This isn't another PIC vs AVR flame war....

Go To Last Post
68 posts / 0 new

Pages

Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I just visited the Homebrew Robotics Club in San Jose, CA last night.

http://www.hbrobotics.org

I had a great time but I noticed a very disturbing thing... I was the only one in the room of 35 people that is using an AVR!! :shock: These people are obsessed with PIC.

Does anyone know why PIC is SO popular? Is it just because they got to the market first? Is there something I'm missing? I got really turned off by the high cost of the tools (programmers, compilers, etc). The fact that many of the PIC parts are 7 bit or 14 bit also left a bad taste in my mouth.

So seriously fellow Freaks, what is it that makes the PIC so much more popular? I mean 34 out of 35 people recommend PIC!!!!

Don't let this turn into yet another PIC vs AVR flame war. I'd like to hear the positive things about PIC. Don't worry, you're preaching to the "pry it out of my cold, dead fingers" AVR guy here.

Go electric!
Happy electric car owner / builder

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

I think the microchip data books have oodles of tutorial sw examples all ready worked out and ready to run. Could THIS be the only reason??

Imagecraft compiler user

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

I've been forced to do a lot of assembly programming for a PIC 16F84A as part of some coursework, and I haven't been impressed. There were a number of issues including one single interrupt vector for all interrupts along with a problem of missing external interrupts if you happen to read the port pins within a clock cycle of the interrupt event. Add to that the low throughput per clock speed and high part cost... IMO, the popularity is there because it's been around so long, and so many of the old peeps were raised on them... that and you can get them in SO many different packages with EXACTLY the subystems you need.

I couldn't be happier with my AVR-GCC and ATmega8! :D Well, I certainly wouldn't mind an ATmega168 if I could track 'em down.

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

Did you ask any of the PIC people why they liked it?

Perhaps they just need to see the light?

Go forth and evangelize! Yea, brother! Pick up thy mantle and show these poor oppresed souls the path to REDEMPTION!!!!

.... .. well y'know what I mean.... :D

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

EW wrote:
Did you ask any of the PIC people why they liked it?

Perhaps they just need to see the light?

Go forth and evangelize! Yea, brother! Pick up thy mantle and show these poor oppresed souls the path to REDEMPTION!!!!

.... .. well y'know what I mean.... :D

[The altitude is apparently getting to him.]

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

couldn't someone re-text that pulp fiction scene for AVRs? :twisted:

I tend to post off-topic replies when I've noticed some interesting detail.
Feel free to stop me.

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

EW wrote:

Go forth and evangelize! Yea, brother! Pick up thy mantle and show these poor oppresed souls the path to REDEMPTION!!!!

We should all shout from the roofs! ;)

But seriously: one of my best friends happens to be doing PIC development, and it's awesome the amount of information we share, all in an excellent spirit of colaboration and mutual learning.

That said, less than 5 MIPSat 20MHz does kinda kill him sometimes......

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

First let me say I'm glad this didn't turn into the typical PIC bashing session!

EW: No I didn't want to bring up that type of subject at my FIRST meeting! You know how it is...people get pretty defensive. My evil plan ( :twisted: ) is to enter a robot in the next "challenge" and THEN (if it does well) tell them it's an AVR! Of course if it explodes into a ball of flames then I may keep it a secret :lol:

These are a good bunch of people - very helpful and friendly. If anyone is in the area (San Jose) then I highly recommend showing up. The club dues are quite resonable -- IT'S FREE!

Also they have "challenges" every year. It is NOT a "competition". I like that attitude. Basically, it's everyone against some type of challenge. Pretty kewl.

Go electric!
Happy electric car owner / builder

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

Hi,

Convince your club to come to the AVR Seminars taking place soon - just register at Atmel.com. Tell them there is a free dev tool, and a free lunch. If it is a lot of hobbiests, the prospect of free stuff should convince them ;-)

