As it stands now, the Prop 1 is sufficiently low power when in its lowest state. I get about 4 to 7 micro amps when choosing a single cog with the slow internal RC oscillator, and waiting for a PINEQ.
With a 2 cog device, that power would not drop appreciably, but it's good enough now as it is.
Peter, my comments were concerning a respun 8-cog Prop I using the proposed Prop II fab process, which would draw more current in the standby mode.
Mike, is there an estimate on what the leakage current would be for the Prop I with the smaller geometry of the Prop II?· Would it be possible to run this at a very low voltage to reduce current draw?
Leon said...
I just can't see the point of using an expensive device when something a lot more suitable is available at a fraction of the price!
IMO, if a Prop is to compete in the SX / PIC / AVR market place, it has to match what those and similar chips offer to be considered a contender -
* Comparable pricing
* Single chip solution
* On-chip Eeprom and/or Flash
* Eeprom/Flash overwrite protection
* Zero-time boot ( not 1.5 seconds )
* On-chip full-speed osc with +/-2% manufacturing accuracy
* Wide operating range 1V8 to 5V5
* On-chip multi-channel 10-bit ADC ( not s/w sigma-delta )
Not all of that will be needed by everyone at all times, and yes some may be achieved in some other way, particularly by adding external components, and the Prop has its own unique advantages but that's to look at it the wrong way round. With all that it's a 'contender for all projects'; drop things off and you reduce its universal appeal.
Tweaking the Prop, adding or lopping I/O or Cogs, reducing price extends the Prop market, may fit particular replacement needs, but there's more to it in creating something which can compete with ubiquitous 'utility microcontrollers'.
Dave,
No. Chip did discuss leakage for the current (Prop I) process vs. the new (Prop II) process, but it was all in terms of the Prop II as I recall. With the thinner oxide, the Prop II (or a Prop I redone in the same process) would have to be run at a much lower voltage anyway (1.8V) to avoid damaging the oxide gate insulators. You can't run it at at the higher voltage (3.3V). The leakage is even worse there.
The I/O pins are built with thicker oxide and larger (area) features so they can run at 3.3V (and handle the large currents involved).
Hippy, if Parallax wanted to make a chip with those features, they would probably have been better off just buying them from a company that makes chips with those features. Inasmuch as the combination is even possible (does anyone even make a cpu that can run from 1v8 to 5v5?)
The Propeller exists because Parllax couldn't get a chip from anyone with the features THEY wanted for the next generation of basic stamp like devices -- a device with many general purpose equivalent I/O lines. The I/O capability and low power operation determined the process which determined that on-chip EEPROM was out. The use of serial EEPROM minimizes the use of pins for getting the program out of a really cheap device into the prop where it can be used, in a way that leaves the two pins used to do it available for other purposes once the boot is complete.
I doubt it is possible to fab an oscillator entirely in silicon that will hold to 2% over a full operating temperature range.
The absence of ADC is a point, but at what bit depth and sample rate? For my purposes 10 bits is a waste of silicon, I need a minimum of 16 but I don't need more than 1 khz or so sample rate. For decent audio sampling you need 16 bits at 44 KHz. For video you only need 8 bits but you need megahertz. This is why multiple solutions exist. One advantage of the delta/sigma scheme, if it weren't so picky about stray inductance, is that you can trade bit depth for sample speed in software.
Then there's the whole cogs, hub ram, and creating random peripherals in software which is just ... elegant. Sure, there are other chips that can do things the prop can't do; if you need their features use one of them. But when I want to put together a serial terminal a local printer output, ps/2 keypad, some parallel I/O's, and four VGA display outputs, if you can find another vendor who can do that with a single part for under $10 I'll be really impressed.
Watch it Holly ... someone might start begging to have you banned from the forums
The 1284p is a nice part with big Flash and SRAM. I use 32a and 8515 sometimes.
localroger said...
Hippy, if Parallax wanted to make a chip with those features, they would probably have been better off just buying them from a company that makes chips with those features. Inasmuch as the combination is even possible (does anyone even make a cpu that can run from 1v8 to 5v5?) ... I doubt it is possible to fab an oscillator entirely in silicon that will hold to 2% over a full operating temperature range.
Microchip have 1V8 to 5V5 operation, +/-2% for their internal oscillator across 0C to 85C, for mainstream devices.
I'm not criticising the Prop: What the Propeller is and what it offers isn't really the point here though, sure it can stand on its own two feet, but we are talking about a hacked about Prop in a role of an SX replacement. So the measure has to be how well the Prop matches as a replacement against how well some other alternative matches as a replacement.
