Wow, um, ah, awkward. At least she used an Arduino.
For better or worse, the Prop makes an appearance here too:
"Often in this situation Ill break out something like the Saxo board, with an FPGA and a thick USB 2.0 pipe. In this case, I was dealing with low enough data rates that I could do something even simpler. I plugged the remote into my Propeller Demo Board and wrote a quick program to capture the SPI traffic and send it back in ASCII over the serial port."
As Cluso said in post 22, our government is rolling out a very expensive fibre optic network, but this is being built by older people, for older people. My kids won't use it. They go to Grandma's and ask "what is your WiFi password?" Ditto their friends. They seem to know everyone's password. Everything is portable nowadays.
All very true. And of course, now there's 4G as well.
But, in the end, I do still think that Bill has the right general idea: Provide the basic (requisite) connections "free" out-of-the-box and let the rest be done with dongles etc. I can see where that might have strong appeal, as compared to other options, in many situations (and anyway, you can't give 'em everything they might want -- but you can go broke trying).
I like Bill H's approach. Basically make it a option on the PCB to keep the boards price competitive with all the other low cost embedded controller offerings out there.
As many have suggested Parallax needs WiFi in its product portfolio.
Restated, again ....
One way to do it is to source the Netgear device (or equivalent) as I mentioned before. It should be available for resale on the web site. I never suggested by-passing the 10BaseT (MagJack) connector. It is useful to provide a regulated power connection for devices like the Netgear adapter.
The other is providing a USB host ability. For USB all one needs are: 1) the right connections, 2) the right driver low level drivers, and 3) the right software for each device. Number 3 is the hardest to do, although many software packages are available for Linux that can be reverse-engineered.
I agree with bill...price is a BIG issue...you can get a raspy berry pi for only 25 to 35...if the board cost more than $50...I start looking at other options
The other is providing a USB host ability. For USB all one needs are: 1) the right connections, 2) the right driver low level drivers, and 3) the right software for each device. Number 3 is the hardest to do, although many software packages are available for Linux that can be reverse-engineered.
So from a hardware point of view, it would be a matter of adding some USB sockets. From a software point of view - is it possible to write USB host software?
I like Bill H's approach. Basically make it a option on the PCB to keep the boards price competitive with all the other low cost embedded controller offerings out there.
No way to argue with this. It provides the basics of USB at very low cost and allows the rest of the options to be added for a relatively small additional cost. The cost of a P2 board with USB and dongles for both WiFi and Bluetooth would most likely be lower than a P2 board with the three built in. Nothing beats piggybacking a design on mass market modules as far as cost and flexibility goes.
I believe I read a review for the $1.80 dongle where someone said that it works with USB 1.1. That would be 12.5 MBPS, right? We ought to be able to do that in the Prop2. How do we find the USB-side protocol, though? Is this something that Microsoft publishes a standard for? I see your links answer that question.
A good reference point / reality check for SW-based USB, is the work some have done on AVRs
This works on moderate clock speed AVRs at Low Speed USB rates.
I'd guess the FPGA Prop 2 boards would clone this easily, @ Low Speed, and might even get to 12MBd ?
This works on moderate clock speed AVRs at Low Speed USB rates.
I'd guess the FPGA Prop 2 boards would clone this easily, @ Low Speed, and might even get to 12MBd ?
I could be wrong, but this is my limited understanding...
If you use the USB dongle to WiFi (or Bluetooth), then don't you then have to do both the USB stack and the WiFi (or Bluetooth) stack as well. Won't this take far more resources than interfacing a (admittedly more expensive) WiFi (or Bluetooth) module?
You are correct, however most of the market is so price sensitive that the extra software/resources are well worth it.
FYI, we need a USB stack anyway, for keyboards, mice etc., and once that low-level stack is done, then all we need is the added bluetooth and tcp/ip layers.
Even with chips we would still need the Bluetooth and TCP/IP stacks anyway, except VERY expensive modules like Wiznet.
While the lowest level bit-banging code will need to be in pasm, the higher layers would likely be in CMM with PropGCC, Spin byte codes, Forth or similar, taking far less space than LMM code.
Concrete example:
Prop2 board with just USB/Magjack guesstimate $50-$80 each depending on quantity etc. User adds $2 Bluetooth dongle, $8 WiFi dongle for a total of $60-$90 depending on quantity.
Prop2 board with on-board WiFi & BlueTooth (would still need USB) would need to sell for $100-$150 due to certification, licensing etc costs. Do not underestimate FCC etc certification hassles/costs.
