We have always dreamed of some general purpose SERDES on the P2. And perhaps some configurable logic as appears on some PICs and other MCU's.
Sounds kind of tricky to me to come up with such configurable logic blocks that are of general use whilst actually being usable by normal people. I notice the PIC guys get around this by providing a GUI tool that generates the required code/configuration for such logic elements.
Single use hw is the wrong direction. Just the basic pieces to get decent speed is all that should be included. Software can do the rest.
And I'm all for having HDMI capabilities but the above goes for any HDMI assistance also. Any serialiser that can be used for HDMI must be a generic serialiser that can be re-purposed.
I know that this is the premise of the Propeller philosophy which has been much discussed over the years. Yet despite such a supposed strict philosophy, even the P1 had special video generation hardware which wasn't really suitable for other uses.
We don't have to worry about HDMI hardware because we know it won't be supported, but a parallel interface comes in handy for many things (and not just interfaces to HDMI/LVDS chips and LCDs).
And while I might agree for the most part that soft peripherals is a good way to go, I see USB as a major exception, at least where the processor simply is not fast enough for a software implementation. That's because of the ubiquity of USB that I mentioned earlier. Even with the P1, many of the common digital interfaces could be performed sufficiently so specialized hardware wasn't necessary, but proper USB was well beyond its capabilities. If the P2 manages to do USB in software, it'll probably be just barely, and that would be a shame, especially considering specialized hardware doesn't appear to be too costly.
You guys are all talking about features for the P3, right?
That's what I thought.....because how could folks be proposing changes and additions to a P2 when we really don't know all it REALLY will or won't have and it's coming up on another release milestone a little over one year after the latest re-re-re-design?
We zeroed in on what the majority feature set will be just prior to Chip going away to really get it done. I would be surprised to see much has changed. The team he's working with is likely to trigger some changes to optimize performance or make it possible to put on chip. We need those. And I'm quite sure the feature list got vetted and pruned as needed during this time too.
Truth is, I've enjoyed each iteration. The main thing is getting it done. We will put it to all sorts of uses, check box features or not.
I'm personally eager to see how the HUB memory plays out, and what about those Smart Pins? Good times ahead, when it gets done. We will have basically 16X HUB transfer potential! Insane! Can't wait.
We would commune in the awesome silence of the cosmic vibration. Who needs the internet and forums? I mean, like, we are all connected man. Can't you feel it?
Meanwhile..here on Earth...I'm still patiently waiting for Chipmas.
I don't know about USB being a mess. It works and pretty darn well. The only question is whether or not the P-2 can emulate it software, if it can't then it means the P-2 is dependent on 3rd party devices for USB support. Either way it all come out in the testing as to what is and is not possible.
Anyway, USB is a catastrophic mess, spawn of the devil. Well, Intel anyway. We should not go there.
And yet so major that it would nearly be a catastrophe not to support it properly. I maintain that a lack of proper Hi-speed USB support at even a minimal level (ULPI+ basic SIE) will preclude it from being used in consumer/prosumer products that it otherwise might have been extremely well suited for. While it's great that many of the member's industrial products might not need USB, or at least the high mark-up allows additional components to be used when it is necessary, these extremely low volume products do practically nothing for Parallax's bottom line (I'm not saying they don't still appreciate it) unlike high-volume customers which Parallax seems to be severely lacking (did they even hit the millionth unit milestone yet?). I just don't see how implementing the most significant interface standard hurts anyone as much as it could hurt Parallax not to implement it. Chip was confident that the P2 Hot would have been able to do USB, and that was great. Not being able to is a step backwards, and please don't compare it to loss of other wacky features such as perspective-correct texture mapping and what have you.
I know that this is the premise of the Propeller philosophy which has been much discussed over the years. Yet despite such a supposed strict philosophy, even the P1 had special video generation hardware which wasn't really suitable for other uses.
Except perhaps for the NTSC chroma generator, the P1's so-called "video" hardware is nothing more than a very fast generic serializer that has been used for many non-video apps, including my PWMx8 pulse-width modulator. So it doesn't really violate the "no app-specific hardware" philosophy.