Say we have a cheap cut-down Prop, 2V7-3V6 only, requires off-chip Eeprom, no code security, has slow boot time, needs an external crystal, external ADC or s/w sigma-delta - just how much of a realistic and viable "replacement for the SX" is that really ?
Maybe okay for some, but how would it compete against a non-Prop "SX replacement" which is a better fit ?
If the LMM is about as fast as PASM is, one could totally do video, BTW. The higher resolutions would be off the table, as would higher colors, but a simple tile map, like the Parallax Reference Driver does would run nicely.
Is there any reason why the waitvid instruction can't be used in tandem with LMM? One would have to consider instruction timing to make sure a lot got done between waitvids, and there would be a pause at the end, waiting for the waitvid to sync up for the next frame, but that happens now anyway in PASM.
Done that way, most of the driver could exist outside of a COG. That means being able to pack a few things into a COG, with it being able to do one of those things at a time modal style, or perhaps a few of them at a time with threading, depending on what was on the table.
Seems to me, if simple video output is used, like is done in the older 8x8 drivers, and some of Cardboardgurus early work, the video driver doesn't have to work that hard, leaving lots of time between waitvids. Smaller frame sizes, like for 8x8 text, are a problem, but larger ones like the Parallax driver are much less so. 256x200 pixels would be possible, bitmap or tiles.
A monochrome bitmap, say 480x200 would work too, as that could have a nice, large frame size. That's 12K for a buffer. If resolution isn't needed, something like 160x100 in 4 color would only be 10 waitvids per scan line. If it's two color, then that's only 5. That means being able to do stuff like video, and other lower level, or perhaps supervisory tasks while maintaining a display.
It's all about the resolution and the color and tile expectations. If video needed to be done, then that video cog would be the supervisor cog, with it's video kernel written to provide the display, with a lot of that driver not even having to be in the COG, allowing for various services in that COG, including draw primitives done during the blank to eliminate double buffering.
If we see a two cog Prop, with rapid HUB access, we will find it can do a lot more than we think. Honestly, I'm kind of jazzed about it now [noparse]:)[/noparse]
Did anyone mention RAM in that configuration? Would it still be the same 32K HUB / ROM split? If so, I think that would kind of rock and would be a serious trade-off from the Propeller we have now. Worth doing, IMHO.
Edit: I'm still mulling over what faster HUB access really means. One thing it means is being able to fetch a whole scanline, or a lot of it during a blank. If that's done, with an unrolled LMM video loop, it's gonna run nice. 40 character text displays and such are probably doable. A lot of the challenges with video is HUB data fetch related.
Well Leon, everybody knows there are a lot of cool chips out there, and a lot of variations. Seems to me the point here is to muse over what a Propeller variation gets us. Given we sort of like Propellers, that's where the value in the discussion is, right?
It just won't happen, though, given the options that are available off the shelf now, with more performance and lower cost. The market for the proposed device will be very limited.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Leon Heller
Amateur radio callsign: G1HSM
So then, the discussion isn't about absolute share, but about margin and sales. If a prop variation saw success in a clear niche or two, that would make a big impact to Parallax, and those using the chips. Comparing that impact to the bigger players won't look all that sexy, but who really cares?
Ask Apple computer about that idea. They recently took over the top spot from Microsoft, with shares that analysts said were not viable for years. "Apple will die" over and over, because they didn't see Apple taking significant share.
What every one of those people failed to realize is that the viability of an enterprise is margin, not share.
On that basis, a Propeller variation like this is worth discussion.
Leon said...
On my recommendation one SX user switched to a 16-bit PIC24 - it was cheaper, the tools were much better, and it offered a lot more performance.
Leon, I'm sure Parallax is very appreciative of your efforts to move people to the PIC.··· Parallax has gone through great pains to make sure that SX chips will be available for some time -- maybe as long a 2 years or longer.· The idea of an SX replacement was discussed in length in the EOL thread ( http://forums.parallax.com/showthread.php?p=827942·) and a thread with the title "Should Parallax sell an SX replacement?" ( http://forums.parallax.com/showthread.php?p=829412·).· I believe the conclusion was that this market segment is extremely competative, and it may not be worth the effort to develop a low-cost chip that could survive in such a market.