Note even the Raspberry Pi, made by a non-profit in HUGE quantities, with very low parts/manufacturing costs does not include Bluetooth or Wifi for cost issues... but does have USB, and ethernet on the $35 version.
I could be wrong, but this is my limited understanding...
If you use the USB dongle to WiFi (or Bluetooth), then don't you then have to do both the USB stack and the WiFi (or Bluetooth) stack as well. Won't this take far more resources than interfacing a (admittedly more expensive) WiFi (or Bluetooth) module?
Does the Prop1 USB driver use the video circuitry to serialize the data stream for either 1.5Mb/S or 12Mb/S?
Will it be possible for the Prop2 to use its video circuitry to serialize the data and free up some cog cycles to execute some of the other driver functions?
Does the Prop1 USB driver use the video circuitry to serialize the data stream for either 1.5Mb/S or 12Mb/S?
Will it be possible for the Prop2 to use its video circuitry to serialize the data and free up some cog cycles to execute some of the other driver functions?
Does the Prop1 USB driver use the video circuitry to serialize the data stream for either 1.5Mb/S or 12Mb/S?
Will it be possible for the Prop2 to use its video circuitry to serialize the data and free up some cog cycles to execute some of the other driver functions?
Ah! I am not the only one having problems with the forum double-posting!
No, you're not the only one. I have seen several others as well as mine. Seems to happen only when the "leaving page" error message pops up. Just prior to that one I posted on one of the other threads and did not get the error message or a duplicate post. Interesting problem.
Don't know about the P1 implementations.
On the P2, 1.5Mbps and 12Mbps can be bitbanged
Yes, I figured they could be bit banged on the P2, but that would require a cog anyway so I was wondering if it was possible to use the video circuitry to do the bit banging and use the cycles freed up to do a little of the low level stuff and possibly use any free cog ram for data buffers.
USB uses differential signaling, and on the P2 video output is to DACS only; but using it for output should be doable anyway. I can't see a way of using it for input.
1.5Mbps low speed should easily be doable as a single task in a cog per USB port
12Mbps may be doable with one task (of two) on a cog; I think one of four is a stretch. It would likely be quite possible to handle two 12Mbps USB ports with one cog.
I don't mind sacrificing one cog to two USB 12Mbps ports if it gets me Bluetooth and WiFi - more with an external hub (though that requires more software). Still much cheaper than adding Bluetooth and WiFi circuitry to the board, and avoids FCC hassles/costs.
Yes, I figured they could be bit banged on the P2, but that would require a cog anyway so I was wondering if it was possible to use the video circuitry to do the bit banging and use the cycles freed up to do a little of the low level stuff and possibly use any free cog ram for data buffers.
You are correct, however most of the market is so price sensitive that the extra software/resources are well worth it.
FYI, we need a USB stack anyway, for keyboards, mice etc., and once that low-level stack is done, then all we need is the added bluetooth and tcp/ip layers.
Even with chips we would still need the Bluetooth and TCP/IP stacks anyway, except VERY expensive modules like Wiznet.
While the lowest level bit-banging code will need to be in pasm, the higher layers would likely be in CMM with PropGCC, Spin byte codes, Forth or similar, taking far less space than LMM code.
Concrete example:
Prop2 board with just USB/Magjack guesstimate $50-$80 each depending on quantity etc. User adds $2 Bluetooth dongle, $8 WiFi dongle for a total of $60-$90 depending on quantity.
Prop2 board with on-board WiFi & BlueTooth (would still need USB) would need to sell for $100-$150 due to certification, licensing etc costs. Do not underestimate FCC etc certification hassles/costs.
Note even the Raspberry Pi, made by a non-profit in HUGE quantities, with very low parts/manufacturing costs does not include Bluetooth or Wifi for cost issues... but does have USB, and ethernet on the $35 version.
My point is that the two stacks will use most of the P2 resources with not much left for anything else. External modules are now available but they are relatively expensive. But with the new WiFi chips, it is likely that with the new TI chip and others, these prices will become available cheaply.
Currently you can buy a MK802 dongle that has 1GHz A10 and 1GB memory with WiFi, HDMI, 2xUSB (one is OTG) and microSD socket, runs *nix and android, and now costs <$40 shipped. There are later versions of these with dual cores, etc. This would be a much better add-on IMHO.
