Alf-Egil Bogen on the beginnings of the AVR

Go To Last Post
34 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi All,

Thought some of you might be interested. Alf-Egil Bogen, of of the creators of the AVR, speaking briefly about the beginnings of the AVR microcontroller:

http://youtu.be/s6JqigdNy9M

What would you like to see more of?

Eric

Last Edited: Mon. Apr 9, 2012 - 04:19 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
What would you like to see more of?

The next episode.
When I started with AVR's, I was happily surprised by the luxury of all those builtin peripherals, flash , RAM and EEprom. Fast and smart. They managed to build a bridge between HW and SW. That was some good thinking when the fundament was defined. And we still benefit from that.

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

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

I think what made them successful was the inclusion of all those features in one piece of silicon. With the exception of the eeprom, microcontrollers already had those features, but not all at once. The manufacturers had you choose a product according to feature, which sucked.

Also, the unbridled speed! Most micros of those times seem to target the 8051s in processing power, which is not very fast by today's standard.

I would love to find out the motivations behind the 1 instruction/clock. This was pure genius, considering that the avrs could have turned out so much slower.

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

EW wrote:
What would you like to see more of
The Mac; are y'all trying to torture us? ;-)
(I'm hoping for non-Microsoft solution)
btw, nice product placement(s) in the video's background.

"Dare to be naïve." - Buckminster Fuller

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

I saw the Raven and smiled.

I saw the Mac and ... Never mind, I'm just not a Mac lover.

I did enjoy listening to the presentation, however.

JC

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

It is interesting to note that the AVR and the SX came into being at about the same time. What is even more interesting is that while they both had 1 clock cycle/instruction for most instructions, the SX started at 50 MHz and quickly went to 75 MHz while the AVR started out at 8 MHz and took forever to reach 32 MHz. At 75 instructions per uS, a lot can be accomplished with a 1 MHZ interrupt rate.

How much longer will it take Atmel to catch up to a 16 year old obsolete microcontroller? Imagine what could be done with a 13 nS instruction cycle. :twisted:

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

MarioRivas wrote:
I think what made them successful was the inclusion of all those features in one piece of silicon. With the exception of the eeprom, microcontrollers already had those features, but not all at once. The manufacturers had you choose a product according to feature, which sucked.

Also, the unbridled speed! Most micros of those times seem to target the 8051s in processing power, which is not very fast by today's standard.

I would love to find out the motivations behind the 1 instruction/clock. This was pure genius, considering that the avrs could have turned out so much slower.


Indeed. And the driving capacity of the pins ... and the very high impedance inputs, yet protected. Selectable pull-ups ... what more did I forget ?
Oh, and the Xmega's are even more sophisticated.

Hmmm. Almost a commercial :lol:

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tricia, and Ulyana. You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

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

I was attracted to AVRs, coming from 8051 and then PIC, was the linear address spaces without bank-switching/paging, and a real hardware stack pointer. Like night-and-day vs. the PICs. (yes, the top-end PICs today reduce bank-switching and stack simulation).

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

RickB wrote:
... the AVR ... took forever to reach 32 MHz.
Flash access speed is the limiter IIRC. Some creators and fabs select/have a "better" process. I assume Atmel and fab-less closes some doors (speed) and opens other doors (price, risk reduction (unfortunately witness Renesas)).

"Dare to be naïve." - Buckminster Fuller

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

During an early AVR workshop I heard a report of how your first meeting with the CEO of Atmel at the time (George P?) went.

My memory isn't the best but I think it went along the lines of:

George:
"I'm running very late guys, I've only got 5 minutes for this new microcontroller idea presentation of yours. OK?"

Alf-Egil:
"Oh I'll summarise the hour long presentation I had planned. Here's the last page and here's how much money we're going to make for Atmel."

George:
"Wait there guys. I just gonna cancel the rest of my day's appointments... Now what were you saying?"

It gave me goosebumps at the time and it does again now.

I would love to know if that really was how it went.

Either way. Well done and thank you.

Pete

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

Quote:
What would you like to see more of?

A personal appearance by yourself with other members of the avr-gcc team telling us how that came about.

The avr-gcc toolchain is the thing that I like most about AVR. At about the time of ATMEGA8 everything about AVR was the best. Recently though the competition has caught up and sometimes past AVR on chip development. They still have their "crapy development system" though as Alf-Egil Bogen puts it.

John

If all else fails, read the instructions.

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

JohnWalton wrote:
Quote:
What would you like to see more of?

A personal appearance by yourself with other members of the avr-gcc team telling us how that came about.

The avr-gcc toolchain is the thing that I like most about AVR. At about the time of ATMEGA8 everything about AVR was the best. Recently though the competition has caught up and sometimes past AVR on chip development. They still have their "crapy development system" though as Alf-Egil Bogen puts it.