As I said before, the Propeller is a unique chip with it's own market niche.· The Prop II has the potention of competing in the video graphics and DSP market, where the profit margins should be somewhat higher.· However, I still think that an 8-cog Prop I redone with the Prop II fab technology would be a useful product.· It would have a smaller die size, and could be packaged in smaller DIP and SMT packages.· Parallax's development time on such a chip would be minimal (i.e., a low upfront investment).· As Mike Green said, it would not work for low power applications, but the orignal Prop I could continue to be used.
Dave
@potatohead, LMM and PASM might be pretty close in performance if the HUB window was adjusted by the number of COGs. Unfortunately there is still the matter of the LMM macros that are required for jmp,call,load and other kernel features. One *could* get away with doing add/sub on the PC for jmp, but calculating those all by hand is a PITA. Giving up any HUB RAM at all is a hard sell for me especially with LMM, but I might survive ....
Having a "product line" is a positive consideration for single sourcing and the perception of commitment to longer term production. If adding products were free or easily funded by demand, that would be easier to do of course. That sounds rather catch-22-ish though. I guess all we can do is be as positive as possible even in the face of some reservations ....
On my recommendation one SX user switched to a 16-bit PIC24 - it was cheaper, the tools were much better, and it offered a lot more performance.
Leon, I'm sure Parallax is very appreciative of your efforts to move people to the PIC. Parallax has gone through great pains to make sure that SX chips will be available for some time -- maybe as long a 2 years or longer. The idea of an SX replacement was discussed in length in the EOL thread ( http://forums.parallax.com/showthread.php?p=827942 ) and a thread with the title "Should Parallax sell an SX replacement?" ( http://forums.parallax.com/showthread.php?p=829412 ). I believe the conclusion was that this market segment is extremely competative, and it may not be worth the effort to develop a low-cost chip that could survive in such a market.
As I said before, the Propeller is a unique chip with it's own market niche. The Prop II has the potention of competing in the video graphics and DSP market, where the profit margins should be somewhat higher. However, I still think that an 8-cog Prop I redone with the Prop II fab technology would be a useful product. It would have a smaller die size, and could be packaged in smaller DIP and SMT packages. Parallax's development time on such a chip would be minimal (i.e., a low upfront investment). As Mike Green said, it would not work for low power applications, but the orignal Prop I could continue to be used.
Dave
I was discussing the viability of a two-cog version, not an upgraded Prop I, as an SX replacement.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Leon Heller
Amateur radio callsign: G1HSM
potatohead said...
If the LMM is about as fast as PASM is, one could totally do video, BTW.
I'm not sure where the "LMM will be faster in the miniProp" idea came from. There are LMMs that are either not (e.g. reverse LMM) or barely bound by hub waits, so a faster hub access will not help. In any event, the LMM emulation loop itself precludes speeds approaching native PASM.
Phil of course is correct in that the LMM can be run at 16 cycle loops meaning LMM is 4 times slower than PASM plus a margin for the jmps which require an extra fetch (16 cycles). There are other areas that will benefit, such as Spin, video, etc. However, you are not going to do much with 2 cogs!!! One cog will be your program and you are left with 1 cog to do all the peripherals. The RAM/ROM would be the same 32KB each. ROM takes so much less realestate - look at the die photo.
Leon: Yes I know there are other chips that will do certain things better than the prop. The prop can do a lot of things that a large number of variations can do with the same prop. BUT, you do not have to continue to PUSH other chips so blatantly!!! And then to take credit when you achieve that is worse!!! I have said it on other threads - please leave this forum!!!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Links to other interesting threads:
Mea culpa on the wide voltage bandwidth chips being available. I've got to wonder what the tradeoff was for that. I suspect the same one that forbids the Prop to have EEPROM -- the process chosen for other more important reasons.
OTOH when I see the 2-cog prop being proposed as an SX replacement, I don't see it being suggested as a replacement for the SX in all its possible roles -- I see it as a replacement for the SX as it was being used by Parallax, which would be like specifying a replacement for the PIC not as a general chip but as Parallax uses it to make Basic Stamps. I think much about the Propeller was driven by the I/O pins. Parallax has that focus where other manufacturers have different focus. A mini chip with the same core and I/O philosophy as the P1 probably does have a place at the table.
Personally, I think two cores is too few. Four would be great, but apparently it won't fit in an SDIP 28. Argh. I could really have used a part like that.
I think we should let Chip concentrate on the Propeller 2 chip and not bother him
with Prop 1 variants. We all need more I/O and ram and the TurboProp will
fix this.
Even though your quantity requirements are quite high, the Propeller 2 will
solve so many more peoples problems.
Here are some options while trying to keep design work to a minimum. I am not sure if any of this is even possible, but here it is anyway.
