio Access Network (RAN) applications for wireless base stations. C-RAN is a development project being led by the China Mobile Research Institute, in which the conventional base station hardware and software resources at individual cell sites are removed and aggregated in a centralized location, connected remotely to the towers over high bandwidth fiber. With pooled resources to serve multiple cell sites, China Mobile proposes that C-RAN can lower network deployment costs with reduced hardware and provide a greener solution that consumes less power. At the 2011 Mobile World Congress, Intel announced that they were collaborating with China Mobile on the C-RAN project, developing a proof of concept for processing TD-LTE signals on the company's Sandy Bridge processors. Intel also highlighted this development at the 2011 Intel Developer Forum.

Tom Flanagan, Director for Technical Strategy in the Wireless Base Station Infrastructure group at TI, says that most of the writings about C-RAN leading to greener base stations have neglected to consider the power that would be consumed by large Intel-based servers. He says that gains in power with C-RAN can't be achieved on Intel class servers, you need dedicated wireless chip sets such as TI's in which hardware accelerators perform the necessary signal processing functions much more efficiently.

TI's KeyStone SoCs for wireless basestations encompass three different configurations to scale from large macro basestations to smaller pico and femto cells.

TI's KeyStone SoC architecture combines the company's fixed and floating-point C66x CorePacs for DSP functions along with ARM cores for control processing, hardware acceleration packs (the green blocks in the illustration above) and I/O. Flanagan says that the architecture enables TI to build three different types of devices that scale solutions from large metro base stations to smaller pico and femto cells. For larger cells, a Transport and Control Processor integrates multiple ARM cores, along with hardware accelerators for network security. One of these devices is typically combined with three Baseband SoCs, integrating DSP CorePacs for L1 and L2 processing, to form a 3-sector macro basestation. The third type of device in the KeyStone family integrates a smaller number of both control and DSP processors for single-sector small cells, i.e. a basestation on a chip.

Flanagan says that even with C-RAN, some processing is still needed at the antenna site in order to send the analog-to-digital converted signals over fibers, that may span a distance of several kilometers. The antenna signals could be formed first into packets or just digitized and transmitted directly to the C-RAN over fiber.The enhancements that TI has made for C-RAN include doubling of the queue manager and descriptor capacity. For software developers, these functions abstract the details of the hardware so an engineer does not need to know when they write their programs, letting them just describe the function they need to use. In C-RAN, Flanagan says, it will take a large pool of chips and the Keystone architecture allows construction of a virtual 800-core device. TI upgraded chip-to-chip communications from 1Gb to 10Gb ethernet, as well as providing a proprietary 100Gb interface over dual links. With the KeyStone architecture, any chip in a cluster can directly write to the processing elements on another device.

According to Flanagan, the number of antennas that KeyStone can support for C-RAN is use-case dependent, varying with the number of users, geographical area, traffic movement for mobile users, and other factors. Basestations are typically constructed with one chip per sector, and the C-RAN configuration of ~800 cores uses 64 chips. Each basestation sector may have 2, 4, or 8 antennas depending on the level of Multiple-Input/Multiple-Output (MIMO) topology that is employed. With KeyStone the software scales across various sized architectures. In order to support 3G/4G migration and backward compatibility, Flanagan claims that TI has a unique capability of supporting simultaneous multimode baseband radio processing, for LTE, WCDMA or even 2G standards in the same chip.

The current markets for C-RAN are primarily China, Korea and Japan, where there are many high-density cities with high bandwidth fiber in place. Regardless of C-RAN usage in such locations, Flanagan says that all networks will be need to be heterogeneous architectures including both macro and small cell basestations. In 4G LTE networks, the need for lower latency of transmission has led to netwrok architectures that push more processing out into the field and closer to the network edge. Since long fiber links may result in too much transport delay, C-RAN deployments may be limited to dense urban environments where distance to the server facility will be relatively short. You can download a copy of TI's whitepaper on "Creating cloud base stations with TI’s KeyStone multicore architecture" at http://www.ti.com/lit/wp/spry183/spry183.pdf

Related articles:

On November 29, SAP hosted a discussion with Geoffrey Moore, the well-known author of "Crossing the Chasm", who has written a new book - "Escape Velocity: Free Your Company's Future from the Pull of the Past." Moore has been a consultant to SAP in the past, so it is understandable that he framed most of his comments at the company's Hillview Speaker Series event on how to apply the book's principles to SAP's business. However, the message which Moore describes in his introduction, that globalization is driving a need to question the working assumptions of successful businesses, is applicable in any industry. The implications for an SAP or Oracle will also flow down to suppliers throughout the Information Technology (IT) ecosystem, from large computer manufacturers to mobile device designers to companies creating the chips and software that run them.

