Atmel in ARM-land

Last post
25 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

A few opinions, please...

I've been working with ARM7 and now Cortex for several years (embedded systems, not consumer items).

I may be mistaken/naive but it seems that this ARM/embedded market is dominated (market share) by NXP and ST Micro. Again, not talking about ARMs with MMUs.

Would you go down the Atmel road with low-end ARM/Cortex? Are there facts that bolster the prudence of Atmel in (low end) ARM-land?

Atmel Studio 6 seems focused on Cortex/M3 and that's a positive sign - if it matures and proves to be a viable alternative to IAR and Keil (in the professional tools space).

some chatter...
http://www.digitaljournal.com/pr/601116

http://www.atmel.com/microsite/sam3ax/

http://dev.frozeneskimo.com/notes/getting_started_with_cortex_m3_cmsis_and_gnu_tools#why_i_like_cortex_m3

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Many years ago I went to an Atmel seminar introducing their low-cost ARM7TDMI device. We got a nice little evaluation board and a Wiggler (both free). I think they were the first company with such hardware. However, they seem to have lost their way subsequently, and are now, belatedly, trying to catch up. They appear to have abandoned the low-end Cortex-M0, perhaps because it competes with the 8-bit AVR.

Leon Heller
G1HSM

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Let me put on my Owner Of One Man Company Hat.... "Hmmm... need an ARM compiler for this Company X project... lets see... this one is $3000, this one is free. Gotta think... $3000? $0?"

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

To expand on Uncle Bob's post. If I was to get into ARM, I would look at what Atmel has to offer simply because I would not have to learn an entirely new toolset. THe price is a home run as well. I will note that given the current issues with Studio5, I have not looked at the concensus of 5.1 is, I would wait to see how things work out with the toolset before jumping in.

I do agree with Bob, $3k as opposed to $0. At the same time though for $3k, one would expect things to work perfectly right out of the gate.

I would rather attempt something great and fail, than to attempt nothing and succeed. - Fortune cookie

 

Don't hang your hat on expectations because this often leads to disappointment.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

stevech wrote:
in the professional tools space
What space is that? Why does not GCC (avr*-gcc) belong there? Is it because it is free?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

not sure how yet this "Atmel Software Framework" works with this "For Atmel's ARM processor-based microcontrollers, the library provides full support for the Cortex Microcontroller Software Interface Standard (CMSIS)."..need to research a bit more on that to get a clear picture.

they also say that AvrStudio 6 supports SAM-ICE which if I understand correctly is an OEM version (subset)of the full version Segger ARM Jtag debugger. So it would be interesting to know if AvrStudio 6 will support Segger ARM Jtag tools and the Atmel Cortex versions...seems like they could/should at some point.

along with Atmel , Silabs also has joined the ARM Cortex train recently

http://www.silabs.com/products/mcu/Pages/32-bit-microcontrollers.aspx

Silabs indicates their software is free (eclipse based like NXP,Code Red,Atolic,TI.etc)...so if cross platform dev is important, there are lots of choices unlike AvrStudio6. Kind of like all the eclipse based vendors though as their IDE's are somewhat the same, so you can have more mfg choices without being locked in too much if at all.

and there are many choice's between $3k and $0 and a lot of them are pretty stable dev tools as they have been released for some time.

what Silabs may bring to the party is a good backgound in Analog and rf capabilities..that is probably where their Cortex parts will ultimately shine especially if they embed some of their other technology.

chips with little or no errata, available from distributor stock in qty, and the stability and reliability of the sofwtare tools/envirionment is the criteria here for almost any micro to be considered...that and is the mfg a reliable vendor moving forward and also in the recent past.

it will be interesting to see if other software only vendors supports ATmel Cortex going forward.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Cortex M0; cheaper than a cup of shitty coffee, 50 MIPs, 32 bit, CMSISS, multiple vendors offering similar silicon.

What bob says about pricey development tools is true, but it is a one time investment and the benefits of M0 over AVR is very compelling.