Prop 1.5X 64 I/O + 64KB Hub ROM + 64KB Hub ROM
64 I/O (i.e. the new Prop 1.5)
32KB hub RAM $0000-$7FFF
32KB hub ROM $8000-$FFFF
32KB hub ROM $10000-$17FFF (new)
32KB hub RAM $18000-$1FFFF (new)
Why this memory layout? To keep compatibility with existing software and the interpreter. It would be nice (but not required) for a cog to have an extra option to swap the new hub RAM into the existing ROM location to allow the interpreter to use 64KB RAM (or maybe just 60KB to keep the ROM interpreter visible).
Of course the die would be larger so it would be a little more expensive. Maybe $12 for a Prop 1.5X. I would pay that for the extra I/O, RAM & ROM.
Here is a picture of the proposed die using the Prop 1 as a reference.
Prop 0.4X·4 Cogs + 20-32 I/O + 8KB Hub ROM + 16KB Hub RAM This layout reduces the Prop 1 die width by almost 50% while height remains the same. The cogs are reduced to 4. However, ROM is reduced to 8KB,·so no character font in ROM and hub RAM is reduced in half to 16KB.·Another cog could be replaced with 8KB hub RAM. While this would achieve a die cost reduction by almost 50%, that does not translate to anywhere near a 50% price reduction due to packaging and handling. My best guess would be 20% at best.
Here is a picture of the proposed die.
Prop 0.2X 2 Cogs + 20-32 I/O + 32KB Hub ROM + 32KB Hub RAM Here is a picture of the proposed die.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔ Links to other interesting threads:
>>> OTOH when I see the 2-cog prop being proposed as an SX replacement, I don't see it being suggested as a replacement for the SX in all its possible roles -- I see it as a replacement for the SX as it was being used by Parallax, which would be like specifying a replacement for the PIC not as a general chip but as Parallax uses it to make Basic Stamps. <<<
I don't think this is the case. If it were, Parallax would be busy porting the BS2 series to the Propeller. As it were, Ken mentioned in another thread that they ordered like 20 years worth of SX chips for the BS2 line.
The thing is, when the Propeller was released, it could afford to be different, because it wasn't the only product being offered by Parallax. They had the BS2 line, as well as the SX, which were & probably are, the main revenue generators. However, with the EOL of the SX line, there is now a hole in their line-up.
What I think Parallax needs to do is develop a real SX replacement. In other words, not a Propeller. It needs to be of a more conventional architecture, with a mix of features that would make it applicable to many of the applications the SX is currently used for. And the sole reason for them to do this is simply to control the IP. Eventually, the 20-year supply of SX is going to run out, or the non-Parallax replacement will again be discontinued, etc. But the Propeller really can't compete in this space, because it simply wasn't designed to do so.
Now, as far as an SX replacement... The Propeller was introduced with nobody outside of Parallax really knowing anything about it. So it had none of the design input that the Propeller II has had from these forums. I think that Parallax could very easily solicit input as to what people would like feature-wise in the design, and then simply build it. It could then be used for the BS2 line, as well as any similar line of future products, offered for sale standalone, and not compete directly with the Propeller, like the SX currently doesn't compete.
Kevin, they aren't porting BS2 to the Propeller because the propeller was meant to be the next step beyond the BS2. All the SX ever was was a faster, better PIC, so it was a natural next step for a design that was already based on PIC in the BS1. But there's a limit to how far that kind of improvement will get you. Going beyond a 100 MHz 8-bit chip either requires dealing with crazy high frequencies or widening the architecture.
Scenix made a strategic decision to horn in on Microchip's market by making their product object code compatible, but that's a risky decision even if it doesn't get you sued. Microchip always had the option, should the market open up enough, to get the memo and make a faster better chip of their own. A small company can't compete effectively with a big one doing effectively the same thing because the small company doesn't have the economies of scale. But a larger company like Ubicom might not feel the need to grab a ready-made slice of market via object code compatibility with another product whose owner will sue you at the drop of a pin. So they targeted a new market which doesn't need object code PIC compatibility.
Parallax seems to have done pretty much what I suggested; instead of cloning their own better PIC / SX, they just laid in a supply from the other people who did it for them. By the time Parallax uses up its inventory of SX's it's very likely that Microchip or someone else will have something as good or even better. Meanwhile, Parallax is going in a direction no one else seems interested in going, which is good for Parallax because it gives them breathing room and it's good for us because they're making something nobody else is bothering to make.
Leon said...
Microchip already has something better, as I've mentioned a couple of times!