Starting with the "Bigger Picture" for IT, Moore said that in the past enterprise IT was about building the "great systems of record", and in the last decade virtualization has been the lead idea, motivated by reducing cost. In the mean time, greater innovation has taken place in consumer IT, he said, so that now many enterprises are looking at how to transfer those innovations to corporate IT. Moore referred to this trend as the "enterprization" of consumer IT. Companies, and their employees, want an enterprise Twitter, an enterprise YouTube, an enterprise Google search and Facebook, just as they use in their personal lives.

Asked by the moderator how this will play out in Silicon Valley, which she referred to as a "special place" and home to most of these technologies, Moore commented that Silicon Valley is a "perceptually entitled" part of the world. It is a problem, said Moore, that these companies are very consumer focused and U.S. centric, and this causes a disconnect that is painfully hard to bridge. The Googles and Facebooks have set the bar for the user experience, but businesses need enterprise class solutions, which creates enormous opportunity for SAP and Oracle according to Moore.

In the mobile industry, this trend is referred to as BYOD - or Bring Your Own Device. In Moore's view, middle managers were never served by the "systems of record", which were more for the transaction worker who employed static views of the business. Middle managers have a greater need for immediate access to information, not a report, on any device at any time.

Hierarchies of Power

The title of Escape Velocity refers to the problem that large companies get caught up in performance commitments. The end result, Moore says, is that resources tend to get consumed supporting legacy businesses, so that they are not made available for innovation.

To outline a set of solutions, Moore constructed a framework to describe "Hierachies of Power" in business. This model is reminiscent of Harvard professor Michael E. Porter's Five Forces that shape strategy, but is more internally focused.

Moore says that his hierarchy of five powers starts with the broadest, which takes the longest to show results, going down to the most focused and immediate.

Moore's Five Hierarchies of Power:

Category Power - derives from the demand for products and services you produce.
Company Power - how powerful is your company within its segment?
Market Power - which markets do you play in, and how valuable are they compared to others.
Offer Power - how successful are your offers?
Execution Power - how successful are you at executing your offers?

Using the example of SAP, Moore said the company has high power in the Enterprise Resource Planning (ERP) category, but it's not a segment that will see further breakout growth. However, innovation for a company in a mature business can be based on rethinking an existing category, more than entering a completely new category, he said.

Market Power

In describing market power, Moore emphasized the importance of global and vertical alignment. In verticals, Moore said that SAP is well aligned, while offering the interesting remark that Cisco is not. With globalization as a theme of his new book, Moore went on to say that, in this century, the U.S. will no longer be the greatest driver (China). He emphasized that growth will be driven more in the developing economies, especially China.

Moore said that tech companies will inherit business dynamics that require "playing away games". In treating these remarks as if they were a new prescription for business, Moore seemed unaware, or perhaps just skeptical, of the global perspective that is already well in place at most if not all Silicon Valley technology companies. He quipped, attributing the joke to John Cleese (though that is not verified) - "if Britain held a World Series we would invite other countries".

Offer power
Offer power, according to Moore, is the differentiation of what you produce. In mature categories it is not so important to be highly differentiated, he said, because by then customers have sorted it out. However, in dynamic categories the offer rules. Moore gave Apple as an example, saying they achieved a unique feat of delivering new compelling offers in three categories. For SAP, Moore proposed that HANA, a system for real-time in-memory computing, can fundamentally transform the offer that they bring to market.

Execution Power

The key execution power idea is hard for large companies, said Moore. The challenge is, "when you decide to do something different can you take it far enough to get to the tipping point?" He stated that until that is achieved, no other measure of a project's performance matters.. not revenue, nor profit or how many users you have. It is a mistake, he said, that fledgling initiatives get held to a series of metrics before enough adoption has been achieved to get to the tipping point.

Example of successful companies
Asked for examples of companies that were good at execution power, Moore listed these:

Companies that have a charismatic leader, such as Steve Jobs.
Companies that go through a "near death" experience, like IBM under Lou Gerstner.
BMC software, and the transition to cloud computing.
NVIDIA, (saying he had visited the company that morning), for adapting to the need to compete with AMD, who was integrating the GPU into an SoC.

Once again Moore cited Cisco as a bad example, saying that it takes executive leadership, but you need to pare down the number of initiatives you pursue, to get down to just one or two.

Asymmetrical resource allocation

To achieve "Escape Velocity", Moore says that company leadership must be capable of making "asymmetrical bets". The asymmetry arises from what he calls a "lead first, manage second" approach. This can't be done during the budgeting process, and ideally is done one quarter before, he said. Management should be aware that they only have 2-4% of company resources to play with, and unless they are focused you end up getting something like the infamous Yahoo "Peanut Butter Manifesto", i.e. being spread too thin. To back up asymmetrical bets, companies must be willing to have all executive and employee compensation based on success of the initiative.

The larger your company is, according to Moore, the less disruptive the initiative has to be... it just has to be transformative. Competitors will not be able to match the commitment to an asymmetrical bet, he said, but when you "peanut butter" it becomes easy for competitors to follow. An asymmetrical bet is not betting the farm, it's betting next year's seed corn.

