XMEGA

Go To Last Post
440 posts / 0 new

Pages

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

Not to pee on anybodies campfire, but is there a company that doe NOT leak product information that is either arrives late or never?

Smiley

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

Joe,

I know one - I've worked here 24 years.

Cliff

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

There were two seminairs the other day in Germany where the X-mega was shown/discussed.

Is there anyone who has some info about it?

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

Nick, do you know if the information provided was under a non-disclosure agreement?

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

Quote:
if the information provided was under a non-disclosure agreement?
Most likely. No one is allowed to know when they REALLY are available :? maybe the odd freak knows...

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

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

This is what I found about the XMega:
http://www.changnamint.com/board...

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

I'm going merge this onto the end of the existing Xmega discussion...

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

Hi,

I just saw it on the seminar page from atmel. They reffered to another company: http://www.msc-ge.com/seminare/

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

Fleemy, thanks for the pdf. I'm still looking for the pdf preliminary data sheets that were posted by are now gine. Can someone pm them to me or tell me where to find them. Thanks!

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

Dean's still got them at fourwalledcubicle hasn't he??

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

Cliff,

It is dark out there :lol:
Can you pm them?

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

I don't have them (I'm waiting for my Atmel FAE to send me the real ones in fact) but, like I say, I'm pretty sure Dean cached a copy on his site so maybe try a PM to abcminiuser (but you'll have to wait for Aus to wake up now I guess)

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

Wouldn't it be neat if an Atmel Fab engineer told us how the first parts are allocated.... How many die can be made on an 8" wafer? How many wafers in a boat? So the whole first run goes to big outfits that have preordered 1000s of xmegas, one chip at a time? So we wait for the 2nd run for the FAEs to get samples? And how are fab runs scheduled? Wait till the orders for a boat load of 168s backs up, then run that one? So you always schedule the highest profit product? There must be a priority scheduling system for these things like a task list? How about a freaks signup sheet?

Imagecraft compiler user

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

I'm in!
JC

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

The prelim overview of the XMEGAs:

http://fourwalledcubicle.com/fil...

Note that this is probably totally out of date. If anyone sends me the leaked datasheets, I'll host those too.

- Dean :twisted:

Make Atmel Studio better with my free extensions. Open source and feedback welcome!

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

I wonder if the boardroom soap opera and CEO jump-ball goings-on will hurt Atmel in being chosen for large new design wins.

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

bobgardner wrote:
There must be a priority scheduling system for these things like a task list?

As I guess Atmel's goal is only really making the greatest return for their shareholders I imagine all the early samples went to the 10 million+ customers but the fact that most of the compilers are likely to have X support on day one means that maybe a few went out in that direction too - now, does anyone know anybody involved in compiler development?

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

If I guess 200 die per wafer and 20 wafers per boat, the initial run would be 4000 chips... which I assume go to the first 4000 customers with the highest orders. Can we assume Eric W is modifying the avr gcc compiler as we speak?

Imagecraft compiler user

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

Quote:

If I guess 200 die per wafer and 20 wafers per boat, the initial run would be 4000 chips... which I assume go to the first 4000 customers with the highest orders. Can we assume Eric W is modifying the avr gcc compiler as we speak?

LOL. Now >>that<< is thought-provoking.

-- EW made a recent post in the GCC Forum which I interpreted like this, relating to WinAVR support of new AVR models: "Since I'm intimately involved with WinAVr & Atmel, users should see WinAVR having timely support for new models."

-- There was another lively exchange a while back where IIRC EW insisted that Atmel did not give any special treatment to any particular compiler brand.

Ergo, if EW is, in fact, busily cooking up XMega recipes, then the same should be happening with Pavel at HPInfoTech and Richard at ImageCraft and at BASCOM and at CrossWorks and ... .

Lee

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

I've never met a pig I didn't like, as long as you have some salt and pepper.

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

Well the men in black from Atmel actually mentioned a number of different C compilers that were likely to have X support on day one - not just GCC. I don't think anyone's going to be too disappointed.

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

My inside info at Imagecraft was saying he was done a year ago when they announced the imminent introduction of the xmega.....

Imagecraft compiler user

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

Though presumably it later required actual chip samples to test that the implementation was OK?

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

We'll have to see once all the details are out, but I can certainly see where the AVR8 compiler vendors could certainly support these chips quickly, perhpas no more than any new AVR model line:

--It >>is<< an AVR, after all.
--I don't remember mentions of any new instructions. The instruction set is pretty much full anyway, right? Little if any room for new op codes.
--I/O registers are I/O registers.