Better than what? Better for wich application?
And why do you keep on trying to recruit·people to join the harvard's jedi force, when so many of us are perfectly happy with the von-neumann's dark side?!?
That as well, but in my last missive I meant better than the SX, in response to this:
localroger said...
Parallax seems to have done pretty much what I suggested; instead of cloning their own better PIC / SX, they just laid in a supply from the other people who did it for them. By the time Parallax uses up its inventory of SX's it's very likely that Microchip or someone else will have something as good or even better. Meanwhile, Parallax is going in a direction no one else seems interested in going, which is good for Parallax because it gives them breathing room and it's good for us because they're making something nobody else is bothering to make.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Leon Heller
Amateur radio callsign: G1HSM
Ok, but by this logic I could as well say that an LPC2101 is even better in nearly all applications, just because it's 70MHz, cost $1.50 and has GCC tools?!?
Given a specific application that maps just right onto one of the hundred of variants that Microchip, Atmel and others have in catalog, is obvious that Leon is right!
But the SX as it was, a GENERIC, simple and FAST processor where you could time multiplex your software defined functions, has only one alternative I would consider: the SiLabs C8051F36x. Maybe a small market segment, but not so crowded after all.
I think the whole propeller concept has many new things compared to SX (video shifters, etc.), but basically it carry on the SX tradition of VPs, just making it much easier by having them already partitioned in COGs.
In some other thread about "how C fits on Propeller", one person defined the propeller "the poor man's FPGA"... if I was Chip I would feel really HONORED by such definition
Exactly, I don't see what is so "poor" about it. It's brilliant. A certain other company is basing it's whole business on multi-core micro-controllers with deterministic timing and I/O closely coupled to the cores. They coined the term "Software Defined Silicon" and push it as an easy to use and flexible alternative to FPGAs in many situations.
The Prop II will be continuing in this spirit and giving said (actually unsaid) company a run for it's money. Excellent.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
Comments
As it stands now, the Prop 1 is sufficiently low power when in its lowest state. I get about 4 to 7 micro amps when choosing a single cog with the slow internal RC oscillator, and waiting for a PINEQ.
With a 2 cog device, that power would not drop appreciably, but it's good enough now as it is.
Cheers,
Peter (pjv)
Mike, is there an estimate on what the leakage current would be for the Prop I with the smaller geometry of the Prop II?· Would it be possible to run this at a very low voltage to reduce current draw?
Dave
IMO, if a Prop is to compete in the SX / PIC / AVR market place, it has to match what those and similar chips offer to be considered a contender -
* Comparable pricing
* Single chip solution
* On-chip Eeprom and/or Flash
* Eeprom/Flash overwrite protection
* Zero-time boot ( not 1.5 seconds )
* On-chip full-speed osc with +/-2% manufacturing accuracy
* Wide operating range 1V8 to 5V5
* On-chip multi-channel 10-bit ADC ( not s/w sigma-delta )
Not all of that will be needed by everyone at all times, and yes some may be achieved in some other way, particularly by adding external components, and the Prop has its own unique advantages but that's to look at it the wrong way round. With all that it's a 'contender for all projects'; drop things off and you reduce its universal appeal.
Tweaking the Prop, adding or lopping I/O or Cogs, reducing price extends the Prop market, may fit particular replacement needs, but there's more to it in creating something which can compete with ubiquitous 'utility microcontrollers'.
No. Chip did discuss leakage for the current (Prop I) process vs. the new (Prop II) process, but it was all in terms of the Prop II as I recall. With the thinner oxide, the Prop II (or a Prop I redone in the same process) would have to be run at a much lower voltage anyway (1.8V) to avoid damaging the oxide gate insulators. You can't run it at at the higher voltage (3.3V). The leakage is even worse there.
The I/O pins are built with thicker oxide and larger (area) features so they can run at 3.3V (and handle the large currents involved).
The Propeller exists because Parllax couldn't get a chip from anyone with the features THEY wanted for the next generation of basic stamp like devices -- a device with many general purpose equivalent I/O lines. The I/O capability and low power operation determined the process which determined that on-chip EEPROM was out. The use of serial EEPROM minimizes the use of pins for getting the program out of a really cheap device into the prop where it can be used, in a way that leaves the two pins used to do it available for other purposes once the boot is complete.
I doubt it is possible to fab an oscillator entirely in silicon that will hold to 2% over a full operating temperature range.