In-house innovation vs. M&A

Moore was asked how to resolve the choice between internal R&D and Mergers & Acquisition (M&A) to move into a new category. His advice was that the larger a company is the harder it is to do organic innovation. His recommendation is that organic innovation is a good way to enter a new category, but at a certain point look for an acquisition to "bolt something on" to grow to the tipping point.

How to make the Escape Velocity concepts actionable

Finally, Moore described three possible modes of offer innovation:

Radical differentiation, such as the examples given of Apple and the iPod, iPhone, and iPad.
Neutralization innovation, for which he gave as an example Microsoft.
Productivity innovation.

For Moore, radical differentiation means creating value that is beyond compare. This is to be contrasted with neutralization, where it is more important to be quick enough and just good enough. Moore's example here was how Google responded to Apple's move into the mobile space with Android. At least initially, Android wasn't close to a match for the iPhone, but it was good enough and to market fast enough to neutralize.

The lesson here for incumbent market leaders comes from Nokia, as Moore said, "the one who didn't do anything". The largest manufacturer in the cell phone industry is now "on the bubble" because they didn't neutralize. Geoffrey Moore said that a neutralization strategy is not celebrated enough in Silicon Valley, which values most highly the ability to innovate.

On November 30, Electronic Design Automation (EDA) vendor Magma Design Automation (Nasdaq: LAVA) announced that they had entered into a definitive agreement to be acquired by Synopsys (Nasdaq: SNPS), for $507M. With Magma ranked as the 4th largest company in the EDA industry in terms of revenue, this low valuation stands in dramatic contrast to other large EDA acquisitions over the last 12 to 18 months. At the close of trading just before the deal was announced on Wednesday, Magma had a market capitalization of $391.53M, and a share price of $5.72. Synopsys' offer then represents a premium of just $115M - less than Magma's most recent reported annual revenue of $139.29M at the end of April, 2011.

Contrast the Synopsys-Magma marriage to the most recent large EDA M&A transactions:

(May 13, 2010) Cadence announces plan to purchase Denali Software Inc. for $315M. Estimated annual revenue = $43 M
(June 10, 2010) Synopsys announces plan to acquire Virage Logic for $315M. Virage's previous year annual revenue = $47.4M
(June 30, 2011) ANSYS announces plan to acquire Apache Design Solutions for $310M. Apache reported 2010 revenue of $44M.

With three times the annual revenue of these recent EDA acquisition targets, should Magma have commanded a much higher premium over their smaller industry colleagues? Did Synopsys just steal a competitor, who they had previously sued for patent infringement? (For which Magma paid Synopsys a $12.5M settlement).

Much of the answer to the apparent discrepancy in valuations can be discerned from the chart of Magma's 5-year trailing annual revenue. Comparing financial performance for Magma's most recently closed fiscal year with their 2007 results shows a Compound Annual Growth Rate (CAGR) of approximately -5%.In their 2011 Annual Report, Magma stated:

We have a history of losses, except for fiscal 2003 and fiscal 2004, and we had an accumulated deficit of approximately $387.1 million as of May 1, 2011. If we continue to incur losses, the trading price of our stock may decline.

If Magma were to continue to decline at a rate of 5% per year, their next 5 year annual revenues (in thousands) would look something like this:

2012 projected: $132,321.70
2013 projected: $125,705.62
2014 projected: $119,420.33
2015 projected: $113,449.32
2016 projected: $100,000.00
5-year Total: $590,896.97

In other words, based on this model, Synopsys achieves a positive return on their $507M investment in the 5th year.

Compare this to the most recent acquisition from the $40M+ club, ANSYS' acquisition of Apache for $310M. The Form S-1 which Apache had filed for their aborted public offering shows a 3-year revenue history of:

2008 Annual Revenue: $25,695
2009 Annual Revenue: $34,601
2010 Annual Revenue: $44,047

Prior to their acquisition, Apache had a CAGR of 20%. Projecting that growth forward for a 5-year period, the same ROI period as the Synopsy-Magma acquisition, we see:

2011 projected: $52,856
2012 projected: $63,428
2013 projected: $76,113
2014 projected: $91,336
2015 projected: $109,603
5-year Total: $393,336

So, just like Synopsys, ANSYS would also achieve a positive return on their investment in the 5th year. While these models are somewhat simplistic, they nevertheless serve the purpose of demonstrating how a company like Magma, at more than three times the size (in revenue) of Apache, could agree to be acquired at a price that it only 63.5% higher. With a deficit of $387M, and prospects for further share price declines, LAVA stockholders should be thrilled to cut their losses in this deal.

Related articles:

Home

		Fiscal 2012
Named Executive Officer		Annual Base Salary($)
Rajeev Madhavan		$510,000
Peter S. Teshima		$320,000
Roy E. Jewell		$510,000

Subscribe to the EE Daily newsletter

About Us

Recent Tweets

Article Archive

Archived Publications