NXP produces multicore (dual Cortex M0 & M4) processor
Cluso99
Posts: 18,069
I just noticed this article http://www.techfocusmedia.net/archives/articles/20110119-nxp/
I couldn't resist posting a comment about the prop there
I couldn't resist posting a comment about the prop there
Comments
Its only a matter of time before they produce something similar with more cores.....
I really hope Prop II can gain a foothold in the marketplace before they do.
Regards,
Coley
http://www.nxp.com/news/content/file_1701.html
I'm working on a little 14-DIL PCB for them that will plug into a breadboard or prototyping board.
And your comment has already been challenged I see.
Exactly, that's the beauty of the Prop, it has 8 blank CPU's that can be moulded into whatever peripherals you want in software.
My Prop projects have 7 I/O processors. I adopt the model that one COG is my "main program" and the rest is peripherals.
I don't have the patience to get an account there and reply, perhaps Cluso will.
Now having said that, one of my dream chips would have a 1GHz ARM hooked up to megabytes of RAM and FLASH with all it's Ethernet and USB etc interfaces. It would then have an on board Propeller to wiggle 64 or 96 I/O pins at my command:)
Hmm...Do you think Parallax could licence some ARM core IP and put an ARM core in one corner of the Prop II or III:)
I've got a 1 GHz ARM (Qualcomm Snapdragon) in my Dell Streak tablet, and Qualcomm has chips with two 1 GHz ARM cores. The Snapdragon includes all the usual interfaces like GPS, Ethernet, 3G, graphics, etc. They are working on a 2 GHz version.
I would add that I'm a hobbyist because I don't yet know how to market what I've created. I have more ideas and it is only a matter of time before I learn to successfully reach my future consumers.
Would it not make more sense to 'wiggle' the pins on the ARM chip not to mention somewhat cheaper as well!
With the new NXP 2mm ARM that Leon mentioned its only a matter of time before we see multi core (more than 2) ARM devices.
Ross.
Leon: Here's your chance to tell them about your favorite chip too!
No. That question indicates that you have overlooked some of the key features of the Prop as well.
With the Prop I get:
1) Eight cores all running with deterministic timing.
2) Cores have their own RAM and round robin access to shared RAM. They can run in total isolation for speed or access shared RAM with no timing contention jitter.
2) Eight cores means not having to use interrupts.
3) Not having interrupts means I get to use the deterministic timing without worrying how tasks disturb each other timing wise.
4) Result: I can wiggle the pins to drive esoteric hardware, promptly respond to external events, or generate weird signals all very easily.
With the ARM I would be bogged down in interrupts and timing issues trying to do anything complicated on the I/O core.
Or course if they produced an 8 core ARM with each core having it's own RAM that might change the game a bit. But they would still be far away from the real-time handling available with the Prop and dare I say it XMOS.
I don't believe the likes of ARM are interested in moving into the real-time space of the Prop and XMOS. ARM has the general purpose mobile CPU market sewn up already.
And I don't believe Parallax and XMOS want to move into the general pupose CPU arena. The Prop II will make this gulf between general purpose and real-time embedded silicon even wider.
Conclusion: So far no ARM is any threat to the Prop (or XMOS).
Nice summary - why not post it on the techfocusmedia site?
Ross.
I'm in no mood for signing up and logging into yet another forum.
Perhaps someone who is registered there would like to cut and paste or paraphrase my summary over there.
The ARM Cortex architecture offers fully deterministic instruction and interrupt timing!
Also, code can be written in C whereas the Propeller needs assembler to make use of its performance and deterministic features.
Excellent. I don't have much a clue about the ARM instruction set despite having used them for numerous embedded products. Isn't C great:)
Still, that's only part of the equation as you know. Deterministic timing for multiple activities mandates multiple processors (or hardware scheduling of multiple deterministic threads), the non-use of interrupts, tight coupling to the I/O to the CPU, your own memory to play in etc. Features that the Prop and the XMOS have but current ARMS do not.
Of course, given a fast enough CPU and memory you don't need any of that for some real-time jobs. But for any-given level of technology it always wins.
Thing is I compare a Prop and an XMOS, and other multi-core mcu's to FPGA's (or even discreet logic) rather than tradition compute platforms. What we are doing is replacing FPGA logic blocks with whole processors.
ARM is no threat to that market.
Re: MS and ARM.
Aside:
Years ago the company I worked for had a brilliant idea for an embedded product and started on developing its software using Wince. That just did not work. Just as they were about to give up on the idea I was called into the team and we save the whole project by migrating to VxWorks. The later generations of that product were migrated to Linux.
MS is late to the party. They now see the world swimming in mobile devices based on Android or Symbian or whatever. They see that the traditional PC as we know it will be extinct fairly soon. They want part of that action in a growing market.
There is no point trying though, they can't beat "free".
Aside:
For many years I have thought that those big box PC you see in the stores are pointlessly huge, badly made and invariably ugly. Not to mention noisy. How is it that these dinosaurs have survived into 2011?
I have here this Mac mini. Really tiny (do not forget the power brick this model has) besides me. It does more than the big ugly beige box at work. two times more . I fail to see why would someone today buy a big box machine, it just makes no sense from a space point of view, but they are cheaper. These minis lost they appeal through the years... from 500 € to > 700...
Yes indeed, built down to a price, and it shows. I guess I never thought about it much, never having bought a PC, they just come free from work and get used no matter how hideous they are. But then I'm odd, I've never bought a T.V. or VCR or microwave either.
BUT times they are a changin'. My Android phone has more memory and horse power than the PCs I was using a short while back. I'm seriously looking at the idea of using a credit card sized Beagle Board or such as a replacement for that PC megalith I'm typing this on.
M.S. wants to go with me. Good luck.
though because we can use allot of the code that we have written for the Prop I. Also if Chip figured
correctly the Prop II should be around $5 less.
We plan on selling 1000 to 2000 of these terminals a year so this cost difference is substantial.
Heater, the irony of this statement in light of your defense of the Propeller against the ARM is the perfect counterpoint to your argument.
I'm not sure that any of my statements can be taken as "defense of the Propeller against the ARM".
In my mind they are almost not comparable, apples and oranges, they are built with different purposes in mind.
ARM is a general purpose CPU, you know, for building computers.
Prop is a micro-controller, a replacement for logic or FPGA's,
I have said here from time to time that an ideal match would be an ARM and it's memory on a board along with a Prop or two. The ARM handles the big application code, the Prop(s) handle the serious deterministic real-time stuff. In fact the ARM on such a board would be running Linux and provide the development environment for the Prop(s):)
Just recently I refined that idea to, "lets have an ARM core on the Prop III". Best of both worlds on a single chip.
As I said previously, ARM Cortex chips offer fully deterministic I/O and interrupt handling. There isn't much point in using them with Propeller devices.
I can't see the Propeller replacing FPGAs. XMOS chips were designed for that sort of thing, though.