#include

Are we moving to an all-ARM world?

Rich Nass

10/16/2008 1:29 PM EDT

There are many microcontroller architectures available to an embedded systems developer. But as more and more vendors adopt the ARM core, the choices become limited. Of course, those licensees will tell you that every one of their offerings is different. And in many ways they are. But in many ways they aren't.

In an interesting twist this week, one of the ARM competitors--Microchip--decided to hop on the ARM bandwagon, sort of. In a roundabout way, Microchip is attempting to become an ARM licensee. They are doing this by attempting to acquire the microprocessor divisions of Atmel.

Rather than go through the grueling months or even years of development, Microchip is trying to buy its way into the ARM camp. My first instinct (and the one I'm sticking with), is that this is a great move for Microchip, and one that could make them a very powerful player. By adding the ARM processors to its portfolio, which already contains the MIPS-based products, Microchip can span the performance range.

Whether it's reality or not, the perception of the MIPS products is that they play at the high end. This left a hole in the Microchip family. Its products start at the really low-end at 8 bits and traverse through the mid-range at 16 bits. But when it came to the 32-bit products, they skipped the high volume segment in favor of the high performance. Adding ARM changes all that.

I spent the last few days at the Renesas Developer's Conference, and I broached this subject with a few of the company's executives. They took the high road, explaining that they can beat the ARM competitors on performance, thanks to a much higher flash-memory performance. And if a customer wanted it, they could embed 8 Mbytes of flash on-chip.

I still make the argument that it's the supplier with the best ecosystem, which in turn gets their customers to make the fastest, is that one that will win.





DrOctavius

10/16/2008 11:36 PM EDT

The MIPS idea didn't work, now Microchip need the plan B: "If you can't fight them, JOIN them", but they a fast move because they're losing market!

I agree 100% of the ecosystem importance. Now you can get everything you need in a MCU that cost less than $10. Development time is more and more important than if I pay $1 more or less the HW.

Sign in to Reply



daraius

10/23/2008 11:24 AM EDT

2ndGenEmbedded, you are a little out of touch. Do you have any idea how many 8051 derivative parts are out there, and running all the way to 100MIPS? SiliconLabs, ADI, TI etc etc etc. Go to Keil.com and see the list. A huge number of new 802.15.4 products will have embedded 8051 cores. The microchip PIC parts went outdated years ago and they only recently launched what I consider decent micros with dsPIC, and PIC32. A SiLabs 25MHz 8051 runs circles around even the best PIC18's.
Yes, they cultivated a very loyal user base like yourself.
I think them picking up an ARM line is a good idea. They are a good company with good pricing and availability.

Sign in to Reply



Haldor

10/23/2008 11:32 AM EDT

To 2ndGenEmbedded:

The 8051 is a very successful chip that is used in many new designs. Why? Because you get get the exact peripheral and memory mix in the package size you want at a great price. The 8051 also has a fantastic tool chain going for it with Keil C51.

IC vendors with 8051 based processors:

Analog Devices
Atmel
Infineon
Intel (well they used to)
NXP (Philips)
ST Micro
Texas Instruments
Dalla Semi (Maxim)
Hynix (LG)
OKI
Silicon Laboratories
WinBond

I personally think the Arm7 and Cortex processors are the future replacement for the 8051. I can get a 128KB ARM7 processor from NXP with USB, Ethernet and CAN for under $5. If NXP drops the ball I can find a comparable ARM7 based part from one of their competitors for not much more and software porting will mostly consist of rewriting device drivers and recompiling. Compare that with the effort required to port a complicated application from say a PIC to an HC08 for example.

Another big issue is that single vendor processors limit you to only one companies IP. If I need a really high performance AD converter I can look to Analog Devices or TI or Silicon Labs. If I need a tiny (3mmx3mm) processor with 25 MIPS and no external components required (except a decoupling cap) then I can look to Silicon Labs. If I need a sub $1 processor with full feature set and 64KB Flash I can look to Hynix or one of the other Chinese vendors (there are several I didn't list because most people wouldn't recommend them).

There is a huge amount of free or inexpensive IP available for the 8051. Tools are incredibly cheap now. How about a USB interfaced ICE with trace and hardware breakpoints for $40?

Obviously I am an 8051 user. I have also developed with Z80, 6502, 68XX, 680XX, ColdFire, X86, NEX KOS, ARM7 processors. They all have strengths and weaknesses. For me the big issue is portability of designs. I have had to port existing products to new processors more than once and single vendor ICs make this much harder.

Sign in to Reply



bgat

10/27/2008 9:52 PM EDT

Microchip's pursuit of Atmel continues the schizophrenic trend of the PIC product line as a whole. 'Nuff said.

Sign in to Reply



Please sign in to post comment

Navigate to related information

Datasheets.com Parts Search

185 million searchable parts
(please enter a part number or hit search to begin)
Jobs sponsored by

Feedback Form