Single cog Propeller?

richaj45

Musings:

I think the propeller is interesting but i wish they would have made a single cog chip, running at 100Mhz with 4k words of memory. The cog has a nice
instruction set architecture it just needs to address more than 512 word of memory. Increasing the instruction length to reach 4k words would be adequate for these applications. The hub concept is very good but it takes expensive,in die area, hardware to make it as fast as the basic cog.

Any commnets?

cheers,
rcih

<subject line added by moderator>

Post Edited By Moderator (Paul Baker (Parallax)) : 3/7/2008 7:00:02 PM GMT

deSilva

ARM7 based chips are at discount at the moment (LPC21 as well as SAM7) There are also low cost carrier boards...

No, I am not kidding. A COG is quite similar to an ARM, and you will have MUCH more expandibility that way...

It will not be more expensive....

hippy

Large Memory Model (LMM) is well worth looking into. While it's slower than pure PASM execution and brings some hurdles to deal with it allows access to 8K of PASM instructions.

Take it a step further and a Cog can execute PASM from I2C Eeprom or SD Card. Using other Cogs to do caching could make that quite efficient in many cases.

Ariba

I think a 2 Cog version with 8kByte HubRAM in a 18 or 20 Pin package would be great. With the silicon process of the Prop II, the die will be only 1/8 or so of the existing Propeller and that results in a price of perhaps 2$ to 3$.
Such a chip can still do TV or VGA (_Text) output, has 4 32Bit Counters, needs no interrupt system ....

I would prefere such a chip to every other Low cost processer. And I'm shure it would dramatically increase the propagation of the propeller chip.

Andy

stevenmess2004

Ariba, a chip like that would be really nice for things like smart sensors. One core for the sensor and another for communicating with the host. Maybe it can be the PropIII

Bill Henning

Skinny dip-28 version of current prop in new process, 64KB eeprom buried in the package... pinout: P0-P21 = pin 1..22, TxD, RxD, OSCx2, Vcc/Vss... at $2-$3 it would destroy the market for many other microcontrollers. Better yet, also put a 5MHz xtal in the package, then we can have P0-P23,TxD,RxD,Vcc,Vss... droool

Ariba said...
I think a 2 Cog version with 8kByte HubRAM in a 18 or 20 Pin package would be great. With the silicon process of the Prop II, the die will be only 1/8 or so of the existing Propeller and that results in a price of perhaps 2$ to 3$.
Such a chip can still do TV or VGA (_Text) output, has 4 32Bit Counters, needs no interrupt system ....

I would prefere such a chip to every other Low cost processer. And I'm shure it would dramatically increase the propagation of the propeller chip.

Andy

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
www.mikronauts.com - a new blog about microcontrollers

stevenmess2004

Somehow I think that would up the price a bit but it would be great Even put a regulator on it (big enough to drive some sensors) and it would be perfect.

Baggers

Bill, Considering the current prop is $12.95 alone, without eeprom etc, and translates to ~£12 ($24) here in UK, you're $2-$3 price tag is a tad under what it would cost for all those bits.
Yes it would destroy the market, and also Parralax, as the losses would be enormous, don't you think?

Yes everyone wants things cheaper, but you also have to be realistic.

Gadgetman

I think this was discussed in an earlier thread, too...

A sngle or 2COG prop is too limited if you ask me...

A 4-COG, with 16 I/O should fit in a 24pin DIP, though.
Of course, the HUB access needs to be the same as on the 8-COG version so that timing is consistent.

but the chance of getting anything like this?
Very, very close to ZERO.
Yes, Parallax wants to give us what the customers want, but they also need to make money on it, and having to stock more versions of the same chip, which sells to the same markets isn't a good idea.

And it probably wouldn't be much cheaper, either.
Yes, they can reuse most of the design from the original, but it'd need a new round of testing, and the production costs will be about the same anyway.

Besides, Chip is busy designing the next generation Propeller chip, and I'd much rather have him work on that...

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Don't visit my new website...

Graham Stabler

I like your idea Andy, it would be a handy chip for many small applications. Chip made it clear they are going towards bigger rather than smaller however but isn't something like that more of a cut and paste chip design?

Graham

stevenmess2004

It already almost exists and is called the spinStamp and you even get a full 8 cogs

RinksCustoms

Personally, I don't think there's anything wrong with a single chip with 8 cores running @80MHz, is able to serialise 1-4bit video/s-video, vga, or broadcast, while running a realtime program. Bang for the buck, I don't feel it can be beat for the ease of use of the SPIN language. Maybe for a "serial" programmer it's too many cores, but for a hobbyist, this thing is great! Spin having a seroius lack of "code jumping" in the IDE makes debugging a snap. I've been involved with embedded systems a whole 2-3 yrs now, and I'm doing things with the prop I could only dream about a few yrs ago. Being an amateur EE (took two yrs before jumping into auto collision - stupid me!), I remeber when it would take me a week or two to just get a few seven segments to work right, with the BS2 or Prop i can have the same circuits up & running in 20 min!

And why is it, with a chip this capable, are we constantly, CONSTANTLY, trying to cram a Battleship, into a mooring made for a dingy when we have the elbow room for a battleship?! I must be missing something..

Basically what I'm saying here is that the prop is a great chip & being an American, bigger is better!! Bring on the Prop gen II!! They did a VERY good job with the chip, so what if you have a few COG's taking a nap...

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
E³ = Thought

http://folding.stanford.edu/·- Donating some CPU/GPU downtime just might lead to a cure for cancer! My team stats.