Anyway the original question was for P2 to be a success. The hobbyist is not the intended audience here (or at least those are my thoughts).
So, I would advocate that a board worthy of demoing the features of the prop would likely include...
* a USB connection - both LS & FS, both master & slave (can be done with links or driven with pins)
* an expansion connector so that external WiFi and Bluetooth modules could be added - adapter/wires may be required
* an optional Magjack would make sense too (agreed these are expensive, so just the pcb holes to fit)
But what is more to the point, is that a demo showing these things can be done with the P2. The TI WiFi chip IIRC is about $5. One of these on an unapproved pcb that could plug into the P2 would help to show what the P2 can do.
BTW I am fully aware of the FCC requirements, having gone through many of these approvals years ago - it is not simple and it is not cheap.
My point is that the two stacks will use most of the P2 resources with not much left for anything else. External modules are now available but they are relatively expensive. But with the new WiFi chips, it is likely that with the new TI chip and others, these prices will become available cheaply.
I don't agree that the two stacks "will use most of the P2 resources". Specifically, anything but the transport layer could run as CMM or byte code, potentially from flash.
Currently you can buy a MK802 dongle that has 1GHz A10 and 1GB memory with WiFi, HDMI, 2xUSB (one is OTG) and microSD socket, runs *nix and android, and now costs <$40 shipped. There are later versions of these with dual cores, etc. This would be a much better add-on IMHO.
I know... I ordered three different ones. One works great, one was DOA, one is yet to be tested.
Anyway the original question was for P2 to be a success. The hobbyist is not the intended audience here (or at least those are my thoughts).
So, I would advocate that a board worthy of demoing the features of the prop would likely include...
* a USB connection - both LS & FS, both master & slave (can be done with links or driven with pins)
* an expansion connector so that external WiFi and Bluetooth modules could be added - adapter/wires may be required
* an optional Magjack would make sense too (agreed these are expensive, so just the pcb holes to fit)
I agree with your list above, except I'd put on two USB connectors.
But what is more to the point, is that a demo showing these things can be done with the P2. The TI WiFi chip IIRC is about $5. One of these on an unapproved pcb that could plug into the P2 would help to show what the P2 can do.
Except as far as I know, you'd need just as much of a stack for the TI chip as for a USB wifi dongle that costs <$2... and we will need a USB stack anyway for other uses.
The price differential is even more important to volume commercial customers, and for retail products the differential is also going to be marked up.
It would IMHO be much more impressive to customers that the P2 can do USB, TCP/IP in software only. I am guessing that one cog could handle two USB ports (bit-banged) and another cog could handle a 10Mbps ethernet bit-banging easily (probably take only half a cog).
The higher layers of USB, Bluetooth, TCP/IP could run as XMM / CMM / byte code, but we would probably need 10k or so for buffers in the hub.
Much more impressive demo of a P2 than having to add WiFi, Bluetooth, etc chips with support circuits, and still have to add protocol stacks.
FYI, I don't consider chips like the Wiz with very limited built-in stacks viable, due to their BOM costs for volume product; they are great for short runs or one offs, but not for volume.
It would be interesting to have a prop2 SoM (System on Module) so that we can buy a few of them to put on custom boards. Resistors and jumpers (or zero ohm) could be set to choose which feature we want. Want ethernet or usb? Just put the connector on your board and connect the right pins. Same thing with AUDIO/VGA/RCA/DVI/HDMI/PS2...
Here is a picture of another SoM (Cortex-M4 I think):
I could see a cheap Wii type console being sold at CVS for $39.95 using the Prop2
Something like the Zone 40 but much better
But depends if someone could do hdmi-widescreen-480p on the Prop2 in software.
VGA/Component + Audio is OK but not ideal.
However I don't have a clue as to how popular it is. A standard eval board design with just the basics like the Freescale Freedom board should be enough. Just make sure and give the eval folks access to all the pins so they can attach whatever they want.
However I don't have a clue as to how popular it is. A standard eval board design with just the basics like the Beagle bone board should be enough. Just make sure and give the eval folks access to all the pins so they can attach whatever they want.
A Bluetooth serial port can be used with SimpleIDE for communications. Programming will require connecting DTR or RTS to the reset circuit. I've done both of these successfully YMMV. WiFi access is another issue. A WiFi virtual serial port using TCP may have potential.
If P2 can do some USB host mode (USB OTG?) we're set for doing lots of good things.
This looks interesting at about $40 for a P2 soft-wired-ethernet 10Mbps solution.