John

I hate to agree with you, but in many regards you are right.

My concern is that Atmel is trying to bridge this performance gap by continuing to pour resources in their ARM stuff. If you look at their product offering with the SAM stuff, you will see a handful of redundant families that overlap in capabilities. On the one hand you have the SAM9, which is badly needed. The included video decoder and LCD controller is simply amazing. On the other, you have the SAM7 at the lower end of the market, which is also necessary. But what about the recently anounced SAM4 and other varients? Excluding the different members within each family, there's significant redundancy within their ARM processors.

Meanwhile we are left with the AVR family. Personally I think it is the best 8bit micro family ever created by man, but I still would like to see more features.

I want higher bit ADC with differential inputs, true 32 bit counters with support for compares, not just input capture. And more speed! I want the xmegas to run to at least 66Mhz, AVR32 to 99MHZ. I want more flexible EBI support with 16bit addressing .

Unfortunately, with the SAM stuff, Atmel seems to be diverting their attention from their AVR family and becoming like their competitors, creating more and more redundant micros hoping for something to stick.

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

JohnWalton wrote:
Quote:
What would you like to see more of?

A personal appearance by yourself with other members of the avr-gcc team telling us how that came about.

I'm planning on being at the Atmel Technology Live conference in San Jose in September.

Also, tentatively, I'll be at ESC Silicon Valley, again in San Jose, at the end of March.

I realize that might not help, you being in Stockholm. :(

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

I read John as "a personal appearance [in a video] by yourself and [...]".

If you do that, even I might consider joining FB.. :wink:

I concur with others that one of the greatest things about AVRs is ar-gcc/avrlibc, and I'd love t hear the "How it all started" story from those who where involved.

As of January 15, 2018, Site fix-up work has begun! Now do your part and report any bugs or deficiencies here

No guarantees, but if we don't report problems they won't get much of  a chance to be fixed! Details/discussions at link given just above.

 

"Some questions have no answers."[C Baird] "There comes a point where the spoon-feeding has to stop and the independent thinking has to start." [C Lawson] "There are always ways to disagree, without being disagreeable."[E Weddington] "Words represent concepts. Use the wrong words, communicate the wrong concept." [J Morin] "Persistence only goes so far if you set yourself up for failure." [Kartman]

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

I agree, it would be interesting to hear about the philosophy behind having such a broad range of models. A related subject would be the cost of producing the chips, although I suspect that is all proprietary information.

The point is, last time I looked there were about 30 Xmega models. Obviously the chips scale in capability as one adds more ports and memory, but what is the real difference in profit when one removes a DAC and has a new model? The indirect costs for design, scheduled production, inventory, shipping, documentation, etc., are significant for each model added to the line-up.

I guess one factor at play is the chip manufacturing houses model of charging per transistor, or per unit area of silicon? Eliminate a module and a couple of K of memory and the chip is fractionally cheaper.

Perhaps the other driving force is the mindset of the end users, "I need X number of pins, Y number of Timers, and Z number of ADC's and not a transistor more". The goal is to give the User exactly what they think they want?

This just strikes me as an archaic approach, but one which the industry is based upon and still holding to.

JC

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

Yes I did mean an appearance in a video.
Sorry misunderstood.

John

If all else fails, read the instructions.

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

A-ha. I see. :)

I'll have to get over my aversion to cameras..... *cough*

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

MarioRivas wrote:
... but I still would like to see more features.
IIRC, there is a hint of a forthcoming XMEGA C series (CAN?).
Keep improving picoPower with AVR's ruggedness (recent examples are XMEGA B series and XMEGA256A3BU).
Add a Bluetooth Low Energy (BLE) series to: 1. compete against TI, and, 2. integrate with Android (XMEGA BLE?).
Does Bluetooth avoid the registration issues of USB and IEEE 802?

"Dare to be naïve." - Buckminster Fuller

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

EW wrote:
A-ha. I see. :)

I'll have to get over my aversion to cameras..... *cough*

Give them a doctored video that substantiates your thumbnail picture Forum avatar: an Invisible Man wearing a derby! (sadly, don't know how hard this is to produce on a shoestring)

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

Levenkay wrote:
Give them a doctored video that substantiates your thumbnail picture Forum avatar: an Invisible Man wearing a derby!

:lol:

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

They make green screen head covering just for this purpose. You'd put on the covering the derby and suit and stand in front of green screen with some suitable background substituted. Sort of like this only with a hat:
http://www.youtube.com/watch?v=a...
But are you sure you want Atmel represented by an 'empty suit'?

Smiley

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

smileymicros wrote:
They make green screen head covering just for this purpose. You'd put on the covering the derby and suit and stand in front of green screen with some suitable background substituted. Sort of like this only with a hat:
http://www.youtube.com/watch?v=a...