The absence of ADC is a point, but at what bit depth and sample rate? For my purposes 10 bits is a waste of silicon, I need a minimum of 16 but I don't need more than 1 khz or so sample rate. For decent audio sampling you need 16 bits at 44 KHz. For video you only need 8 bits but you need megahertz. This is why multiple solutions exist. One advantage of the delta/sigma scheme, if it weren't so picky about stray inductance, is that you can trade bit depth for sample speed in software.
Then there's the whole cogs, hub ram, and creating random peripherals in software which is just ... elegant. Sure, there are other chips that can do things the prop can't do; if you need their features use one of them. But when I want to put together a serial terminal a local printer output, ps/2 keypad, some parallel I/O's, and four VGA display outputs, if you can find another vendor who can do that with a single part for under $10 I'll be really impressed.
From the ATmega1284p datasheet
The 1284p is a nice part with big Flash and SRAM. I use 32a and 8515 sometimes.
Cheers,
--Steve
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Propeller Pages: Propeller JVM
Microchip have 1V8 to 5V5 operation, +/-2% for their internal oscillator across 0C to 85C, for mainstream devices.
I'm not criticising the Prop: What the Propeller is and what it offers isn't really the point here though, sure it can stand on its own two feet, but we are talking about a hacked about Prop in a role of an SX replacement. So the measure has to be how well the Prop matches as a replacement against how well some other alternative matches as a replacement.
Say we have a cheap cut-down Prop, 2V7-3V6 only, requires off-chip Eeprom, no code security, has slow boot time, needs an external crystal, external ADC or s/w sigma-delta - just how much of a realistic and viable "replacement for the SX" is that really ?
Maybe okay for some, but how would it compete against a non-Prop "SX replacement" which is a better fit ?
Is there any reason why the waitvid instruction can't be used in tandem with LMM? One would have to consider instruction timing to make sure a lot got done between waitvids, and there would be a pause at the end, waiting for the waitvid to sync up for the next frame, but that happens now anyway in PASM.
Done that way, most of the driver could exist outside of a COG. That means being able to pack a few things into a COG, with it being able to do one of those things at a time modal style, or perhaps a few of them at a time with threading, depending on what was on the table.
Seems to me, if simple video output is used, like is done in the older 8x8 drivers, and some of Cardboardgurus early work, the video driver doesn't have to work that hard, leaving lots of time between waitvids. Smaller frame sizes, like for 8x8 text, are a problem, but larger ones like the Parallax driver are much less so. 256x200 pixels would be possible, bitmap or tiles.
A monochrome bitmap, say 480x200 would work too, as that could have a nice, large frame size. That's 12K for a buffer. If resolution isn't needed, something like 160x100 in 4 color would only be 10 waitvids per scan line. If it's two color, then that's only 5. That means being able to do stuff like video, and other lower level, or perhaps supervisory tasks while maintaining a display.
It's all about the resolution and the color and tile expectations. If video needed to be done, then that video cog would be the supervisor cog, with it's video kernel written to provide the display, with a lot of that driver not even having to be in the COG, allowing for various services in that COG, including draw primitives done during the blank to eliminate double buffering.
If we see a two cog Prop, with rapid HUB access, we will find it can do a lot more than we think. Honestly, I'm kind of jazzed about it now [noparse]:)[/noparse]
Did anyone mention RAM in that configuration? Would it still be the same 32K HUB / ROM split? If so, I think that would kind of rock and would be a serious trade-off from the Propeller we have now. Worth doing, IMHO.
Edit: I'm still mulling over what faster HUB access really means. One thing it means is being able to fetch a whole scanline, or a lot of it during a blank. If that's done, with an unrolled LMM video loop, it's gonna run nice. 40 character text displays and such are probably doable. A lot of the challenges with video is HUB data fetch related.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Propeller Wiki: Share the coolness!
8x8 color 80 Column NTSC Text Object
Wondering how to set tile colors in the graphics_demo.spin?
Safety Tip: Life is as good as YOU think it is!
Post Edited (potatohead) : 7/3/2010 7:30:56 PM GMT
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Propeller Wiki: Share the coolness!
8x8 color 80 Column NTSC Text Object
Wondering how to set tile colors in the graphics_demo.spin?
Safety Tip: Life is as good as YOU think it is!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Leon Heller
Amateur radio callsign: G1HSM
Post Edited (Leon) : 7/3/2010 7:36:30 PM GMT
Fricking huge, that's how large it is.
So then, the discussion isn't about absolute share, but about margin and sales. If a prop variation saw success in a clear niche or two, that would make a big impact to Parallax, and those using the chips. Comparing that impact to the bigger players won't look all that sexy, but who really cares?
