I completely concur with the subject line here. Stop debating new features and get a 16 core faster PII out the door that has enough RAM to do bigger jobs.
Now is a relative term, assuming Parallax and their guru corps skip a lot of testing and just release it with Spin and Pasm maybe by the end of the year.
It won't be something to take to Design West, that's for sure.
The way I see it, is nobody is discussing anything about what is getting done what way, as Chip has rightfully removed himself from the forum, so he can think clearly and concentrate on getting what he needs to do done. then if he so desires to look into the discussions of how to implement, then that's up to him.
Don't forget people are wanting it fully functional as soon as, yet he still has to finalise the instruction set, do it, then add whatever features he wants to add, be it cordic or float or whatever, then he also has to redo PNut, then he also has to create the new SPIN language, and write the compiler and decoder.
That's quite a major workload for one guy alone! so if you guys keep chatting too much, when he comes back in about two weeks with his huge workload done or almost done, he'll have to sift through a shedload of banter.
It's bad enough just missing a single day on here, let alone couple of weeks! the more that's gonna be on here, if it was me, the more I'd want to skip! otherwise it'd take him another two weeks to catch up. lol
Don't forget people are wanting it fully functional as soon as, yet he still has to finalise the instruction set, do it, then add whatever features he wants to add, be it cordic or float or whatever, then he also has to redo PNut, then he also has to create the new SPIN language, and write the compiler and decoder.
Chip did post the proposed instruction set, around 200 instructions iirc. Yes Cordic will have a slightly different loading and retrieving mechanism, but that should be easy to cut across to the final scheme. Video will be slightly different too.
In short there's not much stopping us from developing using what we have already. Stick to the reduced instruction set and It'll be more similar than different.
Yeah, you say that, but if the video doesn't have the composite, which it probably won't then for the likes of potatohead who prefers using composite, as he has composite input on his laptop, so mobile developing for prop is useful for when he's mobile, and saves carrying a VGA monitor with him to play when working away.
I also use composite, quite a bit, more-so than VGA but I'm not as restricted as potatohead, the driver will have to be totally re-written. and the code for colours, as it won't support 24bit RGB like P2 does at the moment.
But I do agree a lot of what the P2 emulation we have can be pretty code compatible or pretty close to what the final chip will be like, it also runs at 80MIPS, so won't have to redo the timing on stuff either.
Heater, I know he does, it's just taking the time to sift through it all, in one day this forum covers a lot of ground at the moment, let alone a couple of weeks.
Yes I need composite, too. I'm not worried there will be some way to do it, because you can always bitbang something at 80 MIPs. Not as elegant, but once an object exists we can run with it.
So I guess thinking about in-cog sprites is on hold for now, but we can work on hub bitmapped based stuff with confidence in the meantime.
The DE0 still constrains things a bit, but if we get 4 cogs onto it we can look at how to get sdram based images across to a video cog via the hub. A test like this might sort out whether we need any kind of hub slot sharing/mooching
I think the real question is whether or not WAITVID keeps the modes, or just ends up driving the DACS. The color management and pixel engine are beastly. Deffo gone, but the modes maybe not!
Re: mobile
I may get to use "home base" a bit more in the future.
The thing with the modes, is that it output RGB values, that will be gone for sure as it was expensive, like the pixel engine.
Basically if we can have some way of having 4 bytes cycle to the DACs we should be able to do some form of pixel bit banger to the DAC by having the colour burst for said pixel in the 4 bytes. it would then just be a case of figuring out desired palettes.
I have to laugh without reading this whole thread. Maybe the impatient people should pony up the investment capital necessary to produce a chip, then perhaps Chip may throw something together to satisfy your urges.
As for me, I am prepared to wait for Chip's masterpiece
I have to laugh without reading this whole thread...
I'm not sure why you are laughing. Six or eight years development and 4 million dollars investment is not a laughing matter.
Perhaps, possibly, maybe, if it were not for a silly failure on the last shuttle run "Chip's masterpiece", as it was a year ago, it would be in our hands already.
As it is, yes, we will wait. Things are looking good again.
I have to laugh without reading this whole thread. Maybe the impatient people should pony up the investment capital necessary to produce a chip, then perhaps Chip may throw something together to satisfy your urges.
As for me, I am prepared to wait for Chip's masterpiece
I don't believe Parallax would accept any investment capital in the form of cash anyway. They will happily accept continuing "sweat capital" contributions however.
I'm not sure why you are laughing. Six or eight years development and 4 million dollars investment is not a laughing matter.