It is setting up and >>using<< all these fancy new features such as DMA and port re-mapping that is the hard part. The compilers themselves probably care less--the code generation is going to be the same. It is going to be stuff like the app building Wizard that may lag way behind. [So again WinAVR is obviously the best since it doesn't concern itself with trivia such as app builders or IDEs or the like.]

Lee

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

I've never met a pig I didn't like, as long as you have some salt and pepper.

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

Don't know about the other compilers but to add a new entry to the -mmcu= compiler parameter to GCC actually requires changes in several places. Then the supporting io.h needs fixing up and an io???.h for the specific part(s) to be supplied.

But yes, as long as it conforms to the "avr5" architecture otherwise it should be pretty much a "shoe in" I guess.

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

Isn't there a couple of new instructions? A multiply and accumulate or something?

Imagecraft compiler user

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

It would be nice to have dsp instructions - proper pixed point arithmetic. Good dac's / adc's, analog comparators and plenty of ram/flash. Then I could make me a software radio. But maybe I'm dreaming and should find another processor company that have them on the shelf.

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

I don't mean to offend, I'm simply curious. What is the big deal about these XMEGAs? I feel I must be missing the point. Looking at the preliminary info Dean posted, they don't look THAT special.

Matt.

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

I concur. The chips will be outdated the moment they are launched. Right now I'm looking at luminarymicro's stuff.

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

Matt,

Depends what you are after - you might just look at them as an ongoing development in the line of Tiny - Mega - Xmega. The reason folks find Mega's "better" (whatever that means) than Tiny is probably the same reason why folks will find Xmega "better" than Mega's. I know some specific things that I thought "hey, it's about time AVR8's got that" when it was described to me. There are improvements in areas like ADCs (and added DACs), much better internal oscillators, DMA, interrupt and "event" priority systems and a whole bunch of other stuff that are just a natural progression. I guess users of mega88P probably prefer it to mega88 and users of that prefer it to mega8 and users of that prefer it to AT90S4433. It's just the natural course of development and maturity. (like I'd prefer to be typing this on a multi GigaHertz core 2 Duo than a 4.77MHz 8088!)

Cliff

PS That isn't to say that there's aren't some jobs for which the most lowly of Tiny's are still the obvious choice!

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

Thanks for the info Cliff. I was only going by the info presented in the chart Dean gave a link for so was not aware of DMA etc.

It'll be interesting to see what the pricing and power consumption are like for these devices. Things are getting very competitive in the high-end 8-bit/low-end 32-bit category so they're going to have to fight their corner!

Matt.

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

With rumors of being able to directly address 16 MB external SRAM and/or SDRAM in the previously linked PDF (http://www.changnamint.com/board...), I suppose at the very least there'd have to be an addition to the instruction set.

(Maybe an Extended LD family of instructions in conjunction with RAMPZ, as well as the RAMPX, RAMPY, and RAMPD registers that have been mentioned in the official instruction set for quite some time but never actually implemented thus far, in the same vein as the ELPM instruction that was added to support the >64 kB Flash parts.)

It seems to me that IAR's AVR-related documentation has advertised "far" and "huge" memory models that are intended to be able to access a 16 MB data address space for quite some time.
http://supp.iar.com/FilesPublic/...

It'd be very interesting to see how avr-gcc would adapt itself to that sort of development, since it's currently tied to a single, 16-bit data addressing mode.

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

lfmorrison wrote:
It'd be very interesting to see how avr-gcc would adapt itself to that sort of development, since it's currently tied to a single, 16-bit data addressing mode.
One of the avr-gcc developers thought that to support beyond 16-bit addressing avr-gcc would need to expand to a 32-bit addressing, bypassing 24-bit as well as not supporting a mixed memory model like IAR. My suspicion is the initial avr-gcc support will use a library function to access memory beyond 16-bits and will still be a 16-bit pointer compiler.

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

Wonder is 24 bits would be enough anyway? (he asked knowingly).

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

Forget 24-bit pointers, I'd sell my soul for (amongst a pile of other things including a never-ending chocolate bar) 24-bit integer support. Using 32-bits when 24 would do is so inefficient!

- Dean :twisted:

Make Atmel Studio better with my free extensions. Open source and feedback welcome!

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

Xmegas are going to be announced in the next couple weeks according to our local FAE here in South Florida. Took Atmel a while to get them out, but worth the wait to get DMA, event support, 384K flash, and 32Mhz clock.

OTOH, the AVR32 B series of mc's is also quite nice, and would seem to have otherwise overtaken the field that the Xmegas were supposed to fill.