Ask Apple computer about that idea. They recently took over the top spot from Microsoft, with shares that analysts said were not viable for years. "Apple will die" over and over, because they didn't see Apple taking significant share.
What every one of those people failed to realize is that the viability of an enterprise is margin, not share.
On that basis, a Propeller variation like this is worth discussion.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Propeller Wiki: Share the coolness!
8x8 color 80 Column NTSC Text Object
Wondering how to set tile colors in the graphics_demo.spin?
Safety Tip: Life is as good as YOU think it is!
As I said before, the Propeller is a unique chip with it's own market niche.· The Prop II has the potention of competing in the video graphics and DSP market, where the profit margins should be somewhat higher.· However, I still think that an 8-cog Prop I redone with the Prop II fab technology would be a useful product.· It would have a smaller die size, and could be packaged in smaller DIP and SMT packages.· Parallax's development time on such a chip would be minimal (i.e., a low upfront investment).· As Mike Green said, it would not work for low power applications, but the orignal Prop I could continue to be used.
Dave
Having a "product line" is a positive consideration for single sourcing and the perception of commitment to longer term production. If adding products were free or easily funded by demand, that would be easier to do of course. That sounds rather catch-22-ish though. I guess all we can do is be as positive as possible even in the face of some reservations ....
Cheers [noparse]:)[/noparse],
--Steve
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Propeller Pages: Propeller JVM
I was discussing the viability of a two-cog version, not an upgraded Prop I, as an SX replacement.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Leon Heller
Amateur radio callsign: G1HSM
Post Edited (Leon) : 7/3/2010 8:18:14 PM GMT
-Phil
I had completely forgotten about that, being fixated on what faster HUB access would do. Doh!!
Well, most of my post still applies, but for the part with a lot of the driver not being in the COG. Oh well.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Propeller Wiki: Share the coolness!
8x8 color 80 Column NTSC Text Object
Wondering how to set tile colors in the graphics_demo.spin?
Safety Tip: Life is as good as YOU think it is!
Leon: Yes I know there are other chips that will do certain things better than the prop. The prop can do a lot of things that a large number of variations can do with the same prop. BUT, you do not have to continue to PUSH other chips so blatantly!!! And then to take credit when you achieve that is worse!!! I have said it on other threads - please leave this forum!!!
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)·
· Prop OS: SphinxOS·, PropDos , PropCmd··· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBlade Props: www.cluso.bluemagic.biz
OTOH when I see the 2-cog prop being proposed as an SX replacement, I don't see it being suggested as a replacement for the SX in all its possible roles -- I see it as a replacement for the SX as it was being used by Parallax, which would be like specifying a replacement for the PIC not as a general chip but as Parallax uses it to make Basic Stamps. I think much about the Propeller was driven by the I/O pins. Parallax has that focus where other manufacturers have different focus. A mini chip with the same core and I/O philosophy as the P1 probably does have a place at the table.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
www.mikronauts.com E-mail: mikronauts _at_ gmail _dot_ com
My products: Morpheus / Mem+ / PropCade / FlexMem / VMCOG / Propteus / Proteus / SerPlug
and 6.250MHz Crystals to run Propellers at 100MHz & 5.0" OEM TFT VGA LCD modules
Las - Large model assembler Largos - upcoming nano operating system
with Prop 1 variants. We all need more I/O and ram and the TurboProp will
fix this.
Even though your quantity requirements are quite high, the Propeller 2 will
solve so many more peoples problems.
Russ
Here are some options while trying to keep design work to a minimum. I am not sure if any of this is even possible, but here it is anyway.
Prop 1.5X 64 I/O + 64KB Hub ROM + 64KB Hub ROM
Why this memory layout? To keep compatibility with existing software and the interpreter. It would be nice (but not required) for a cog to have an extra option to swap the new hub RAM into the existing ROM location to allow the interpreter to use 64KB RAM (or maybe just 60KB to keep the ROM interpreter visible).
Of course the die would be larger so it would be a little more expensive. Maybe $12 for a Prop 1.5X. I would pay that for the extra I/O, RAM & ROM.
Here is a picture of the proposed die using the Prop 1 as a reference.
Prop 0.4X·4 Cogs + 20-32 I/O + 8KB Hub ROM + 16KB Hub RAM
This layout reduces the Prop 1 die width by almost 50% while height remains the same. The cogs are reduced to 4. However, ROM is reduced to 8KB,·so no character font in ROM and hub RAM is reduced in half to 16KB.·Another cog could be replaced with 8KB hub RAM. While this would achieve a die cost reduction by almost 50%, that does not translate to anywhere near a 50% price reduction due to packaging and handling. My best guess would be 20% at best.
