Yes, I was not really arguing with your main point - just highlighting the fact that while the Prop can't do everything, it's so extraordinarily flexible that it is a cost-effective solution across a huge range of applications.
I don't think there is another chip on the market that comes so close to being a panacea.
Ross.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Catalina - a FREE C compiler for the Propeller - see Catalina
Some times it appears that Chip is an evil genius that is playing with us. It's as if he purposely set the limits of the Propellers speed and memory capacity at such a levels as to make common things doable but only just. Tantalizingly in the realm of "it must be possible". He wanted to drive us all mad attempting to do those things.
USB - Only just doable.
Z80 or other emulation only just.
Ethernet - Only just, bit banging UDP.
C language support - Only just.
Floating point - Only just.
I'm sure others have many more examples to support my thesis.
Chip's evil and twisted mind is not in a hurry to put us out of our torment with the Prop II. He's enjoying watching us squirm to much. Notice how he intentionally did not release the 64 pin Prop I.
Err..OK. Only joking.
Still you must admit that those Prop limits have spawned a lot of interesting solutions to all kinds of problems over the years. If the limits were lower there would be no chance and no one would bother. If they were a bit higher it would be boringly easy.
I nominate Bill's invention of LMM as the most brilliant of those solutions to work around the Prop's limits.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
Mike Green said...
I'm really struck how much "mainstream" computer science has moved away from what has been a very rich area. Sophisticated people used to develop their own programming languages or language extensions for a project.
I've been out of the academic loop for a long time now, so I'm not familiar with today's curricula. But if, as you imply, university undergrad comp-sci programs are just teaching how to program in C or Java, say, they've been reduced to little more than trade schools. Not that there's anything wrong with trade schools -- far from it; but being proficient in one or two languages does not, by itself, merit a college degree.
It doesn't matter what the processor is or how powerful. There will always be apps for which that processor will be almost -- or only just -- capable. It's our nature to probe those limits. But I'd have to say that doing so with the Prop has been more fun than with any other MPU I've used. I think it's because there always seems to be one more wrinkle to explore and with which to eke out yet another iota of performance.
Phil: "It's our nature to probe those limits" - Yep
Phil: "doing so with the Prop has been more fun than with any other MPU..." - Yep.
Phil: "...one more wrinkle to explore..." - Hope so.
I think the magic is that there are, of course, limits and that Prop has such an unorthodox architecture to let you get around those limits. But that unorthodox architecture forces you to think in somewhat different ways to do it. If it were any old boring run of the mill design one would not bother to push the limits one would just go out and get a different chip.
Bill's LMM is a case in point. That technique is just not applicable to most other chips with variable length instructions for example. But here we are, after that little bit of sideways thinking with two C compilers, some CPU emulators and whatever else on the Prop using it that looked stupid to attempt before.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
heater said...
Some times it appears that Chip is an evil genius that is playing with us.
He is a genius, of that I'm certain. Before I actually got out in the working
world and saw first hand how important making a profit was I would have said
that the prop should have lots more features. But I now understand that
Chip had to make it a realistic product that would have a market.
no profit would mean we would have no propeller at all to play with :-(
This amazing chip is so wonderful that I don't think I will ever discover
everything there is to know about it. Just the joy of not having to write
interrupt code is reason enough to use the prop whenever you possibly
can. I almost decided against using the propeller. I have a mild case of
OCD and if something seems to be not exactly the way it should be then it
is hard for me to use it. I considered the lack of on board program storage
to be "just not right" but the 8 cogs and no interrupts kept making me look
again and again at the prop. It's pretty funny now but I got around this at
first by gluing the eeprom chip on top of a dip prop and wiring it up along
with a resistor so that it would be a complete package..lol .. this made me
comfortable using it and then I moved on gradually to where I could consider
the prop and it's necessary external eeprom chip to be a complete package
of parts that was a lot like a mcu with flash program storage built in. Pretty
weird huh.
@Sal Ammoniac
What you are doing, combining a prop with an ARM chip is a perfect example
of using the prop to best advantage.
I guess the thing I would like the most for the prop is some more cog ram!
I keep running out of ram and having to go back and rewrite code to cram
everything in. Would be nice to have enough room to store a lot more data
where you could access and alter it quickly.
Holly: OCD or not I think many people have viewed the Prop the same way. Including myself. When you are searching around for a micro-controller to do some job and you have seen many micro-controllers before you have some fixed ideas about what you are looking for. As I have said here before I skipped over the Prop on seeing it in a catalog the first few times. "To small" I thought, "You cant' do anything in 496 instructions" I thought, "to weird" I thought.
Eventually it starts to dawn on you...
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
localroger said...