-Colin

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

The first uC I tried was a PIC. I though it was very cool, but I really disliked the core architecture and the instruction set. An accumulator based machine is a pain in the @ss to program, and the instrution set mnemonics are ... at least very non-intuitive (and I have programmed assembly in 2 or 3 different machines).
I ended up doing an assembler syntax converter so that I could program in an assembly that made more sense to me. I was unable to find myself confortable with the original assembly.
Meanwhile I stopped having time. Recently I got back to uC, and while gathering some info on the new PICs, I found AVR... well, after seeing the datasheet of AT90S1200 I've simply thrown away all PIC datasheets. I loved AVR's architecture, it is almost a programmer's dream. I honestly can't find *any* advantage on PIC.
I also do not want to enter a "flame war", but AVR *really* is better. I know both architectures, I have worked a bit with both of them. I think microchip should redesign their micros, because I have no doubht that otherwise AVR will end up "ruling the world" :).

p.s. Eco guys don't worry, I didn't throw away the piles of paper PIC datasheets; I'm using their clean backs for draft AVR project drawings :)

Embedded Dreams
One day, knowledge will replace money.

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

Microchip spends more effort marketing to education and hobby folks. A lot more. This is a relatively small market.

Atmel focuses on the embedded processor world - high volume.

On a technical level, having used both, the Atmel is superior because it is not a paged memory architecture. This weakness is the root of many problems and inefficiences, especially in C compilers trying to cope.

It's sort of like the original 8080 micro with paging - versus the linear memory address model of the 68xx and the large memory mode of the later 80486 and pentiums.

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

I agree, that the architecture was definitively not the reason of its popularity.

I found the following reasons:

- the long and good aviability of the PIC
- the first free Windows tool chain for the PIC
- the easy getting of free samples also for kids and students
- in the sum, the very good advertising

Most PIC people seems knowing, that its architecture was not the best. Thus, I think, they react often very aggressive against other µC developers.
E.g. 8051, AVR, MSP developers can speak relative friendly with each other, but a PIC developer tells always, only the PIC was the Holy Grail, no others can exist beside it.
But this is only my personal impression from discussions with other developers.

Peter

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

Having written loads of code for mid-range PICs, e.g. 16F877, the real weakness to my mind is the lack of a proper stack. Apart from the problem of only having an 8 level hardware stack, there's no way to detect stack overflow, or to manipulate stack contents. Now, if I get a seemingly random crash problem and suspect it's caused by stack overflow, I call a dummy subroutine from the interrupt service routine to "prove the point". Having written assembler for many processors over the years, I've yet to find anything as elegant as the 6809. I still use it regularly writing slot-machine rebuild software. In fact I think of 6809 assembler as my second language, after English. If somebody could make a 6809 core with onboard FLASH, RAM and peripherals, I'd be first in the queue to buy it. It's not Harvard though, so I guess you couldn't squeeze the same performance out of it. Ultimately I think that people tend to favour the processor they learn first, but in reality they all have good points and bad points. I sincerely hope that most of the ant-PIC stuff in this forum is largely tongue in cheek. Is there an emoticon for tongue in cheek? Anyway, time for my medication now..

Four legs good, two legs bad, three legs stable.

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

I've done four or five commercial projects using PICs. One of the early ones used a PIC because it was the only device available in a small enough package with the peripherals I needed; others were used because of the relatively low development costs. I've always used the Hi-Tech C compiler, which does an execellent job (and gives you a detailed graph of stack use, so you can be _sure_ you won't get stack overflow).

I haven't made use of any PICs beyond the PIC16 generation, but I suspect that PIC18s get round quite a few of the architectural deficiencies (but IMHO the biggest deficiency with PICs is MPLAB).

Unfortunately, free dev tools seem to be the bane of all these micros. I find AVRStudio crashes every ten minutes; I use the IAR compiler (which is certainly not free!), but it doesn't support the ICE50, so I am having to debug with AVRStudio.

And as a final point (and it just illustrates the way everybody has a different perspective) I used to program 6502s in assembler (now that was a nice instruction set), but I found the PIC assembler (used on the first project I did) to be very easy to develop with.

Regards,

Colin

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

Quote:
I loved AVR's architecture, it is almost a programmer's dream. I honestly can't find *any* advantage on PIC.

Hallelujah! Now we have testimonials. :wink:

Quote:
- the easy getting of free samples also for kids and students

Atmel seems to be pretty good actually - they have a university program for example where they fit a lab with some STK500s and everything.

However I would say it seems to me a bit like Atmel's whole scheme is a bit off, not just for students. If anyone wants to get samples, its done through Atmel distributors / sales offices - which can take a long time. And some of the chips don't exist even...

