What do you see as the main factors behind the growing trend for 32-bit MCU designs in markets once dominated by 16- and even 8-bit chips?
I think three factors combine to create the explosive growth of 32-bit MCUs (and the decline of 8-bit and 16-bit MCUs):
- Increasing software content in embedded applications
- Availability of cost-effective 32-bit MCUs
- Increasing performance and connectivity requirements in embedded applications
When it comes to choosing an MCU for an embedded application, it’s all about the software. Industry analysts VDC found that by 2004, 48% of total production cost of embedded applications was attributed to software development costs. With today’s 32-bit MCUs, software can be efficiently written in C, helping to contain software development cost while the software content continues to grow.
Semiconductor process technology and MCU architecture have now combined to make 32-bit MCUs as cost-effective as their 8-bit and 16-bit cousins – and significantly more capable. With the ARM Cortex-M3 core, the MCU market has the first open standard truly deterministic microcontroller core, making 32-bit MCUs built on Cortex-M3 fully capable of replacing legacy 8-bit and 16-bit designs.
Finally, as the worldwide consumption of power increases exponentially, governments and markets are being driven more by a need to increase the efficiency and reduce the power footprint of applications. Fundamentally, this increase in efficiency and decrease in power is facilitated by increasing MCU performance (to be able to run the computationally complex algorithms required to increase efficiency, especially in motion applications) and by increasing connectivity (so that embedded applications can communicate to increase efficiency and decrease power of the entire system, for example, in intelligent power metering systems with intelligent consumer appliances in the home).
Spelling out GNU and Linux stories
Your MCU’s are based on the ARM Cortex-M3 processor core, how do you see this core evolving and are there any changes you would like to see implemented?
ARM did a fantastic job of architecting and designing the Cortex-M3 core for microcontroller applications. It’s small, fast, fully deterministic, efficient in on-chip memory utilization, and possesses excellent mathematical computational capability. In fact, we frequently refer to our Stellaris MCUs as “an easy-to-program DSP with truly portable code”. The creation of the Cortex-M3 core was a foundational pillar to the creation of Luminary Micro – if our only choices were to create the company around an ARM7 or ARM9 or MIPS core, Luminary Micro would not exist.
We continue to work closely with ARM on extensions of the Cortex-M3 core and extensions of capabilities for the microcontroller market. I can’t significantly tip our hand on this, but it would be safe to expect further innovations to foster new market developments, at least one of which will be similar in scope to the conventional microcontroller market transformation currently underway.
Express Logic’s royalty-free RTOS has been ported to Stellaris, so do you see the move to open-source and even Linux platforms having an impact on your markets?
Since we understand that “it’s all about the software”, we believe that open source is very important and we make significant use of open source to enable customers to get to market quicker. We’ve been using the open source FreeRTOS.org as our RTOS for internal development purposes since the development of our first Stellaris MCUs in late 2005. We use the open source uIP and lwIP Ethernet stacks with our Stellaris MCUs with fully integrated 10/100 Ethernet MAC+PHY. The example web servers that ship with our Ethernet-based evaluation kits show a no-RTOS web server implementation using lwIP, and an RTOS version using FreeRTOS.org and uIP.
We’ve also used the open source PTPd implementation to show how to use our hardware assist for IEEE 1588 Precision Time Protocol applications, and the open source MatrixSSL from PeerSec Networks to show how to add SSL to Stellaris-based Ethernet implementations. Luminary Micro also provides development platforms and support to the OpenOCD community.
Traditional Linux implementations are a bit less likely to impact our markets, as they typically have a large enough flash and SRAM footprint to require embedded computing platforms rather than microcontroller platforms, where all or at least most of the memory is on the chip. In addition, microcontroller applications are more likely to have real-time requirements, which Linux does not handle well. However, some smaller footprint Linux variants are in use in our customer base today.
Do multi-core processor architectures have any relevance the microcontroller market?
Multi-core has become more prevalent in computing/embedded computing markets (PCs and iPhones, for example) because of scaling issues reaching to 1GHz and beyond, which do not apply to the microcontroller market.
The challenges of multi-core implementations traditionally centre around programmability. It’s certainly not hard for semiconductor companies to create multi-core implementations – especially a non-SMP implementation such as you might see in a microcontroller version – but programming and debugging multi-core is quite difficult. To the extent that an embedded control application is easily fragmented for separate-core control functions, and that the resultant multi-core MCU implementation can retain determinism, I think that multi-core MCUs can fill a very useful niche.
What are the challenges and benefits of building a company on a single processor architecture?
The benefits of building a company on a single processor architecture far outweigh the challenges, at least when the single architecture is the ARM Cortex architecture. Being able to provide an unprecedented price/performance range (from 8-bit equivalence to 1GHz) of instruction-set-compatible implementations in the ARM Cortex line among multiple vendors working in an open architecture community is exceedingly powerful to the embedded developer community. It never hurts to be the provider of the solution that the embedded developer base wants!
For us, being focused on Cortex-M3 MCU implementations means that there’s absolute clarity in our mission and in our product strategy. Unlike some of our competitors with legacy proprietary solutions, we are never confused about the solutions we can bring to bear to help a customer get to market quickly – and our customers do not have to worry about which architectural line will be the next to be discontinued.
See also: Q5 – Interviews with electronics industry leaders
Read all the Electronics Weekly Q5 interviews. From ARM’s chairman, Sir Robin Saxby, to touchscreen technology firm Zytronic’s MD, Mark Cambridge, the business leaders share their particular insights on the UK electronics industry.