I had a glance at the SAM3 stuff, it looks exactly like the M3 stuff that other vendors offer. No big advantage no big disadvantage. With all things being equal, I would pick a company that has it's own fabs or is bigger so it has a bigger axe to swing when competing for space from 3rd party fabs.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Development tools... for my job, the benefits of tools in the class of IAR + Segger JTAG/Flash breakpoints - makes the price insignificant, given one (like me) works for several years in ARM-land. And the IAR tool license applies to all vendors' ARMs. People gripe about the cost, but the difference in productivity with IAR is night and day versus GCC.

Indeed, GCC is great for students/hobby and some one of a kind product development.

BUT, the topic's theme is is Atmel in ARM-land, not a debate on tools.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ezharkov wrote:
stevech wrote:
in the professional tools space
What space is that? Why does not GCC (avr*-gcc) belong there? Is it because it is free?

IMO AVR studio and AVR-GCC is every bit as professional as stuff from keil and IAR. Once you get into more exotic stuff like RTOS the non free stuff starts to flex it's muscle.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

stevech wrote:
BUT, the topic's theme is is Atmel in ARM-land, not a debate on tools.
I understand. Sorry for dragging the thread in a different direction. But in my defense, I thought you were asking for that by the phrasing that you used. I have not done any ARM. I cannot comment on that. But I have been doing AVR "professionally" for several years already. Therefore, the cost of IAR would not be an issue. But I just do not see what "night and day" advantage IAR would give. I have to say, I have not tried IAR for AVR (or anything else, for that matter). On the other hand, I have been doing software for quite some time. Therefore, I just do not believe that there is something out there that will just revolutionize my productivity.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The Value Added by Big Companies is libraries and appnotes targeted to the Big Company products. Rabbit for example. Make a lot of modules with a microprocessor, a ram chip or two, a couple of flash roms, an ethernet chip, and they have libraries for everything... complete tcpip stack, just include the h file and call the connect function. Same with Good Old MicroChip. Great appnotes with tutorial info. Atmel is Catching Up... they have good appnotes, and now they have a complete suite of evaluation boards with every Atmel cpu, and the AtmelSoftwareFramework looks like a complete point and click library for every peripheral on every Atmel cpu. No more writing a uart putchar. Thats last millenium. Link configuration manged object files written by professional software engineers. Guaranteed success.

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
BUT, the topic's theme is is Atmel in ARM-land, not a debate on tools.

Granted.. but Atmel marketing is saying that one of the big advantages to using their ARMS is the fact that their IDE Avrstudio 6 is free and supports it and other ATmel chips, one ide to learn. So if you wonder how relevant Atmel's ARMs will be in the ARM Cortex market then it might be based on who and how many think $0 for dev tools is important AND if they like Avrstudio 6. Atmel certainly seems to imply they think it will be an important consideration in the adoption of their Cortex ARM.
Atmel in that marketing interview video tends to suggest that Avrstudio 6 with ARM support IS a strategic marketing tool in Atmel's mind. For those users that have not already abandoned AVR's for other ARM Cortex vendors, Atmel is probably counting on those still using Avrstudio for AVR to transition to their ARM at no cost and thus not look at other vendors. As that IMO creepy Atmel marketing dude in the video said when talking about Studio5/6.."it will be negative experience for someone to switch to another micro" (and IDE).
That was not the case in our lab experience, other eclipse vendors tools were actually quite easy to adopt and were of very high quality...experience so far with studio 5/6 has not been as positive and issue free..so his marketing bs did not measure up here...but in fairness if Atmel can improve Studio it may capture some interest they might not have otherwise.
Watch that video and listen to his words...see how much actually does and does not add up in your experience/opinion.
http://video.eetimes.com/video/atmel-avr-studio-5-interview-with-haakon-skar/820373018001

Software tools importance? Not for your lab and certainly not our's, but maybe someone's....

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

At the end of the day it's surely about cost and features isn't it? An ARM is an ARM is an ARM - pick the one that has all the features you need and costs the least. I'm not sure if that makes SAM3 competitive or not. The Atmel M3 tend to be at the upper end of the M3 scale so are not competing with those lowest end STMs and NXPs. Presumably Atmel have some plans for M0's - it'll be interesting to see where they stack up against the competitors and whether they might offer features (such as 5V operation) not offered by the others. OTOH M0's will be in direct competition to megas and especially Xmegas so maybe Atmel see their Xmega as an "M0 beater"? (not sure where UC3 is supposed to fit into the big picture though - I guess they are more in the M3 realm or M4?). Talking of M4 - the AS6 announcement mentions M4 - exactly how does that fit alongside UC3?