I have several of these adapters. It is how my Spinneret in my chicken coop serves up html to the world. It's a good product in my opinion.
As for P2 connectivity, (from a user stand point) I'd want USB and WiFi for sure. Bluetooth is good too.
GHI and Beagles and the like have a lot of nice goodies. All far too expensive. Like over a hundred dollars. Which is lot to spend on an experiment. Especially as the also require a investment of time and effort to get into yet another system/OS/language/IDE.
Nowadays you have to aim for a Raspberry Pi price level.
As I said before, the simplest, cheapest possible board that, dare I say it, "leverages" the million plus Raspberry Pis out there in the field might be just the thing.
Plug onto your Pi, download code, bingo! piles of extra I/O, digital and analog. VGA terminal/graphics, and so on. Think Prop as useful peripheral rather than main system here.
Don't forget also that the Raspi's focus is on education. Totally in line with a lot of Parallax work.
For industrial users, I don't know, perhaps an all singing all dancing demo board is appropriate there.
Bill Henning says:
"Basically, in this day and age, the market is extremely price sensitive, so it is paramount to keep that in mind."
This point can't be stressed enough, and it applies to chips themselves, boards, everything -- in practically every market you'd want to consider. The "cool" factor might get you a little extra breathing space, and probably you'll earn some more if you combine power/features + simplicity (ie, your product reduces development time). But there are many many options out there nowadays, and the market can be quite ruthless when it comes to price.
Bill Henning says:
"It would IMHO be much more impressive to customers that the P2 can do USB, TCP/IP in software only. I am guessing that one cog could handle two USB ports (bit-banged) and another cog could handle a 10Mbps ethernet bit-banging easily (probably take only half a cog)."
P2 should be able to do all this without too much difficulty -- that is the Propeller way, after all -- and no doubt it's good to show off a little.
Heater says:
"Nowadays you have to aim for a Raspberry Pi price level.
As I said before, the simplest, cheapest possible board that, dare I say it, "leverages" the million plus Raspberry Pis out there in the field might be just the thing.
Plug onto your Pi, download code, bingo! piles of extra I/O, digital and analog. VGA terminal/graphics, and so on. Think Prop as useful peripheral rather than main system here."
Excellent point. Perhaps it is best to focus more on situations where Propeller is an add-on rather than the system as such.
I personally think the best, most valuable time spent is on USB. USB 1.1 support to start with would be very useful and can be achieved relatively simply (there are plenty of software references).
While bit-twanged Ethernet may be nice from a parts count perspective, it will result in more development effort, just to get a software PHY written. I personally think that it's pointless to offer a lackluster solution simply for the ease of saying "it can be done".
I have a 624J600 board in hand, and given the pricing of the chip, I feel this is a far more elegant and reliable solution. The total BOM is well less than $10, and roughly $6 more than a software only solution, but it does 100BaseT, and I know a software solution can't do proper 100BaseT with a reasonable COG utilization.
Comments
"Often in this situation Ill break out something like the Saxo board, with an FPGA and a thick USB 2.0 pipe. In this case, I was dealing with low enough data rates that I could do something even simpler. I plugged the remote into my Propeller Demo Board and wrote a quick program to capture the SPI traffic and send it back in ASCII over the serial port."
But, in the end, I do still think that Bill has the right general idea: Provide the basic (requisite) connections "free" out-of-the-box and let the rest be done with dongles etc. I can see where that might have strong appeal, as compared to other options, in many situations (and anyway, you can't give 'em everything they might want -- but you can go broke trying).
Restated, again ....
One way to do it is to source the Netgear device (or equivalent) as I mentioned before. It should be available for resale on the web site. I never suggested by-passing the 10BaseT (MagJack) connector. It is useful to provide a regulated power connection for devices like the Netgear adapter.
The other is providing a USB host ability. For USB all one needs are: 1) the right connections, 2) the right driver low level drivers, and 3) the right software for each device. Number 3 is the hardest to do, although many software packages are available for Linux that can be reverse-engineered.
Well, she is definitely not shy or inhibited.
Interesting thought. USB Host. Ok, how cheap can you get a USB to Wifi dongle? Sort by "price+shipping" on ebay and the answer is $4 including shipping http://www.ebay.com.au/itm/WF001AU-Latest-Mini-USB-2-0-WiFi-Wireless-WLAN-150Mbps-Network-Dongle-Adapter-/121050774531?pt=LH_DefaultDomain_15&hash=item1c2f303c03 ($US and $AU are about the same, and I searched on the Australian version of ebay as it can sort by "buy it now" only which then gives the real price rather than an auction price. Lots coming up between $4 and $5).