You are correct, $4,000,000 and years of development is no laughing matter, but you are pulling my comment way out of context. I am laughing because of the people trying to put pressure on Chip and Parallax, when they have no monetary investment in the development of the chip. If the new chip fails to be successful, it will be a huge monetary loss for Parallax. So why should they rush it out the door?
You are correct, $4,000,000 and years of development is no laughing matter, but you are pulling my comment way out of context. I am laughing because of the people trying to put pressure on Chip and Parallax, when they have no monetary investment in the development of the chip. If the new chip fails to be successful, it will be a huge monetary loss for Parallax. So why should they rush it out the door?
They shouldn't rush it out the door but neither should they tinker with it for another 6-8 years. :-)
They shouldn't rush it out the door but neither should they tinker with it for another 6-8 years. :-)
Personally, I would like to have a chance to toy with the new chip before I die However I believe that Chip should tinker with it until he is completely satisfied with his creation.
I haven't given up hope, but I am certainly not waiting around for this chip to instantly pop on the market. In the meantime, I would really like to see Parallax do more with the original Propeller, because there are all kinds of useful boards they could create.
Personally, I would like to have a chance to toy with the new chip before I die However I believe that Chip should tinker with it until he is completely satisfied with his creation.
I haven't given up hope, but I am certainly not waiting around for this chip to instantly pop on the market. In the meantime, I would really like to see Parallax do more with the original Propeller, because there are all kinds of useful boards they could create.
I'm curious about what new kinds of Propeller boards you envision. It seems to me that there are already many Propeller boards maybe even hundreds but certainly dozens.
Of course I have my own interests, so my suggestions will be highly biased
There are many possible they could go, but I see the Propeller as highly suitable for CNC and Robotics. I don't even think Parallax has scratched the surface with the Propeller and robot control, because there are so many possible variants. The same goes for CNC and controlling multiple motors of different varieties. Everyone tinkers with the Prop in their own little way, but I see it as a machinery workhorse. VGA, sound, etc.... (YAWN)
There are many possible they could go, but I see the Propeller as highly suitable for CNC and Robotics. I don't even think Parallax has scratched the surface with the Propeller and robot control, because there are so many possible variants. The same goes for CNC and controlling multiple motors of different varieties. Everyone tinkers with the Prop in their own little way, but I see it as a machinery workhorse. VGA, sound, etc.... (YAWN)
While many have done lots of great things with VGA and Sound, apparently not many volume users use video according to Ken.
But, things like CNC, ROBOTICS, xxxCOPTERS, etc and all the uses where parallel cores make the whole programming simpler and modular are precisely where the prop should shine. Way too little has been done in this arena. But then comes the question about what sort of board(s) do you build. That is the real question.
The same goes for CNC and controlling multiple motors of different varieties. Everyone tinkers with the Prop in their own little way, but I see it as a machinery workhorse. VGA, sound, etc.... (YAWN)
I don't see the P1 (or any of its following versions) as a CNC controller. But maybe that is because I do have a CNC mill and a CNC lathe. And I suppose, you didn't even dream that the Prop will do the VGA too as a side job.
The mill has a Atom 800 on the CPU-board, the lathe has an ARM.
While many have done lots of great things with VGA and Sound, apparently not many volume users use video according to Ken.
But, things like CNC, ROBOTICS, xxxCOPTERS, etc and all the uses where parallel cores make the whole programming simpler and modular are precisely where the prop should shine. Way too little has been done in this arena. But then comes the question about what sort of board(s) do you build. That is the real question.
Per your recent POLL, and the ongoing mega-thread in the making, it seems like the few who want video to be the primary use case of the Prop are willing to overlook your far more useful suggestion.
Which is a real, real shame, as having 16 32K Cores at 100 MIPs and 128K Hub really would be a useful product in the marketplace.
While many have done lots of great things with VGA and Sound, apparently not many volume users use video according to Ken.
LOL... Sorry Ray, my comment was not intended to distract from or degrade the great things accomplished by other users such as yourself and many others, it just does not interest me that much. To be perfectly honest, I could not do the things you guys do, even if I was interested. I am impressed by you guys, but just not enough interest.
But then comes the question about what sort of board(s) do you build. That is the real question.
The answer to that question lies in market research. However, I have a hunch that a 3D printer controller would be a good place to start.
And how does this pertain to your previous comment?
Maybe because I know what:
* Inner and outer loops are
* What feedback is
* What a trajectory planner is
* What G-code is
* What a tool-offset table is for
* What tool compensation is
* How big CNC-files can get
* How many IOs it takes even for a simple machine
To put that discussion into perspective, you might explain what your personal CNC background is and what you'd expect from a CNC controller.