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

Not sure how you consider an overlap between AVR32 and X-AVR8 ? They are quite different devices with quite different purposes. It's like saying an ARM overlaps any form of AVR8 - it doesn't.

(BTW I think your rumour may be just slightly optimisitic - but there is an intersting thing somewhere on the atmel.com web site ;) )

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

Quote:

but there is an intersting thing somewhere on the atmel.com web site

Hmmm--Google hasn't picked it up yet.

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

I've never met a pig I didn't like, as long as you have some salt and pepper.

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

Lee,

I (think!) you have PM (the system froze after [submit] so I'm not entirely sure it went)

Cliff

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

In few weekss, in our company would have a meeting to discuss if it's worth a try to include ARM's int our new 'big' products, and what other choices would be there. Right now, only XMega and new Cypress PSoC (with ARM cores) would be the best chances. Anyway, it seems that an RTOS would be a 'must have', for faster developing, including, possibly TCP/IP and/or USB support. Probably XMega will loose by the delay.

BTW, our Atmel representative told us they don't think any XMega would be released soon.

Guillem.
"Common sense is the least common of the senses" Anonymous.

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

Quote:
Xmegas are going to be announced in the next couple weeks according to our local FAE here in South Florida

Quote:
BTW, our Atmel representative told us they don't think any XMega would be released soon.

LOL, that doesn't exactly instill confidence does it :-) or maybe it is just the difference in perseption of what "soon" is :-)

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

Well, probably is not the same USA than Spain. Potential marked is greater on the first, maybe they are more interested to sell the first batches only there, and left the others for when they reach full production, huh?

Guillem.
"Common sense is the least common of the senses" Anonymous.

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

I'm REALLY curious to see the pricing. And not the PR flack "less than a few $$$ in 1.2 billion quantities"--the price to buy 100 or a flat from a distributor. Looking over the feature list again, an A4 16k/32k would compete with, say, Mega164/324. With all the fancy features, would it cost about twice as much? But that would be a killer as well, as Atmel's own AT91SAM7S32 has a lot of the features, and is $4.32/100; Mega324 is $3.98/100. So can it really be an $8 chip and still be attractive? Stay tuned...hold your breath for yet another year. lol

Lee

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

I've never met a pig I didn't like, as long as you have some salt and pepper.

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

Its an advantage to Atmel to have an ARM Eater right? If you have better features and less price, the only negative is second source availability, but that doesnt happen till something is a 'commodity'

Imagecraft compiler user

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

Quote:
But that would be a killer as well, as Atmel's own AT91SAM7S32 has a lot of the features, and is $4.32/100;

I totally agree. These things are going to need to be damn cheap otherwise there just isn't going to be a market for them. There are so many ARM chips around now offering at least the same kind of features as the current top AVRs, often at lower cost.

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

Doesn't it take an AVR32 to be an "ARM Eater"? Or, are you implying the XMega will have a sufficient increment in capabilities that it infringe on some 32-bit or 16-bit ARM7 Thumb applications?

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

No, what I'm speculating about is that if it os priced much more than the SAM7S it won't have much attraction.

Lee

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

I've never met a pig I didn't like, as long as you have some salt and pepper.

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

I see, Lee, that makes sense. The integrated DAC might save cost on an additional component for some applications, but of course only benefits those applications that depend on that additional component.

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

Well I heard that devices of similar flash size are likley to be priced at similar levels to existings devices of the same size. Although there may be more "goodies' on board this is offset by the fact that they are probably fabbed on a smaller process so the actualy silicon cost is the same or less.

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

Way cool! [downside: it is kind of a general rule that the smaller the geometry/feature size/power consumption, the less "robust" w.r.t. noise and other disturbances. cross your fingers.]

Lee

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

I've never met a pig I didn't like, as long as you have some salt and pepper.

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

In my email today, from Atmel re "Embedded World Nurnberg 2008":

Quote:
... meet with an expert technologist to learn what's new in Atmel's microcontroller offering and be the first to see Atmel's newest, innovative MCU family.
...
What You Will See

Product demonstrations exhibiting the latest technologies and capabilities found in:
8-bit Microcontrollers
A New Member of the AVR® Family
At EW2008, Atmel will unveil a major extension of the AVR 8-bit microcontroller proprietary architecture. This new family will extend the reach to 8/16-bit applications with significant architectural enhancements and best-in-class low power consumption characteristics.

Any day now...

Lee

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

I've never met a pig I didn't like, as long as you have some salt and pepper.

Pages