Do we know this one won't do USB in software at the least?
It will have 16 cogs and huge hub transfer capability. Maybe it takes 2 cogs. Great!
We need to see what the product of the current iteration is before advocating for all sorts of stuff.
While you're certainly correct in that I'm basically jumping the gun, I think it's unlikely that Hi-Speed would be unlikely in software with a multi-cog solution being both tricky and a waste of resources.
@Phil
I know a few around here have made use of the video gen for other things, but they could have been a bit more general purpose still, which is why I said "wasn't really suitable" instead of "wasn't suitable at all"
Single use hw is the wrong direction. Just the basic pieces to get decent speed is all that should be included. Software can do the rest..
I agree - simple HW for things like Bit Stuff/Destuff, Bus State sense, and Tx/Rx shifting could easily be part of the SmartIO.
All byte-level upper layer stuff can be done in SW.
That leaves CRC as a possible HW support, and something like that could go into the shared-region like the Cordic and mathops.
That simple HW is also low risk, get your bit-stuff wrong and it simply will not work. Once it is right, you never need to touch it. Just like a UART state engine.
We are in agreement here. USB chips are a commodity and with our volume it makes no sense to design that into our P2.
Do you mean HS-USB or FS-USB ?
For HS-USB, some form of external support will always be needed.
That can be a ULPI, or you could use a FT232H in the Fast Serial mode, as a HS-PHY.
Fast Serial is a variant of UART handler, with a FSCLK and FSCTS added.
It is spec'd to 50MHz, so is simple, low pin cost, and has intermediate bandwidth ability.
FT232H also has Sync FIFO '245' mode, and again, in the Byte-Wide Streaming p2 has, the handshakes for that mode certainly should be included. This goes faster, but needs more pins.
FS USB could be assisted with some Smart Pin additions, but if Parallax want to do nothing at all on P2 for USB, then you will need to deliver libraries that work well with a chosen commodity chip. (and probably Prop-link libraries for that commodity MCU too, but use the Vendors libraries for the PC end )
Ken,
FYI (just in case you are not familiar with USB)
USB v2.0 has LS (low speed) which is 1.5Mbps, FS (full speed) 12Mbps, HS (480Mbps).
I have discussed with Chip the minor additions required to ensure we can get USB FS working on P2. This involves a simple instruction and perhaps the inclusion of a CRC instruction to aid in calculating the CRC. IIRC Chip agreed to add the simple instruction and look at the possible CRC as soon as he completed other work. I am sure he just needs to be reminded at the appropriate time. I certainly was not after a complex solution like some proposed.
It would be a shame not to have this minimal help hardware to get USB FS in software on the P2. We only require 3 pins and 3 resistors and a USB connector externally.
As for any higher speeds (USB 3.0 is higher again), an external solution will be required.
And now we also have USB-C here which is even faster and way more complex.
Phil & Others,
While we can use the video for other serial output purposes, that was not the original intention. As many of us have lamented, it's a shame it couldn't do serial in as well. It would be nice to have some form of serialiser/deserialiser in P2 and I expect this will be likely in the smart pins washup.
IMHO the Smart Pins, while sounding fantastic, are consuming a hefty time delay to the P2.
In additional to this, I have some concerns about the new hub format. Only time will tell. I do now it sounds good on paper.
I think Chip would need a good USB article and/or USB mentor to make USB a reality. That would probably fit more with P3 than P2.
However, successfully testing those small hardware bit manipulation blocks Cluso requested, on a P1V (as jmg points out), would really divide and advance the case.
Well, there is plenty of USB-related Verilog code with very permissive licenses on opencores.org such as this one http://opencores.org/project,usb (modified BSD license according to the dev's site). Plenty of USB1.1 stuff on there including PHYs too.
Hello all,
I just received an interesting email from Dr Cherng Chin, Chairman of the Taiwan Forth Interest Group.