LOL... Come on... Be real.
So no visual feedback? I have the feeling that you talk about a motor controller, not a CNC controller.
Maybe because I know what:
* Inner and outer loops are
* What feedback is
* What a trajectory planner is
* What G-code is
* What a tool-offset table is for
* What tool compensation is
* How big CNC-files can get
* How many IOs it takes even for a simple machine
That still does not explain:
I don't see the P1 (or any of its following versions) as a CNC controller.
So you honestly don't believe the Propeller can handle forementioned tasks, especially if assisted by a PC or other device?
I have the feeling that you talk about a motor controller, not a CNC controller.
I am simply talking about a chip that can parallel process and control IO pins, what else is there? These pins contol motors, vavles, solenoinds, electo-magnets, lasers, sensors, switches, etc.... The list goes on. If one Propeller does not provide enough pins, just add another.
As for all the items you bring up, I believe they all can be handled one way or another, by an on-board PC which lacks IO pins for device and machine control, but can obviously talk to a Propeller which has IO pins.
One PC motherboard and 1,2, or 3 Propellers could control a very large machine that requires a lot of input and outputs.
You may have your toys and the experience over me, but you are not backing up your statement very well in my opinion.
That still does not explain: <snip> So you honestly don't believe the Propeller can handle forementioned tasks, especially if assisted by a PC or other device?
Yes. That's because I do understand how these things relate to each other.
As soon as you add a PC, you no longer have a CNC controller, but some subsystem that does a fraction of what is involved.
Done that. I have two Propellers running ModBus on my mill. One for the pendant, the other one for IO. The later, I wouldn't do again.
You may have your toys and the experience over me, but you are not backing up your statement very well in my opinion.
How can I backup my experience against someone who seems to have none but knows better?
Here are my "toys" http://motor-manufaktur.de/shop/index.html
I replaced the complete control of the mill with LinuxCNC, added my own electronics.
So what is your background? What makes you think a Propeller can be a CNC controller and why all the controller manufacturers do use multi core processors and not micro-controllers?
Comments
Ship it.
But I don't think new features are being discussed. What's being discussed is how those 16 cores get access to that bigger RAM.
If that access mechanism is different to the current round-robin HUB access scheme then it is indeed a new feature.
It won't be something to take to Design West, that's for sure.
Don't forget people are wanting it fully functional as soon as, yet he still has to finalise the instruction set, do it, then add whatever features he wants to add, be it cordic or float or whatever, then he also has to redo PNut, then he also has to create the new SPIN language, and write the compiler and decoder.
That's quite a major workload for one guy alone! so if you guys keep chatting too much, when he comes back in about two weeks with his huge workload done or almost done, he'll have to sift through a shedload of banter.
It's bad enough just missing a single day on here, let alone couple of weeks! the more that's gonna be on here, if it was me, the more I'd want to skip! otherwise it'd take him another two weeks to catch up. lol
Chip did post the proposed instruction set, around 200 instructions iirc. Yes Cordic will have a slightly different loading and retrieving mechanism, but that should be easy to cut across to the final scheme. Video will be slightly different too.
In short there's not much stopping us from developing using what we have already. Stick to the reduced instruction set and It'll be more similar than different.
I also use composite, quite a bit, more-so than VGA but I'm not as restricted as potatohead, the driver will have to be totally re-written. and the code for colours, as it won't support 24bit RGB like P2 does at the moment.
But I do agree a lot of what the P2 emulation we have can be pretty code compatible or pretty close to what the final chip will be like, it also runs at 80MIPS, so won't have to redo the timing on stuff either.
Heater, I know he does, it's just taking the time to sift through it all, in one day this forum covers a lot of ground at the moment, let alone a couple of weeks.
So I guess thinking about in-cog sprites is on hold for now, but we can work on hub bitmapped based stuff with confidence in the meantime.
The DE0 still constrains things a bit, but if we get 4 cogs onto it we can look at how to get sdram based images across to a video cog via the hub. A test like this might sort out whether we need any kind of hub slot sharing/mooching
Re: mobile
I may get to use "home base" a bit more in the future.
Basically if we can have some way of having 4 bytes cycle to the DACs we should be able to do some form of pixel bit banger to the DAC by having the colour burst for said pixel in the 4 bytes. it would then just be a case of figuring out desired palettes.
As for me, I am prepared to wait for Chip's masterpiece
@Bruce, it's a two way street. Concerns about time are valid all arround.
