Archive for the ‘Operating Systems’ Category

Beware of Chip Companies Bearing False Gifts

Back in the mid 1960’s American Airlines ran a very successful ad program (the ad was “Take me along if you Love me”) in which business travelers were allowed to bring their wives along on their business trip – at no additional cost.

Thousands of business travelers took advantage of the offer and sales soared. Leave it to some marketing moron to contact the traveling “wives” and ask them how they enjoyed their trip. Seems that many of the “take me alongers” were not the wives but someone else.

Hundreds of divorces ensued and the airline was sued by many and sales took a turn for the worse.

Fast forward to 2011 and it seems that the grandsons/granddaughters of those hapless AA marketing mavens might now be working for Freescale.

Freescale is giving away free the MQX operating system – but is Freescale helping or handicapping their customers?

In a recent EMF survey of 660 embedded developers, we were able to compare design outcomes among all of the major operating systems (using our unique Dashboard tool – see video).

Citing just a few highlights,

  ThreadX Micrium VxWorks MQX
         
Time from design start to shipment (months) 10.5 11.3 16.3 15.8
         
Percent of Designs completed Behind Schedule 29.8% 40.0% 51.1% 56.6%

 

So we are left wondering whether Freescale realizes what they are doing to their customers – and whether the grandsons/granddaughters of American Airlines marketers can find a place where their efforts can’t do any more harm. The Obama administration comes to mind.

Interpreting Embedded Survey Outcomes – Different Surveys Report Different Results

Surveys of embedded developers to establish market parameters regarding RTOS use must be careful to avoid unintentional bias. 

I have recently been contacted by senior executives from RTOS companies mentioning UBM’s survey of embedded developers that placed FreeRTOS as the most used RTOS for embedded use in 2011 (14% of respondents). FreeRTOS placed 3rd in the 2010 UBM survey. Year-over-year EMF data is at substantial odds with these findings.

EMF takes no position on how UBM conducts their survey or on their results. EMF data for the past two years have shown that FreeRTOS usage has garnered less than 1% of total developer responses. This encompasses over 1200 responding developers. This is a significantly lower response than the 14% reported by UBM. EMF has no idea of how this discrepancy came about. We stand by our data as our surveys and responses are restricted and carefully monitored. We make no judgment regarding UBM’s methods or results. We are responding to requests to report our findings.

In EMF’s 2011 Annual Survey of Embedded Developers, eleven thousand embedded developers were statistically selected and sent invitations to participate in the 2011 survey. Six hundred and fifty three developers responded to our invitation. PIN numbers were assigned to each request so that we could insure that only those invited participated in the survey – and that they could respond only once.

In 2011 developers reported using an in-house RTOS (20.1%), Android (19.3%), XPE (16.5%) and CE (15.9%). FreeRTOS was used by 0.9% of respondents. From our perspective, the suggestion that FreeRTOS use would exceed that of in-house, Android, XPE, CE, or VxWorks use is beyond any reasonable reality check.

Selecting an Embedded RTOS: Factors of Importance to Developers 2009-2010

Every year, for the past 12 years, EMF has asked embedded developers to respond to a comprehensive and detailed survey that explores all aspects of their design and development activities in a manner that permits EMF to correlate answers to any questions – or series of questions – with any other question or series of questions. The survey is constructed and conducted in a statistically accurate manner to insure valid interpretations (including the ability to create comparative ROIs between similar product offerings).

 

In 2010, 536 developers responded to the survey resulting in a statistical confidence level of 95% +/- 4.5%. This is extremely important to EMF as many of our subscribers are basing their competitive position by needing to  forecast what chip and OS usages will be two years from now.

 

We can, of course, examine these responses from the perspective of any vendor’s individual operating system (which many of our subscribers do to look at their users as well as their competitor’s users).

 

Year-over-year we ask developers to select from among many alternative responses (we limit each developer to a maximum of 4 responses) to the question regarding which factors are most important to their decision to select an operating system. We also ask what factors would enter into their decision to purchase from either a single vendor or from multiple vendors. We may publish this data in another post.

 

The following table presents comparative responses to the factors regarding the OS selection process for the periods 2009 and 2010. The top 15 responses (out of 33 possible) are presented.

 

 

2010

2009

 

Industry

Industry

Acquisition cost

44.6%

36.7%

Availability of source code

33.1%

26.4%

Microprocessor support

30.4%

22.9%

Real time performance

29.8%

34.8%

Compatibility with our development tools

27.6%

29.0%

Includes good development tools

26.6%

24.5%

Reliability

25.2%

31.1%

Compatible with Linux

24.1%

17.8%

Availability of perpetual license

22.7%

20.6%

Availability of req. middleware or networking protocols

18.1%

14.3%

Availability of professional services

16.7%

12.1%

Host platform support

15.0%

15.0%

Quality of support

14.4%

19.2%

Royalty cost (production licenses)

14.2%

9.8%

Must be open source

13.4%

10.7%

 

The lists are comparable between 2009 and 2010 with “realtime performance” and “reliability” taking the steepest drop (but remaining important nonetheless), and “Linux” compatibility “microprocessor support” and “source code availability” being the largest gainers.