So from a hardware point of view, it would be a matter of adding some USB sockets. From a software point of view - is it possible to write USB host software?
No way to argue with this. It provides the basics of USB at very low cost and allows the rest of the options to be added for a relatively small additional cost. The cost of a P2 board with USB and dongles for both WiFi and Bluetooth would most likely be lower than a P2 board with the three built in. Nothing beats piggybacking a design on mass market modules as far as cost and flexibility goes.
A good reference point / reality check for SW-based USB, is the work some have done on AVRs
Mostly derivatives of the same work, but expanded to support more clock choices, here is one example :
http://www.obdev.at/products/vusb/index.html
This works on moderate clock speed AVRs at Low Speed USB rates.
I'd guess the FPGA Prop 2 boards would clone this easily, @ Low Speed, and might even get to 12MBd ?
1.5Mbps should be very easy on the FPGA, 40 clocks per bit.
For 10Mbps Ethernet, we would have 6 clocks per bit on the FPGA, I suspect that should be enough for Manchester en/de coding.
If you use the USB dongle to WiFi (or Bluetooth), then don't you then have to do both the USB stack and the WiFi (or Bluetooth) stack as well. Won't this take far more resources than interfacing a (admittedly more expensive) WiFi (or Bluetooth) module?
FYI, we need a USB stack anyway, for keyboards, mice etc., and once that low-level stack is done, then all we need is the added bluetooth and tcp/ip layers.
Even with chips we would still need the Bluetooth and TCP/IP stacks anyway, except VERY expensive modules like Wiznet.
While the lowest level bit-banging code will need to be in pasm, the higher layers would likely be in CMM with PropGCC, Spin byte codes, Forth or similar, taking far less space than LMM code.
Concrete example:
Prop2 board with just USB/Magjack guesstimate $50-$80 each depending on quantity etc. User adds $2 Bluetooth dongle, $8 WiFi dongle for a total of $60-$90 depending on quantity.
Prop2 board with on-board WiFi & BlueTooth (would still need USB) would need to sell for $100-$150 due to certification, licensing etc costs. Do not underestimate FCC etc certification hassles/costs.
Note even the Raspberry Pi, made by a non-profit in HUGE quantities, with very low parts/manufacturing costs does not include Bluetooth or Wifi for cost issues... but does have USB, and ethernet on the $35 version.
Does the Prop1 USB driver use the video circuitry to serialize the data stream for either 1.5Mb/S or 12Mb/S?
Will it be possible for the Prop2 to use its video circuitry to serialize the data and free up some cog cycles to execute some of the other driver functions?
Does the Prop1 USB driver use the video circuitry to serialize the data stream for either 1.5Mb/S or 12Mb/S?
Will it be possible for the Prop2 to use its video circuitry to serialize the data and free up some cog cycles to execute some of the other driver functions?
Don't know about the P1 implementations.
On the P2, 1.5Mbps and 12Mbps can be bitbanged
No, you're not the only one. I have seen several others as well as mine. Seems to happen only when the "leaving page" error message pops up. Just prior to that one I posted on one of the other threads and did not get the error message or a duplicate post. Interesting problem.
Yes, I figured they could be bit banged on the P2, but that would require a cog anyway so I was wondering if it was possible to use the video circuitry to do the bit banging and use the cycles freed up to do a little of the low level stuff and possibly use any free cog ram for data buffers.
1.5Mbps low speed should easily be doable as a single task in a cog per USB port
12Mbps may be doable with one task (of two) on a cog; I think one of four is a stretch. It would likely be quite possible to handle two 12Mbps USB ports with one cog.
I don't mind sacrificing one cog to two USB 12Mbps ports if it gets me Bluetooth and WiFi - more with an external hub (though that requires more software). Still much cheaper than adding Bluetooth and WiFi circuitry to the board, and avoids FCC hassles/costs.
Currently you can buy a MK802 dongle that has 1GHz A10 and 1GB memory with WiFi, HDMI, 2xUSB (one is OTG) and microSD socket, runs *nix and android, and now costs <$40 shipped. There are later versions of these with dual cores, etc. This would be a much better add-on IMHO.
Anyway the original question was for P2 to be a success. The hobbyist is not the intended audience here (or at least those are my thoughts).