If they had a system like Maxim for example (where it is all done online) it would probably go a long way. But that doesn't seem to be the way things are done at Atmel.

-Colin

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

sgomes,
We have a club like yours in Phoenix, with semi-annual events. The difference here is the controller of choice is the BX-24, an 8535 based stamp clone. :P

Course NetMedia, who makes the BX-24, and is based in nearby Tucson, is one of our sponsers so that might have something to do with it. :roll:

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

theusch wrote:
EW wrote:
Did you ask any of the PIC people why they liked it?

Perhaps they just need to see the light?

Go forth and evangelize! Yea, brother! Pick up thy mantle and show these poor oppresed souls the path to REDEMPTION!!!!

.... .. well y'know what I mean.... :D

[The altitude is apparently getting to him.]

Lee

Hey, hey, hey! The air's thin up here, sure! But I have a great view of the mountains! :P :lol:

To get back on topic, though:
I really don't know why people are obsessed with PIC. I've done a commercial PIC project before. I don't remember offhand any technical comparisons, but doing that project sure felt like poking my eye with a pencil. Repeatedly.

Luckily I have had an opportunity to do projects with a few different types of microcontrollers: 8051, 68HC11, 68HC16, PIC, and AVR. After a while I just got used to all the various kinds of tradeoffs; they all have their advantages and disadvantages. It seems to me, that being an engineer, it helps to be very flexible. I suppose that if a project called for a PIC, I'd do it. But I wouldn't be overly thrilled about it.

If those people want to use PIC, by all means go ahead. I wish you luck in "showing them up". Mabye you can also learn some things as well and bring it back to the AVR community. :wink:

Eric

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

I've never worked with a PIC (so I don't know how PICs handle this).
One weakness I find with the AVR is in the ADC converter.
The AVR limits the references you can use. If you have an external reference then you can't switch to the internal references.
I want to take an absolute measurement and a measurement in relation to an external source.
I also wish there was a way to change the low end of the ADC to something besides ground.
(The HC11 does this, i.e. -I can set the entire range between 2.4 and 2.5 volts)

Besides these limits, I think the AVR is a great micro. I've used almost every version they make.

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

Just my two cents worth - I originally did a project using a 2313 (a few years ago) which was extremely simple (interrupt, timers, sleep mode, that's all). However, the person I did the project for was having problems getting Atmel to guantee availability of 1K quantities. I had to do the project with a PIC and almost lost all of my hair! Had problems finding chip with right combination of timers, interrupts, etc. and I thought their instruction set stunk!

Never again!

Randy

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

In groups, such as the robotics clubs described at the beginning of this thread, its often "strength in numbers". You get one or two who get used to processor X and, whether its a good fit or not, most of the new folks make the same choice.

In many such groups, the people are not electrical engineers or embedded processor programmers. Even simple concepts like a "direction register" is Swahili to them, let alone UART, PWM, and all that other stuff.

The same goes for many companies. You don't want to invest in learning some new processor unless the need is so clear, or overwhelming, that you just cannot avoid it. It means procuring and stocking new components. It means, even with C, non-reuse of code and all those good things that engineering managers try to encourage.

Rather than wringing our hands about "why-not-AVR?", we need to be able to provide concrete examples how our choices result in higher performance, lower development time, added features, lower system cost. And, we should not worry because "X is lower power" or "Y has a better ADC" because any spot measure will always have something else that does some application better.

On the other hand, we need to encourage Atmel to continue steady improvements in the family, because the world does not stand still. Their record has, it seems to me, been a bit spotty here. Certainly, they have made some excellent improvements (increased memory read/write cycles, internal oscillator, more than 8 bit ADC, TWI, SPI, etc). On the other hand, does anyone remember the pain when the 10 bit ADCs first came out? But, lets make sure they don't stop the improviements and lets make sure that the improvements are genuinely useful!

Jim Wagner

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

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

I think the other thing that put PIC's into everyone’s heads was the development of the "ModChip" for the Playstation, I bet Microchip were scratching their heads wondering why sales of the 12C508 (from memory) went through the roof.

Also, alot of hobbyist magazines are PIC biased IMHO.

I learnt asm on the 6809, I still prefer Motorola assembler over AVR, but Motorola just don't make anything as neat as Atmel for $5!!. That's where the AVR is great, just look at some of the low pin count AVR's and they are loaded with features.