It seems that the Taiwan FIG is very much interested in the Propeller 2 FPGA and is looking forward to attempting to use it with Forth as a 3D Printer micro-controller.
The main point here is that the FPGA Propeller 2 is optimal for actually getting software available for the eventual release of a hardware Propeller 2 chip. I suspect there are others out there beside the Taiwan Forth Interest Group that would like to have the FPGA version available so they can get into actually develop software ahead of time.
Lots of us stand ready to contribute. On that last one, we got a long way. C was up and running, we had some basic driver code done, docs were started, etc... wasn't bad at all. Had it not been hot, things would be rocking right now.
Given that taste of how it can go, I personally remain confident in all of it. That model will work well, and should be a lot of fun.
Comments
We have always dreamed of some general purpose SERDES on the P2. And perhaps some configurable logic as appears on some PICs and other MCU's.
Sounds kind of tricky to me to come up with such configurable logic blocks that are of general use whilst actually being usable by normal people. I notice the PIC guys get around this by providing a GUI tool that generates the required code/configuration for such logic elements.
I know that this is the premise of the Propeller philosophy which has been much discussed over the years. Yet despite such a supposed strict philosophy, even the P1 had special video generation hardware which wasn't really suitable for other uses.
We don't have to worry about HDMI hardware because we know it won't be supported, but a parallel interface comes in handy for many things (and not just interfaces to HDMI/LVDS chips and LCDs).
And while I might agree for the most part that soft peripherals is a good way to go, I see USB as a major exception, at least where the processor simply is not fast enough for a software implementation. That's because of the ubiquity of USB that I mentioned earlier. Even with the P1, many of the common digital interfaces could be performed sufficiently so specialized hardware wasn't necessary, but proper USB was well beyond its capabilities. If the P2 manages to do USB in software, it'll probably be just barely, and that would be a shame, especially considering specialized hardware doesn't appear to be too costly.
That's what I thought.....because how could folks be proposing changes and additions to a P2 when we really don't know all it REALLY will or won't have and it's coming up on another release milestone a little over one year after the latest re-re-re-design?
Seriously?
Yes indeed. The P2, get me it NOW!
Never mind that extra fluff.
We zeroed in on what the majority feature set will be just prior to Chip going away to really get it done. I would be surprised to see much has changed. The team he's working with is likely to trigger some changes to optimize performance or make it possible to put on chip. We need those. And I'm quite sure the feature list got vetted and pruned as needed during this time too.
Truth is, I've enjoyed each iteration. The main thing is getting it done. We will put it to all sorts of uses, check box features or not.
I'm personally eager to see how the HUB memory plays out, and what about those Smart Pins?
If only I could get you to agree on all that other stuff...:)
Anyway, USB is a catastrophic mess, spawn of the devil. Well, Intel anyway. We should not go there.
HDMI...much the same.
Meanwhile..here on Earth...I'm still patiently waiting for Chipmas.
And yet so major that it would nearly be a catastrophe not to support it properly. I maintain that a lack of proper Hi-speed USB support at even a minimal level (ULPI+ basic SIE) will preclude it from being used in consumer/prosumer products that it otherwise might have been extremely well suited for. While it's great that many of the member's industrial products might not need USB, or at least the high mark-up allows additional components to be used when it is necessary, these extremely low volume products do practically nothing for Parallax's bottom line (I'm not saying they don't still appreciate it) unlike high-volume customers which Parallax seems to be severely lacking (did they even hit the millionth unit milestone yet?). I just don't see how implementing the most significant interface standard hurts anyone as much as it could hurt Parallax not to implement it. Chip was confident that the P2 Hot would have been able to do USB, and that was great. Not being able to is a step backwards, and please don't compare it to loss of other wacky features such as perspective-correct texture mapping and what have you.
It will have 16 cogs and huge hub transfer capability. Maybe it takes 2 cogs. Great!
We need to see what the product of the current iteration is before advocating for all sorts of stuff.
YUP!!! Heck I would have been happy for a Prop-II with only the extra 32 IO pins that were rumored since the release of the Prop-I!!