So, I would advocate that a board worthy of demoing the features of the prop would likely include...
* a USB connection - both LS & FS, both master & slave (can be done with links or driven with pins)
* an expansion connector so that external WiFi and Bluetooth modules could be added - adapter/wires may be required
* an optional Magjack would make sense too (agreed these are expensive, so just the pcb holes to fit)
But what is more to the point, is that a demo showing these things can be done with the P2. The TI WiFi chip IIRC is about $5. One of these on an unapproved pcb that could plug into the P2 would help to show what the P2 can do.
BTW I am fully aware of the FCC requirements, having gone through many of these approvals years ago - it is not simple and it is not cheap.
I don't agree that the two stacks "will use most of the P2 resources". Specifically, anything but the transport layer could run as CMM or byte code, potentially from flash.
I know... I ordered three different ones. One works great, one was DOA, one is yet to be tested.
I agree with your list above, except I'd put on two USB connectors.
Except as far as I know, you'd need just as much of a stack for the TI chip as for a USB wifi dongle that costs <$2... and we will need a USB stack anyway for other uses.
The price differential is even more important to volume commercial customers, and for retail products the differential is also going to be marked up.
It would IMHO be much more impressive to customers that the P2 can do USB, TCP/IP in software only. I am guessing that one cog could handle two USB ports (bit-banged) and another cog could handle a 10Mbps ethernet bit-banging easily (probably take only half a cog).
The higher layers of USB, Bluetooth, TCP/IP could run as XMM / CMM / byte code, but we would probably need 10k or so for buffers in the hub.
Much more impressive demo of a P2 than having to add WiFi, Bluetooth, etc chips with support circuits, and still have to add protocol stacks.
FYI, I don't consider chips like the Wiz with very limited built-in stacks viable, due to their BOM costs for volume product; they are great for short runs or one offs, but not for volume.
I figured you'd know, but some reading this thread may not.
Here is a picture of another SoM (Cortex-M4 I think):
Something like the Zone 40 but much better
But depends if someone could do hdmi-widescreen-480p on the Prop2 in software.
VGA/Component + Audio is OK but not ideal.
Ghielectronics does the very thing with their upper end products.
Link:
http://www.ghielectronics.com/catalog/product/398
However I don't have a clue as to how popular it is. A standard eval board design with just the basics like the Freescale Freedom board should be enough. Just make sure and give the eval folks access to all the pins so they can attach whatever they want.
Ghielectronics does the very thing with their upper end products.
Link:
http://www.ghielectronics.com/catalog/product/398
However I don't have a clue as to how popular it is. A standard eval board design with just the basics like the Beagle bone board should be enough. Just make sure and give the eval folks access to all the pins so they can attach whatever they want.
As for P2 connectivity, (from a user stand point) I'd want USB and WiFi for sure. Bluetooth is good too.
Paul
Nowadays you have to aim for a Raspberry Pi price level.
As I said before, the simplest, cheapest possible board that, dare I say it, "leverages" the million plus Raspberry Pis out there in the field might be just the thing.
Plug onto your Pi, download code, bingo! piles of extra I/O, digital and analog. VGA terminal/graphics, and so on. Think Prop as useful peripheral rather than main system here.
Don't forget also that the Raspi's focus is on education. Totally in line with a lot of Parallax work.
For industrial users, I don't know, perhaps an all singing all dancing demo board is appropriate there.
This point can't be stressed enough, and it applies to chips themselves, boards, everything -- in practically every market you'd want to consider. The "cool" factor might get you a little extra breathing space, and probably you'll earn some more if you combine power/features + simplicity (ie, your product reduces development time). But there are many many options out there nowadays, and the market can be quite ruthless when it comes to price.
P2 should be able to do all this without too much difficulty -- that is the Propeller way, after all -- and no doubt it's good to show off a little.
Excellent point. Perhaps it is best to focus more on situations where Propeller is an add-on rather than the system as such.
While bit-twanged Ethernet may be nice from a parts count perspective, it will result in more development effort, just to get a software PHY written. I personally think that it's pointless to offer a lackluster solution simply for the ease of saying "it can be done".
I have a 624J600 board in hand, and given the pricing of the chip, I feel this is a far more elegant and reliable solution. The total BOM is well less than $10, and roughly $6 more than a software only solution, but it does 100BaseT, and I know a software solution can't do proper 100BaseT with a reasonable COG utilization.