And the best part is that there's not other vendor offering any comparable part (Leon and his XMOS fetish notwithstanding) so if anybody tries to rip off what we do, it will be kind of obvious what they did.
It's horses for courses - XMOS is more suitable from some things than the Propeller, and vice versa. For instance, the Propeller can't do high-speed USB and Ethernet on the same chip, and it's harder to do VGA on the XMOS chip than on the Propeller.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Leon Heller
Amateur radio callsign: G1HSM
I was fortunate to not have known about the Prop until greater minds had got past all those initial dismissals. I wanted more ram for an ethernet project, I wanted a chip that could do VGA, I wanted ...
Then I saw the YBOX2 and that was that.
Mike is right about the stubborness in resisting new languages, unfortunately I caught that one back with the Z80!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point
Comments
Yes, I was not really arguing with your main point - just highlighting the fact that while the Prop can't do everything, it's so extraordinarily flexible that it is a cost-effective solution across a huge range of applications.
I don't think there is another chip on the market that comes so close to being a panacea.
Ross.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Catalina - a FREE C compiler for the Propeller - see Catalina
USB - Only just doable.
Z80 or other emulation only just.
Ethernet - Only just, bit banging UDP.
C language support - Only just.
Floating point - Only just.
I'm sure others have many more examples to support my thesis.
Chip's evil and twisted mind is not in a hurry to put us out of our torment with the Prop II. He's enjoying watching us squirm to much. Notice how he intentionally did not release the 64 pin Prop I.
Err..OK. Only joking.
Still you must admit that those Prop limits have spawned a lot of interesting solutions to all kinds of problems over the years. If the limits were lower there would be no chance and no one would bother. If they were a bit higher it would be boringly easy.
I nominate Bill's invention of LMM as the most brilliant of those solutions to work around the Prop's limits.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
-Phil
It doesn't matter what the processor is or how powerful. There will always be apps for which that processor will be almost -- or only just -- capable. It's our nature to probe those limits. But I'd have to say that doing so with the Prop has been more fun than with any other MPU I've used. I think it's because there always seems to be one more wrinkle to explore and with which to eke out yet another iota of performance.
-Phil
Phil: "doing so with the Prop has been more fun than with any other MPU..." - Yep.
Phil: "...one more wrinkle to explore..." - Hope so.
I think the magic is that there are, of course, limits and that Prop has such an unorthodox architecture to let you get around those limits. But that unorthodox architecture forces you to think in somewhat different ways to do it. If it were any old boring run of the mill design one would not bother to push the limits one would just go out and get a different chip.
Bill's LMM is a case in point. That technique is just not applicable to most other chips with variable length instructions for example. But here we are, after that little bit of sideways thinking with two C compilers, some CPU emulators and whatever else on the Prop using it that looked stupid to attempt before.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
He is a genius, of that I'm certain. Before I actually got out in the working
world and saw first hand how important making a profit was I would have said
that the prop should have lots more features. But I now understand that
Chip had to make it a realistic product that would have a market.
no profit would mean we would have no propeller at all to play with :-(
This amazing chip is so wonderful that I don't think I will ever discover
everything there is to know about it. Just the joy of not having to write
interrupt code is reason enough to use the prop whenever you possibly
can. I almost decided against using the propeller. I have a mild case of
OCD and if something seems to be not exactly the way it should be then it
is hard for me to use it. I considered the lack of on board program storage
to be "just not right" but the 8 cogs and no interrupts kept making me look
again and again at the prop. It's pretty funny now but I got around this at
first by gluing the eeprom chip on top of a dip prop and wiring it up along
with a resistor so that it would be a complete package..lol .. this made me
comfortable using it and then I moved on gradually to where I could consider
the prop and it's necessary external eeprom chip to be a complete package
of parts that was a lot like a mcu with flash program storage built in. Pretty
weird huh.
@Sal Ammoniac
What you are doing, combining a prop with an ARM chip is a perfect example
of using the prop to best advantage.
I guess the thing I would like the most for the prop is some more cog ram!
I keep running out of ram and having to go back and rewrite code to cram
everything in. Would be nice to have enough room to store a lot more data
where you could access and alter it quickly.
Eventually it starts to dawn on you...
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
It's horses for courses - XMOS is more suitable from some things than the Propeller, and vice versa. For instance, the Propeller can't do high-speed USB and Ethernet on the same chip, and it's harder to do VGA on the XMOS chip than on the Propeller.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Leon Heller
Amateur radio callsign: G1HSM
Post Edited (Leon) : 5/28/2010 7:34:39 AM GMT
Then I saw the YBOX2 and that was that.
Mike is right about the stubborness in resisting new languages, unfortunately I caught that one back with the Z80!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point