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
An ARM is an ARM is an ARM - pick the one that has all the features you need and costs the least.
It makes it more obvious that the peripherals provided on a microcontroller are as important (or more important) than the cpu core. A CM3 with the familiar AVR peripherals would presumably be more attractive to an existing AVR user than having to switch to another vendor's chips that does things differently. If you were programming in C, you might not even notice that the core had changed. (Though I haven't kept track of whether Atmel has done this.)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
A CM3 with the familiar AVR peripherals would presumably be more attractive to an existing AVR user than having to switch to another vendor's chips that does things differently

that would be nice but do not see how that would be competitive or even possible for AVR8 or Xmega..(not sure about AVR32)
generally speaking for example the SPI peripheal on AVR8 is pretty basic and sometimes lacking in some apps.

edited..clawson in his later post explained for example the spi peripheal for the AVR8,Xmega,UC3,Mx part differences a lot better.

that's why this Atmel note is a bit fuzzy to me without further research
"Atmel Software Framework" works with this "For Atmel's ARM processor-based microcontrollers, the library provides full support for the Cortex Microcontroller Software Interface Standard (CMSIS)."

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

westfw wrote:
Quote:
An ARM is an ARM is an ARM - pick the one that has all the features you need and costs the least.
It makes it more obvious that the peripherals provided on a microcontroller are as important (or more important) than the cpu core. A CM3 with the familiar AVR peripherals would presumably be more attractive to an existing AVR user than having to switch to another vendor's chips that does things differently. If you were programming in C, you might not even notice that the core had changed. (Though I haven't kept track of whether Atmel has done this.)

Generally across the board for any uC, peripherals are roughly the same in topology and function. I guess if they keep the naming conventions of registers similar then that might qualify for "familiar".

To me, if Atmel kept the M3 datasheet similar in structure and format as an AVR datasheet, that is all the familiarity I could hope for.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
To me, if Atmel kept the M3 datasheet similar in structure and format as an AVR datasheet, that is all the familiarity I could hope for.

I'd agree with that for an M0, but not for the higher end chips. Not at all.

A new topic-twist: In this embedded-ARM market (not ARMs with MMUs), T.I.'s adventure in going into ARM is arguably a comedy. Anyone agree?

This is all curious - and one harkens back to the days of "Who has the best 8051-alike", since that was licensed out too, as are ARM cores' intellectual property.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
A new topic-twist: In this embedded-ARM market (not ARMs with MMUs), T.I.'s adventure in going into ARM is arguably a comedy. Anyone agree?

if you are referring to the Luminary stuff..agree..like the features but the Luminary stuff has way to much errata for this lab.
do expect some great things from Silabs ARM though..they are very focused and execute as a company very well...and they made a pretty hot "51! :wink:

tough to imagine anyone new coming to the ARM Cortex party without bringing some features and technology to make them standout in the crowd.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

stevech wrote:
In this embedded-ARM market (not ARMs with MMUs), T.I.'s adventure in going into ARM is arguably a comedy. Anyone agree?
Using TI’s Code Composer Studio IDE for free
bluegoo wrote:
... but the Luminary stuff has way to much errata for this lab.
With the addition of TI's fabs to Luminary's products, are the errata getting fixed fast enough?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

So if Atmel made their M3 SPI like AVR8 they would be at a competitive disadvantage. The xmega SPI might be more like a NXP/ST SPI but last time I looked it also is lacking (but better than AVR8).

