Embedded Industry Statistics (2013 Market Study)

This is a review of the 2013 Embedded Market Survey undertaken by UBM Tech Electronics

infographic

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

The latest report throws up a whole host of interesting embedded computing system & industry statistics, we must urge some caution as respondents are 80% USA based, which of course is a different, though not dissimilar market to which we see in Europe.

We see that a higher percentage of new embedded projects are actually upgrades to legacy equipment, this bodes well for product ranges that retain legacy functionality (i.e. our insistence at still providing full ISA bus and support for operating systems such as DOS).

For projects utilising wireless technology, over 50% utilise WiFi, interesting given the difficulty of configuring this (in terms of entering SSID passwords) in a headless system. This report also tells us that headless operation is evident on over half of all embedded deployments.

Software effort accounts for over 60% of the typical project cost (vs hardware), this is to a degree to be expected, though could this also account for the simplification of hardware design based on System-On-Module type designs, as opposed to ‘from ground up’?

A typical custom project spans less than 12 months (70%), though nearly half (46%) of projects are delivered late. This is a particularly worrying statistic as late means slower time to market, which means lost profits and decreased market penetration when the product is eventually released. This figure surprised me, DSL keep a tight lid on project time-scales and project post mortems show 95% are delivered on time, the remainder due to customer invoked change mid-project, rather than any engineering caused delay.

 Want it on time? Talk to us about your next bespoke electronic design project.

Programming in C is still true of 60% of projects, the fact that code is re-hashed from previous projects in 79% of cases would add weight to this. This suggests to me that not only has hardware performance (driven by OS) not increased significantly, where hardware has had to change one would perceive this is due to obsolescence rather than a driver for increased performance/complexity.

The developers key want for improvement area is in debugging tools, we’ve encountered this recently with the main objection to using DOS being the lack of modern debugging tools (DSL are currently investigating this). Interestingly real time capability is the most important aspect when choosing an embedded Operating System similarly availability of full source code, this consequently excludes any Microsoft OS – actually only 15% of correspondents agreed they’d be using a Microsoft OS in the next 12 months at all!

A high percentage intend to utilise Linux with their next development, citing low cost as the main driver for this (interestingly DSL wrote an article on the often false lure of ‘low cost’ Linux many years ago, though this may now be less true with improved development platforms)

An equal number of those surveyed considered the CPU support (development) tools were as important as the CPU’s capabilities itself, this isn’t surprising, projects get delayed by ‘uncooperative’ hardware, especially that with poor documentation.

A mere 5% of respondents considered world CPU leader Intel to have the superior support package, rating TI, Freescale, Microchip & ARM ahead of them (all RISC based).

What scared DSL is DMP, the manufacturer of the ARM competing low power x86 CPU (DX/DX2) wasn’t recognised at all (again though this is 80% USA). This suggests a lack of marketing and a need to raise awareness of the product – after all, one can have the best product in the world, but if no one knows about it…

Finally, we saw FPGA usage has decreased 3-4% year on year for the last 5 years, the reason for decreasing utilisation seems to be cited as lack of experience within the design team. Given a dedicated IC is usually a far better method of data processing than relying on the CPU (with requiring a higher performance CPU introducing a host of power consumption, heat generation & cost issues), I see in the future a wider requirement for outsourcing FPGA design, a service DSL offers and have seen a substantial increase in, in recent years.

DSL regularly reviews such surveys to ensure we keep on top of market trends and can offer future proof advice to customers designing new products, especially utilising new technologies.

Always feel free to discuss with any DSL Engineer any questions or concerns you have – contact us now!