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

On a similar thing to the ModChip use for PIC, the AVR seems quite popular with the satallite cracking community. In fact when the AVR was very hard to get a while ago you could find them at huge mark-ups on E-Bay, often targeted at that market.

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

I am considering (vaguely) using a PIC for a small run (about 20) of a PC_keyboard_to_MIDI_Notes project (#186 in the project forum here).

I'm looking for an 8-pin device that has a UART, internal oscillator of about 1-2 MIPS and two or three port pins.

And is cheap. The cheapest that I can find 8-pin AVRs is at DigiKey and they are in the $2-$3 range (with no internal UART). The PIC 12F629 does most of what I want (again it has no internal UART) and sells for $1.34 on GlitchBuster.com.
I would like to use the Tiny2313. However it appears to be non-existant as its release date keeps being pushed ahead another quarter every time the projected release date comes around.

However, I probably won't go with PIC because I dread having to learn another new complex new processor along with paying for another set of expensive programmers and development tools.

I agree that Atmel could be a lot more flexable with their AVR feature set. They should make it a lot more easier to get the chips also. It wouldn't hurt to FIX the Windows assembler so that it can handle source files bigger than 20K.

But the ISP feature and the ICE200 emulator and occasional availablity of relatively inexpensive AVRs on Ebay make me want to stick with the family.

By the way, where was that highway in the mountains photo taken?

And, yes, the Motorola 6809 was the most flexable microcontroller ever. A inexpensive 6809 with Flash In-Circuit-Programmable memory, on board timers, UART, I2C, EEPROM, SRAM, and ADC would blow every other microcontroller out of the water.

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

I quess PIC's populairity is because they have been around for a while.
It's simple instruction set and architecture.
For a lot of people it most have been their first encounter with the subject of microcontrollers.
Almost all hobby projects I have seen are PIC based, even I have done a lot of PIC programming.
But after a while you simple get fat up with it's shortcomings and start looking for alternatives.
Why I choice AVR? I don't know really, it could also have been the H8 or ST7 or MPS430 or anything else.

"Digital is a special case of analog" - George Philbrick

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

Simonetta wrote:

And, yes, the Motorola 6809 was the most flexable microcontroller ever. A inexpensive 6809 with Flash In-Circuit-Programmable memory, on board timers, UART, I2C, EEPROM, SRAM, and ADC would blow every other microcontroller out of the water.

It would, but then the price of the dev system from Motorola would blow a hole in the VISA card. I've looked at the HC08 and HC12 and they are loaded with goodies and I've already praised the Motorola instruction set in my previous post, but they are all to expensive not only per part but the dev systems are pretty pricey too.

Motorola don't care, I saw recently Zilog offering a $49 dev kit, everyone seems to except Motorola.

If the Tiny2313 ever surfaces that will be a great choice in the AVR range, I use the 90S2313 at the moment and it would be great to have the extra features of the Tiny version.

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

My opinion on Microchip popularity :

Before 1993 hobbyst must have or build an Eprom programmer to use microcontrollers (8751, 68705...) Packages where a little too expensive (UvProm package)

With Microchip 16C84 hobbyst got :
-low cost DIP plastic package,
-low cost home made programmers,
-data books and applications books with plenty of application notes,
-FREE BBS access !!!

That's why PIC chips became so popular among hobbysts.

In 1997 AVR new design arrives with plenty of qualities : good architecture, speed, Flash...

Microchip resists with new chips PIC12 and PIC18 and Flash design.
All their chips are well distributed, prices are good, samples are very easy to get.
Parallax Stamp is popular is many schools....

I had to struggle 2 month before getting 2 Mega8 and 2 Mega 16 sample chips !!!
How long before putting hands on Tiny2313 ?

So the challenger Atmel has to struggle hard ! AvrFreaks is a good support !!

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

Maybe this attachment can be the answer :shock: :lol:

Have a nice day,

- -big bang- -

Attachment(s): 

Take a good look around. Admire our world, our galaxy, the whole universe. Then you just feel there must be a
'Theory of everything'

Maybe, someday, the human race is ready to discover it.

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

Simonetta wrote:
By the way, where was that highway in the mountains photo taken?