It's not a question of what they will do. The SAM3 chips have had datasheets available for a year or two and became generally available a few months ago. For their SPI (for example) the SAM3N00A datasheet (a 16K small one) shows these registers:

SPI_CR (control) (4 of 32 bits used)
SPI_MR (mode) (18 of 32 bits used)
SPI_DR (Rx data) (20 bits used - 16 for data)
SPI_TDR (Tx data) (21 bits of 32 used - 16 data)
SPI_SR (status) (12 of 32 used)
SPI_IER (int enable) (11 of 32 bits used)
SPI_IDR (int disable) (11 of 32 bits used)
SPI_IMR (int mask) (11 of 32 bits used)
SPI_CSRn (chip select, 4 of these) (all 32 bits used)
SPI_WPMR (write prot mode) (25 of 32 bits used)
SPI_WPSR (write prot status) (11 bits used)

Compare that to Xmega that for an SPI has:

CTRL (control) (8 of 8 bits used)
INTCNTRL (int control) (2 of 8 bits)
STATUS (err status) (2 of 8 bits)
DATA (err data) (8 of 8 bits used)

Or a mega:

SPCR (control) (8 of 8 bits used)
SPSR (status) (3 of 8 bits used)
SPDR (data) (8 of 8 bits used)

For completeness this is what a 16K UC3 offers:

CR (control) (5 of 32)
MR (mode) (18 of 32)
RDR (Rx data) (16 of 32 bits)
TDR (Tx Data) (25 of 32 bits)
SR (status) (8 of 32 bits)
IER (int enable) (7 of 32 bits)
IDR (int disable) (7 of 32 bits)
IMR (int mask) (7 of 32)
CSRn (4 of these, chip select) (32 of 32)
WPCR (write prot control) (25 of 32)
WPSR (write prot status) (11 of 32)
FEATURES (features) (21 of 32)
VERSION (version) (16 of 32)

So the SAM3 and UC3 are at least an order of complexity greater than the mega/Xmega. It's also interesting to note how similar they are - I'm guessing either SAM3 reused UC3 peripherals or more likely both re-used SAM7/SAM9 peripherals.

Anyway, as such, knowing how to operate mega or Xmega is unlikely to give you much grounding in using UC3 or SAM3 (SAMn generally?) peripherals.

So packaging these things together in AS6 does not imply that programming SAM3 (or UC3) is going to be anything like AVR8 programming.

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

You have forgotten the Freescale cortex m4, they have precision analog like 16bit adc, flex flash etc… and available in different shapes, if I want to choose a cortex device I definitely go with NXP and freescale... others are still way back, Most of them lack a LCD controller, and freescale parts with TFT LCD controller are in BGA!!! (That’s a shame!)
I think NXP is the leader in Cortex M land… but freescale has also very attractive parts…and then we have ST in the third place

I love Digital
and you who involved in it!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

clawson wrote:
At the end of the day it's surely about cost and features isn't it? ...

Hmmm--How about "availability" and "longevity"?

E.g.: Xmega when announced had decent features, an advance on AVR8. Pricing ain't bad. Too long to market IMO; the window closed. Errata list nagging, to put it politely. Will compatible models be here a few years from now?

Can we apply the same litmus test to Atmel CortexM? Will we be fighting to get supplies of models announced but vapourware for a long time? Meanwhile other vendors have a good selection of readily available models in various M numbers. Is the window closing fast, or already closed?

You can put lipstick on a pig, but it is still a pig.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If any of us could predict the future I guess we'd all be rich! ;-) Who knows what the future is for anyone's line of Cortex chips. I suppose the key thing is maybe picking one that has a "workalike" available from a second source? Or maybe you just gotta back a particular horse and hope it makes it to the finish line?

It was interesting to see a post today in the Xmega forum which may be suggesting that virtually all the original devices could be replaced by the bug-fixed, USB enabled variants - I suppose as long as they (mainly) keep pin/register compatibility it doesn't matter too much? Atmel certainly made the right choice in the Xmega to keep them all to one common "family" architecture so that hopping about between models is much easier than in the mega arena - just so long as there are SOME models left at the end of the day.

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

feeling a bit lucky,installed the Segger Jlink driver on the same pc as AS6 is installed on. Opening a SAM project in AS6, it listed the Segger Jlink Debugger in the Tools drop down(it called it a SAM-ICE)..so it appears AS6 may support Segger Jlink tools. Of course a bit of further testing is required, but it seems logical that it should since ATmel SAM-ICE is a OEM sub-version of Segger tools. That would make the ARM Jtag debug tools a non-issue for those that already have Segger ARM tools! :)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Freescale Cortex M4 devices fall into long levity program, and they will produce chips between 10-15 years!

I love Digital
and you who involved in it!