Perhaps, possibly, maybe, if it were not for a silly failure on the last shuttle run "Chip's masterpiece", as it was a year ago, it would be in our hands already.
As it is, yes, we will wait. Things are looking good again.
You are correct, $4,000,000 and years of development is no laughing matter, but you are pulling my comment way out of context. I am laughing because of the people trying to put pressure on Chip and Parallax, when they have no monetary investment in the development of the chip. If the new chip fails to be successful, it will be a huge monetary loss for Parallax. So why should they rush it out the door?
Personally, I would like to have a chance to toy with the new chip before I die
I haven't given up hope, but I am certainly not waiting around for this chip to instantly pop on the market. In the meantime, I would really like to see Parallax do more with the original Propeller, because there are all kinds of useful boards they could create.
Of course I have my own interests, so my suggestions will be highly biased
There are many possible they could go, but I see the Propeller as highly suitable for CNC and Robotics. I don't even think Parallax has scratched the surface with the Propeller and robot control, because there are so many possible variants. The same goes for CNC and controlling multiple motors of different varieties. Everyone tinkers with the Prop in their own little way, but I see it as a machinery workhorse. VGA, sound, etc.... (YAWN)
But, things like CNC, ROBOTICS, xxxCOPTERS, etc and all the uses where parallel cores make the whole programming simpler and modular are precisely where the prop should shine. Way too little has been done in this arena. But then comes the question about what sort of board(s) do you build. That is the real question.
I don't see the P1 (or any of its following versions) as a CNC controller. But maybe that is because I do have a CNC mill and a CNC lathe. And I suppose, you didn't even dream that the Prop will do the VGA too as a side job.
The mill has a Atom 800 on the CPU-board, the lathe has an ARM.
Nick
Per your recent POLL, and the ongoing mega-thread in the making, it seems like the few who want video to be the primary use case of the Prop are willing to overlook your far more useful suggestion.
Which is a real, real shame, as having 16 32K Cores at 100 MIPs and 128K Hub really would be a useful product in the marketplace.
Why not? Please explain.
And how does this pertain to your previous comment?
LOL... Come on... Be real.
LOL... Sorry Ray, my comment was not intended to distract from or degrade the great things accomplished by other users such as yourself and many others, it just does not interest me that much. To be perfectly honest, I could not do the things you guys do, even if I was interested. I am impressed by you guys, but just not enough interest.
The answer to that question lies in market research. However, I have a hunch that a 3D printer controller would be a good place to start.
Maybe because I know what:
* Inner and outer loops are
* What feedback is
* What a trajectory planner is
* What G-code is
* What a tool-offset table is for
* What tool compensation is
* How big CNC-files can get
* How many IOs it takes even for a simple machine
To put that discussion into perspective, you might explain what your personal CNC background is and what you'd expect from a CNC controller.
So no visual feedback? I have the feeling that you talk about a motor controller, not a CNC controller.
Nick
That still does not explain:
So you honestly don't believe the Propeller can handle forementioned tasks, especially if assisted by a PC or other device?
I am simply talking about a chip that can parallel process and control IO pins, what else is there? These pins contol motors, vavles, solenoinds, electo-magnets, lasers, sensors, switches, etc.... The list goes on. If one Propeller does not provide enough pins, just add another.
As for all the items you bring up, I believe they all can be handled one way or another, by an on-board PC which lacks IO pins for device and machine control, but can obviously talk to a Propeller which has IO pins.
One PC motherboard and 1,2, or 3 Propellers could control a very large machine that requires a lot of input and outputs.
You may have your toys and the experience over me, but you are not backing up your statement very well in my opinion.
Yes. That's because I do understand how these things relate to each other.
As soon as you add a PC, you no longer have a CNC controller, but some subsystem that does a fraction of what is involved.
Done that. I have two Propellers running ModBus on my mill. One for the pendant, the other one for IO. The later, I wouldn't do again.
How can I backup my experience against someone who seems to have none but knows better?
Here are my "toys" http://motor-manufaktur.de/shop/index.html
I replaced the complete control of the mill with LinuxCNC, added my own electronics.
So what is your background? What makes you think a Propeller can be a CNC controller and why all the controller manufacturers do use multi core processors and not micro-controllers?
Nick
You are right, I have no experience or knowledge, I am just a dumb *****. You win.
On second thought, maybe I am not so dumb.... Gee.... I build my own CNCs from scratch.
http://forums.parallax.com/showthread.php/129612-My-Current-Prop-Based-CNC-Photos