Hmmm.... :shock:

smileymicros wrote:

But are you sure you want Atmel represented by an 'empty suit'?

:lol: No. No, no. I guess that's not appropriate either.

I've just never been fond of being in front of a camera.

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

I agree built-in Bluetooth would be a useful feature for many products. Now, which Profiles to include...

Additionally, any Power Supply designers at Atmel?

If one switches from a Mega or Xmega and an FTDI chip to do the USB interface to using a micro which includes the USB interface one can save on the extra chip, board space, parts count, etc. (ignoring the cost trade offs for the moment).

BUT, for some projects powered off the USB bus using the FTDI chip eliminates the need for a 5 V to 3.3 V regulator. The FTDI can source 3.3 V at 50 mA to power the micro and associated circuitry, (sensor, memory, etc.).

An Xmega series, with USB, that included the (linear) voltage regulator on chip would further simplfy a number of designs. I suggest an LDO Linear Voltage Regulator as the internal power loss from a 5 V USB supply to Vout 3.3 V is low, and the linear mode is apt to contribute less noise to the ADC & DAC modules already incorporated with the micro's core.

This would make more sense to me than adding Xmega Ver #31, which doesn't offer much more or much less than its nearest neighbors in the line-up.

JC

Edit: I like Photos, the old "A picture is worth a thousand words" concept. Here is a FTDI chip providing the USB interface for a Mega168, used as a PDI programmer. The FTDI powers both the Mega168 and the Xmega Target, Gabriel's Xprotlab in this case.

This could easily be a "one chip project" using a USB enabled micro if it, like the FTDI, included the voltage regulator.

Attachment(s): 

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

DocJC wrote:
An Xmega series, with USB, that included the (linear) voltage regulator on chip would further simplfy a number of designs.
V-USB Hardware Considerations Solution A second variation is an alternative (minus pull-up and series R).
Typical on-MCU voltage regulators can be destroyed when the input voltage reverses into a low impedance and the current flow from the regulator's total output capacitance over-currents the pass transistor.
The on-MCU voltage regulator can be designed per a reverse current spec but typically it's much more robust and option capable to use an external regulator.
If the MCU is powered from a supercap, cell, or battery the regulator is the charger.
Charger - some MCUs do have an on-board buck, buck-boost, or boost regulator or regulator controller to recharge the long term bulk storage; this is on some of the Freescale i.MX application processors (ARM9, Cortex-A?).
Instead of having an external PMIC (power management IC) it's internal.
Powering an XMEGA B from a LR44/SR44 cell or two PV series cells would be useful for smart medical field instruments.

"Dare to be naïve." - Buckminster Fuller

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

Quote:

NO! To make better use of tiny series where every pin counts.

How does re-ordering the pins have any bearing on that? Surely it's down to your PCB routing software as to how it copes with a fixed layout?

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

Quote:

Remapping enables you to put what you feel like on any pin you feel like.

Which processor are you talking about - I've used processors with muxes but actually you didn't even have as much control as UART+SPI on 13. It was groups of functions could move to groups of pins (on a ~500 pin BGA).

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

Is it worth going back to see how this sticky:
Alf-Egil Bogen on the beginnings of the AVR
Turned into pin remapping?

Smiley

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

RickB wrote:
What is even more interesting is that while they both had 1 clock cycle/instruction for most instructions, the SX started at 50 MHz and quickly went to 75 MHz while the AVR started out at 8 MHz and took forever to reach 32 MHz.

But that came with a price. I've seen SX28's with heat-sinks on 'em. ;-) The SX was about double the power consumption of the 90S1200 at the same speed. By the time you got to 50MHz @ 5V you were cooking off like a third of a watt just in the core! Throw in an instruction set and architecture from 1980 with pretty dodgy silicon and tools and it's easy to see why they just sorta... disappeared...

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

Quote:
Throw in an instruction set and architecture from 1980 with pretty dodgy silicon and tools and it's easy to see why they just sorta... disappeared...

The instruction set and architecture was primitive, but the tools from Parallax worked well. It was taken off the market for legal reasons.

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

Quote:
Which processor are you talking about

As I stated it is not Atmels.
Please request one more time before I tell you.
Best google for yourself.

John

If all else fails, read the instructions.

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

Hint - if John tells you, then he'll be accused of pulling a "Leon" :)

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

Hi All,

Let's stay on topic to the thread. If you want to talk about remapping, start a new thread. Thanks!

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

Can't you or rather a moderator cut out the crap so that folks in future won't have to skim over it to find the relevant stuff?

Smiley

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

smileymicros wrote:
Can't you or rather a moderator cut out the crap so that folks in future won't have to skim over it to find the relevant stuff?

Smiley

Done. Posts moved to https://www.avrfreaks.net/index.php?name=PNphpBB2&file=viewtopic&t=117869