Except perhaps for the NTSC chroma generator, the P1's so-called "video" hardware is nothing more than a very fast generic serializer that has been used for many non-video apps, including my PWMx8 pulse-width modulator. So it doesn't really violate the "no app-specific hardware" philosophy.
-Phil
While you're certainly correct in that I'm basically jumping the gun, I think it's unlikely that Hi-Speed would be unlikely in software with a multi-cog solution being both tricky and a waste of resources.
@Phil
I know a few around here have made use of the video gen for other things, but they could have been a bit more general purpose still, which is why I said "wasn't really suitable" instead of "wasn't suitable at all"
All byte-level upper layer stuff can be done in SW.
That leaves CRC as a possible HW support, and something like that could go into the shared-region like the Cordic and mathops.
That simple HW is also low risk, get your bit-stuff wrong and it simply will not work. Once it is right, you never need to touch it. Just like a UART state engine.
Do you mean HS-USB or FS-USB ?
For HS-USB, some form of external support will always be needed.
That can be a ULPI, or you could use a FT232H in the Fast Serial mode, as a HS-PHY.
Fast Serial is a variant of UART handler, with a FSCLK and FSCTS added.
It is spec'd to 50MHz, so is simple, low pin cost, and has intermediate bandwidth ability.
FT232H also has Sync FIFO '245' mode, and again, in the Byte-Wide Streaming p2 has, the handshakes for that mode certainly should be included. This goes faster, but needs more pins.
FS USB could be assisted with some Smart Pin additions, but if Parallax want to do nothing at all on P2 for USB, then you will need to deliver libraries that work well with a chosen commodity chip. (and probably Prop-link libraries for that commodity MCU too, but use the Vendors libraries for the PC end )
FYI (just in case you are not familiar with USB)
USB v2.0 has LS (low speed) which is 1.5Mbps, FS (full speed) 12Mbps, HS (480Mbps).
I have discussed with Chip the minor additions required to ensure we can get USB FS working on P2. This involves a simple instruction and perhaps the inclusion of a CRC instruction to aid in calculating the CRC. IIRC Chip agreed to add the simple instruction and look at the possible CRC as soon as he completed other work. I am sure he just needs to be reminded at the appropriate time. I certainly was not after a complex solution like some proposed.
It would be a shame not to have this minimal help hardware to get USB FS in software on the P2. We only require 3 pins and 3 resistors and a USB connector externally.
As for any higher speeds (USB 3.0 is higher again), an external solution will be required.
And now we also have USB-C here which is even faster and way more complex.
Phil & Others,
While we can use the video for other serial output purposes, that was not the original intention. As many of us have lamented, it's a shame it couldn't do serial in as well. It would be nice to have some form of serialiser/deserialiser in P2 and I expect this will be likely in the smart pins washup.
IMHO the Smart Pins, while sounding fantastic, are consuming a hefty time delay to the P2.
In additional to this, I have some concerns about the new hub format. Only time will tell. I do now it sounds good on paper.
Anyway, waiting for Chipmas like some others.
However, successfully testing those small hardware bit manipulation blocks Cluso requested, on a P1V (as jmg points out), would really divide and advance the case.
I just received an interesting email from Dr Cherng Chin, Chairman of the Taiwan Forth Interest Group.
It seems that the Taiwan FIG is very much interested in the Propeller 2 FPGA and is looking forward to attempting to use it with Forth as a 3D Printer micro-controller.
The main point here is that the FPGA Propeller 2 is optimal for actually getting software available for the eventual release of a hardware Propeller 2 chip. I suspect there are others out there beside the Taiwan Forth Interest Group that would like to have the FPGA version available so they can get into actually develop software ahead of time.
Lots of us stand ready to contribute. On that last one, we got a long way. C was up and running, we had some basic driver code done, docs were started, etc... wasn't bad at all. Had it not been hot, things would be rocking right now.
Given that taste of how it can go, I personally remain confident in all of it. That model will work well, and should be a lot of fun.