Here is a picture of the proposed die.
Prop 0.2X 2 Cogs + 20-32 I/O + 32KB Hub ROM + 32KB Hub RAM
Here is a picture of the proposed die.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBlade,·RamBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: CPUs Z80 etc; Micros Altair etc;· Terminals·VT100 etc; (Index) ZiCog (Z80) , MoCog (6809)·
· Prop OS: SphinxOS·, PropDos , PropCmd··· Search the Propeller forums·(uses advanced Google search)
My cruising website is: ·www.bluemagic.biz·· MultiBlade Props: www.cluso.bluemagic.biz
Post Edited (Cluso99) : 7/4/2010 6:07:28 AM GMT
I don't think this is the case. If it were, Parallax would be busy porting the BS2 series to the Propeller. As it were, Ken mentioned in another thread that they ordered like 20 years worth of SX chips for the BS2 line.
The thing is, when the Propeller was released, it could afford to be different, because it wasn't the only product being offered by Parallax. They had the BS2 line, as well as the SX, which were & probably are, the main revenue generators. However, with the EOL of the SX line, there is now a hole in their line-up.
What I think Parallax needs to do is develop a real SX replacement. In other words, not a Propeller. It needs to be of a more conventional architecture, with a mix of features that would make it applicable to many of the applications the SX is currently used for. And the sole reason for them to do this is simply to control the IP. Eventually, the 20-year supply of SX is going to run out, or the non-Parallax replacement will again be discontinued, etc. But the Propeller really can't compete in this space, because it simply wasn't designed to do so.
Now, as far as an SX replacement... The Propeller was introduced with nobody outside of Parallax really knowing anything about it. So it had none of the design input that the Propeller II has had from these forums. I think that Parallax could very easily solicit input as to what people would like feature-wise in the design, and then simply build it. It could then be used for the BS2 line, as well as any similar line of future products, offered for sale standalone, and not compete directly with the Propeller, like the SX currently doesn't compete.
Scenix made a strategic decision to horn in on Microchip's market by making their product object code compatible, but that's a risky decision even if it doesn't get you sued. Microchip always had the option, should the market open up enough, to get the memo and make a faster better chip of their own. A small company can't compete effectively with a big one doing effectively the same thing because the small company doesn't have the economies of scale. But a larger company like Ubicom might not feel the need to grab a ready-made slice of market via object code compatibility with another product whose owner will sue you at the drop of a pin. So they targeted a new market which doesn't need object code PIC compatibility.
Parallax seems to have done pretty much what I suggested; instead of cloning their own better PIC / SX, they just laid in a supply from the other people who did it for them. By the time Parallax uses up its inventory of SX's it's very likely that Microchip or someone else will have something as good or even better. Meanwhile, Parallax is going in a direction no one else seems interested in going, which is good for Parallax because it gives them breathing room and it's good for us because they're making something nobody else is bothering to make.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Leon Heller
Amateur radio callsign: G1HSM
And why do you keep on trying to recruit·people to join the harvard's jedi force, when so many of us are perfectly happy with the von-neumann's dark side?!?
Come on, join the dark side... we have cookies!
·
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Leon Heller
Amateur radio callsign: G1HSM
Post Edited (Leon) : 7/4/2010 3:53:27 PM GMT
Given a specific application that maps just right onto one of the hundred of variants that Microchip, Atmel and others have in catalog, is obvious that Leon is right!
But the SX as it was, a GENERIC, simple and FAST processor where you could time multiplex your software defined functions, has only one alternative I would consider: the SiLabs C8051F36x. Maybe a small market segment, but not so crowded after all.
I think the whole propeller concept has many new things compared to SX (video shifters, etc.), but basically it carry on the SX tradition of VPs, just making it much easier by having them already partitioned in COGs.
In some other thread about "how C fits on Propeller", one person defined the propeller "the poor man's FPGA"... if I was Chip I would feel really HONORED by such definition
Exactly, I don't see what is so "poor" about it. It's brilliant. A certain other company is basing it's whole business on multi-core micro-controllers with deterministic timing and I/O closely coupled to the cores. They coined the term "Software Defined Silicon" and push it as an easy to use and flexible alternative to FPGAs in many situations.
The Prop II will be continuing in this spirit and giving said (actually unsaid) company a run for it's money. Excellent.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.