I-70 in Colorado, a few miles west of the Denver metro area, you are looking west from the Genessee Park exit. The snow-capped peaks in the back is the Continental Divide, where all of those mountains are above 14,000 feet (4,267 meters). The bridge where the picture is taken from is about 8,000 feet in altitude. Oh, and it's about a 5-10 minute walk from where I work. :D

Alas, it's not a recent picture. Spring is not completely here yet so all the grass is not as green as in the picture. But it will look that way probably in about another month or less.

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

I wouldn't mind seeing this topic expanded some ...

There are gazillions of microcontroller families out there, with all sort of features, feechures, bugs, whatnot. I'm sure that there are plenty of people here that are not completely monogamous to the AVR, or at least have experience with other families.

What's good about them ? What's bad about them ? Tools, cost, availability, quirks, coolthings, and so on. A comparison would be very enlightening.

Dean.

Dean 94TT
"Life is just one damn thing after another" Elbert Hubbard (1856 - 1915)

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

c_oflynn wrote:
Quote:
I loved AVR's architecture, it is almost a programmer's dream. I honestly can't find *any* advantage on PIC.

Hallelujah! Now we have testimonials.

grunf, very funny :p :)
Ok, I admit that the sentence sounds like my religion is AVR :), but I was very disgussed at PIC, so that was what I felt when I saw AVR's architecture for the first time!

Embedded Dreams
One day, knowledge will replace money.

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

I wouldn't mind seeing this topic expanded somewhat as well. I'd never heard of that Philips chip with the ARM7 core for example, and it sounds v. interesting to me. But I guess the people who run this forum would argue that it's not the place for such discussion, and I can't think of any arguments.

Four legs good, two legs bad, three legs stable.

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

Hi,

I just attended TI's MSP430 day seminar a couple of day ago. They were showing off the MSP430. I was pretty impressed by it. It's a 16-bit mcu and the thing that i loved about it - it's got a DMA. They had this small app where they had a small voice recording and playback unit. According to the presenter, that app needed negligible CPU time, i.e the data went from the ADC -> DMA -> Memory. And while playback, it went from Memory -> DMA -> DAC. TI is really pushing on its ultra low power consumption. Also, because of its low power consumption and hence, low EMI, the ADC can be run without any problems with the processor running. It has pretty much all the features the AVR does and more. Some are available with a large amount of RAM, good for running RTOSes. And they're pretty cheap. Now, don't get me wrong, I'm still a hard core AVR fan.

However, I was just trying to compare the MSP430 and the AVR and was trying to figure out which would be better suited for 16 bit applications. I'm assuming that power isn't the most important consideration, i.e. we don't exactly need the app to run for 10 yrs off a coin cell.

The MSP430 has lower instruction throughput, I think ~8 MIPS as compared to the 16MIPS of the AVR. But AVR would need 2x or more cycles to complete a 16-bit calculation. Also, the MSP430 has a hardware multiplier. (Does AVR have one??) And they also have some efficient instructions pertaining to moving data around the individual blocks. And for some situations, the DMA is extremely useful

The only disadvantage I see is that TI doesn't have an eqvt of AVRStudio and simulation would be a problem. Also, I wish they had more registers (has 15) and a Harvard arch (instead of von neuman). But I guess, they compromised on both counts for power.

Thoughts?

Cheers,
Jimmy

PS: Does Atmel have plans of a 16 bit AVR?

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

I've used MSP430. As Jimmy suggests, its got strong points and weak ones.

Low power is certainly a biggie. And, their price point it generally targeted to meet or beat PIC.

We seemed, sort of, to be forced into IAR compiler. Didn't like that so much.

There is one thing that constantly did frustrate me. It was so "flexible" that you could do 4 or 5 things with almost every pin. There were lots of registers. And, they were immensely confusing. I'm generally pretty good at being able to decipher spec sheets and there were places where it took 3-4 heads to make sense of it. Its the sort of thing where a bit does one thing if other config bit is in one sense, then something else if the config bit is complemented; sometimes the configs were several bits in combination so that you might have several meanings for a register bit depending on the various configurations. Great if you need it, a real pain if you want to do something really simple.

I was also frustrated by the lack of any package but fine pitch smt.

DMA is nice if you need it. But, I, personally, have never needed it.

Jim Wagner

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

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

Quote:
Also, the MSP430 has a hardware multiplier. (Does AVR have one??)