 

This bodes well for the Linux community. Such items as “safety certifiable (DO178B, etc.)”, “visualization” and ““security certification (NSA, Common Criteria)” received middle single digit response levels.

 

We have seen this result for many years – and, due to the proliferation of communications and consumer devices, this is why we have forecast the growth in ThreadX, Micrium, MontaVista Linux and Nucleus use. These OSes have been deployed in hundreds of millions of devices worldwide. Unless the application calls for a MILS level certification, alternative OSes are being used in place of the high power OSes of the past.

Criteria which developers use to select an Embedded Operating System

Insights from the 2010 EMF Survey of Embedded Developers

  

Each year EMF conducts a thorough and detailed survey of embedded developers. Using the EMF Dashboard – a web based tool that permits vendors and developers to correlate information between any responses to any question, EMF presents selected insights from its analysis of the 2010 data. The Dashboard enables vendors to look at the responses of their customers (and potential customers) as well as to their competitor’s customers. This provides invaluable insights for strategic and sales planning. Developers can see what their fellow developers are considering in their selection criteria

 

Product opportunity windows are fleeting and time-to-market issues dominate design considerations. The two factors that invariably make the difference between success and failure are first, the knowledge that comes with an accurate insight into the internal and external forces which drive product markets and second, an insight into the concerns, desires and thought processes of those customers who make the decision to purchase a particular product or deal with a particular vendor.

 

The following data, taken from the 2010 EMF Embedded Developer Survey, addresses many issues of which embedded vendors and developers should be aware. The data presented here represents embedded industry averages across many application verticals, many vendors, many OSes, etc. Subscribers to the 2010 EMF Market Intelligence Program have access to their custom Executive Dashboard with which they can create unlimited cross tabs to further examine the data presented here.

 

 Criteria Most and Least Important to Developers in Selecting an OS

 

Developers were asked to indicate which criteria were most important to their decision in selecting an OS. The top (most important) responses and the bottom (least important) responses are presented in the following tables.

 

There was an interesting reordering of priorities in 2010 with cost remaining the principal factor. Realtime performance fell (no surprise here) and safety certifiable and virtualization remained as a small part of the collective consciousness of embedded developers.

 

 

Criteria most important for selecting an OS:         Top 8 Responses

2010

2009

 

 

 

Acquisition cost

44.6%

36.7%

Availability of source code

33.1%

26.4%

Microprocessor support

30.4%

22.9%

Real time performance

29.8%

34.8%

Compatibility with our development tools

27.6%

29.0%

Includes good development tools

26.6%

24.5%

Reliability

25.2%

31.1%

Compatible with Linux

24.1%

17.8%

 

 

 Criteria Least important for selecting an OS

2010

2009

     
Supports virtualization

4.9%

5.1%

Preferred vendor or on company approved list

4.3%

9.1%

POSIX or SCA compliant

4.3%

4.9%

Provides memory protection

3.9%

4.0%

Security certification (such as Common Criteria or NSA)

2.6%

4.2%

Must not be based on GPL

2.6%

2.1%

ARINC 653 compliant

1.6%

0.7%

Subscription licensing available (annual or fixed term)

1.4%

0.9%

Response to RFP

1.4%

2.8%

Forecast 2010: What Is in Store for Embedded Developers

Taking a “dog’s-eye” view of what we might expect in 2010

Light at end of tunnel

 

 

 The year 2010 is just around the corner, and we are doing what we do best — forecasting. After all it’s our name. But we aren’t just guessing — we base our forecasts on historical facts and data. For the past 12 years, we have been tracking what developers are doing, what tools, OSes and processes they are utilizing and what their design experiences have been. We also report on what issues trouble them the most.

Now we are preparing our 2010 detailed and comprehensive EMF Executive Survey of Embedded Developers and Managers. We will be inviting you to take the survey to see how you our “loyal readers” compare with the larger statistically based responses (please contact us at surveys@embeddedforecast.com if you are willing to participate). Respondents who take the survey will receive a complimentary copy of our survey overview (a $399 value).

Read the rest of this entry »

Is Cavium’s Acquisition of MontaVista Good or Bad for Commercial Linux?

Question 1

 

 

On November 10, 2009, Cavium, a publicly traded provider of highly integrated semiconductor processors that enable intelligent networking, communications, storage and security applications, signed a definitive agreement to acquire MontaVista. Cavium stated that MontaVista would continue to operate separately and their customers would not be restricted to using only Cavium processors.

Immediately, questions arose: 

  • Who is Cavium and why did they make this acquisition?
  • Was Jim Ready clipping Burger King Coupons?
  • Was MontaVista motivated by Intel’s acquisition of Wind River?
  • Does this mean that commercial Linux is facing financial do-do?
  • What is really going on behind the scenes and is this a good or poor marriage? 

Read the rest of this entry »