Yes. Atmegaxxx have hardware multiplier.

Quote:

PS: Does Atmel have plans of a 16 bit AVR?

I think not. I recall reading a comment that 16-bit architecture will tend to disappear, because 32-bit architecture are getting cheaper. Also, Atmel already has the ARM architecture (AT91 ARM Thumb).

Regards,
Alejandro.
http://www.ocam.cl

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

Quote:

PS: Does Atmel have plans of a 16 bit AVR?

There have been similar threads on this sort of topic - basically it ends up that although we all love the AVR, it is really just an 8-bit micro. It isn't designed to do lots of fast math or floating point. So if you want to do that - the AVR isn't that sort of processor.

So there will probably never be such a thing as a 16-bit AVR, as then it isn't really an AVR ;-)

-Colin

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

ka7ehk wrote:

There is one thing that constantly did frustrate me. It was so "flexible" that you could do 4 or 5 things with almost every pin. There were lots of registers. And, they were immensely confusing.

Ya, I hate that too. On paper the features are impressive, but then when you actually sit down and try to work, invariably, the combination you need, doesn't work. :roll:

aweinstein wrote:

I think not. I recall reading a comment that 16-bit architecture will tend to disappear, because 32-bit architecture are getting cheaper. Also, Atmel already has the ARM architecture (AT91 ARM Thumb).

Yeah, I guess that does make sense.

c_oflynn wrote:

So there will probably never be such a thing as a 16-bit AVR, as then it isn't really an AVR

:)

Cheers,
Jimmy

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

One thing that would be "cool" was to see higher speed versions of AVR. They seem to be "stuck" at 16MHz. I know that PIC doesn't go that far, but take a look at the SX family of ubicom:
http://www.ubicom.com/products/sx/sx.html
Unfortunately its architecture is similar to PIC's, although the assembly makes more sense! But they are very cheap (~4$ ~ 9$).

Embedded Dreams
One day, knowledge will replace money.

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

Nuno wrote:
One thing that would be "cool" was to see higher speed versions of AVR. They seem to be "stuck" at 16MHz.

The newer ones (Tiny13, Tiny2313, Mega48/88/168) are rated for 20MHz. As far as I know the basic problem is the speed of the internal flash. There is just no 10ns flash around.

Markus

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

Weren't the original AT90S1200s rated at 24 MHz? I guess they were unstable or something...

-Colin

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

c_oflynn wrote:
Weren't the original AT90S1200s rated at 24 MHz? I guess they were unstable or something...

I haven't seen that, but I haven't been with the AVR all the way from the beginning. Only almost.
I do have a 16 MHz version of the 1200, and a handful 2313 at 12 MHz.

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

MarkusK wrote:
As far as I know the basic problem is the speed of the internal flash. There is just no 10ns flash around.

Well, then maybe they should start using other technics, like copy the flash to a special program sram at startup :)

Sram is another point where they could improve: currently they take 1 extra clock cycle, right?

Embedded Dreams
One day, knowledge will replace money.

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

Nuno wrote:
One thing that would be "cool" was to see higher speed versions of AVR. They seem to be "stuck" at 16MHz.

My Feb 04 copy of the atmel quick reference quide shows mega48, mega88 and mega168 all rated 24MHz

Imagecraft compiler user

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

Nuno wrote:
One thing that would be "cool" was to see higher speed versions of AVR. They seem to be "stuck" at 16MHz. I know that PIC doesn't go that far, but take a look at the SX family of ubicom:
http://www.ubicom.com/products/sx/sx.html
Unfortunately its architecture is similar to PIC's, although the assembly makes more sense! But they are very cheap (~4$ ~ 9$).

I don't have any datasheet in front of me right now, but I am almost sure the PICs do go up to 20 MHz. The problem there is, though, that at 20 MHz they deliver less (usually significantly less) than 5 MIPS. AVR got them beat, pants down.

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

gianmario wrote:
I don't have any datasheet in front of me right now, but I am almost sure the PICs do go up to 20 MHz.

Last time I saw, PICs could go up til 40MHz. But at this speed, they deliver only up to 10Mips, because they execute an instruction every 4 clock cycles.
I guess I'll be really happy when a high-speed USB 1.1 (12Mbps) device can be done with something similar to an 2313 :) (low-speed USB already exists).

Embedded Dreams
One day, knowledge will replace money.

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

bobgardner wrote:

My Feb 04 copy of the atmel quick reference quide shows mega48, mega88 and mega168 all rated 24MHz

Datasheet change from 01/04 to 04/04:
Speed Grade changed from 24MHz to 20MHz.

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

Quote:
Well, then maybe they should start using other technics, like copy the flash to a special program sram at startup

I think again that seems to be losing what the AVR is - as the architecture is designed to execute from FLASH. If they needed all sorts of more SRAM that would just jack up the price... the AVR is great at what it does, no point making it a "jack of all trades - master of none".

Regards,

-Colin

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

c_oflynn wrote:
I think again that seems to be losing what the AVR is - as the architecture is designed to execute from FLASH.

Well, the first AVR probably didn't have UART, ADC, SPI, ... do you think ATMEL shouldn't improve AVR, that AVR is just right as it is right now :)?
It is not relevant if the micro executes directly from flash or from an automatically startup-made copy of the flash in sram. You won't notice the difference.

c_oflynn wrote:
If they needed all sorts of more SRAM that would just jack up the price...

There could be 2 versions, a "faster" one, a bit more expensive, and the "classical" one, the same price as today. What is the problem?
Besides, I think you know that an IC price is basically to pay its development. SRAM is something that already exists, it doesn't seems to me that the development of such a mechanism be much harder than the development of, let's say, a new peripheral. But of course, I'm not from ATMEL :)
Anyway, other things make up the price.
Look at the SX family from ubicom; they aren't very expensive and they run at 50/75 MHz.

c_oflynn wrote:
the AVR is great at what it does, no point making it a "jack of all trades - master of none".

I really don't see how a speed upgrade may do that to AVR.

Embedded Dreams
One day, knowledge will replace money.

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

Hi,

What I was talking about is that the SRAM might be external then - like most processors that copy it from FLASH to SRAM. When you do this it increase the complexity of it, and they also need a bigger pinout.

If they did the RAM internally, that is a LOT of SRAM to cram on a chip. Yes they COULD do it, but I kinda doubt they wood. Especially if this really is SRAM, and not DRAM.

For example 16Mbit (not Mbyte) of SRAM is about $88 (Canadian) at Digikey. I'm sure it could be had cheaper, but it still shows the cost of this stuff.

Quote:
There could be 2 versions, a "faster" one, a bit more expensive, and the "classical" one, the same price as today. What is the problem?

By that time IMHO it seems you should just go for Atmel's 32 or 16 bit processors (ARM).

Regards,

-Colin

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

The original problem was flash speed. As such, I was talking about an internal sram, similar size of what they have now. Just to increase flash speed. No need for bigger (not on the megabit, that's for sure) sram, just keep the low sizes (now up to 128Kb, right?), nor for external.

Quote:
By that time IMHO it seems you should just go for Atmel's 32 or 16 bit processors (ARM).

I don't want an ARM processor; I like the AVR simplicity, architecture and price range. I just would love it would be faster, because then I could use it for a new range of applications keeping the simplicity (There's one principle I love: "KISS" = Keep It Simple, Stupid).

Embedded Dreams
One day, knowledge will replace money.

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

Quote:
(There's one principle I love: "KISS" = Keep It Simple, Stupid).

Doesn't adding this internal sram you mention contradict the KISS principle ? :) (At least from the AVR designer point of view ).

Quote:
I like the AVR simplicity, architecture and price range. I just would love it would be faster, because then I could use it for a new range of applications...

Just for curiostity, which kind of applications will benefit from an increase in speed ?

Regards,
Alejandro.
http://www.ocam.cl

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

aweinstein wrote:
Doesn't adding this internal sram you mention contradict the KISS principle ? :) (At least from the AVR designer point of view ).

You already answered :). Doesn't contradict, from the AVR user's point of view.

aweinstein wrote:
Just for curiostity, which kind of applications will benefit from an increase in speed?

Personally, these are the things I would like to be able to do with an AVR: CCD/CMOS video capture/(light) processing, high-speed USB host interface (for the hw low cost :)), higher speed "BASIC" interpreters.
These little guys are so simple and general purpose that it seems that we can do everything with them. Even more, if they were faster.

Embedded Dreams
One day, knowledge will replace money.

Pages