gtechn
Posts: 156
Joined: Thu Jan 07, 2016 5:32 pm

Why the Raspberry Pi 4 wishlists are tough sells

Tue May 09, 2017 3:04 pm

Hello,

We all know that the speculation route has been gone down a million times and never ends well, with an almost weekly new thread on mythical upgrades from people who want to use a Raspberry Pi instead of using a (somewhat more expensive) Mini-ITX computer (or the like) which are more suited to their purposes.

So, here is some logic on the mythical Raspberry Pi 4 (that is certainly in development) so people will hopefully start understanding what they are saying when they make these wish lists. I also am putting in some speculation of my own at the end, but I think it will disappoint many.

1. The Raspberry Pi 4 will, almost certainly, use the VideoCore. Otherwise, the Raspberry Pi risks falling apart because it will no longer be backwards-compatible and a new chip (like Mali) would have employees in the RPF tied up for months trying to learn how to use the new chip, causing updates and repairs to slow to a crawl. It would also open the door to Raspberry Pi Clones and seriously damage the Raspberry Pi's reputation.

2. The 1GB of RAM will be harder to fix than most people think. The RPF almost certainly has to use VideoCore, as mentioned above. The problem is that the VideoCore IV has a RAM limitation in the chip itself. To remove this will cost, according to another Raspberry Pi Engineer, several million dollars to repair (by rough estimate). This is big stuff. On top of that, when and if they get the RAM fixed, they now have to pay more money for a 2GB RAM chip (which is more expensive than a 1GB RAM chip).

3. USB 3 is in the same boat as the RAM, except even more expensive. If you want faster Ethernet and USB 3, pretty much the entire I/O on the VideoCore needs adjustment, which is (almost certainly) harder and more expensive to fix than the RAM. Several million (again) please, with a few million more for comfort!

4. New Codecs (like H.265) are also hard.. Same reasons as above, plus a license fee for adding support for such codec. This is a niche category.

5. The VideoCore is currently built at 40nm. The next release will probably need 28nm. Otherwise, the Raspberry Pi 4 will overheat because of the build size being too large, and will run slower. Again, more work.

6. The RPF doesn't have the money to do everything everybody wants. Read the financial statements for 2015 on the Raspberry Pi website. Doing this is going to be tough. It is a matter of finding out what is most beneficial to the community and what is not.

7. The Raspberry Pi Foundation has almost maxed out every upgrade that is not crazy expensive. There are only so many upgrades you can do and keep the price $35. The Raspberry Pi Foundation has done almost all of them. It is into the big, expensive stuff now. Think: The RPF is limited on money. The upgrades are in the multi-million dollar range. How can those upgrades pay off when you are selling the work of those upgrades for $35?

So, what will Raspberry Pi 4 look like? This is my personal, community opinion. There are two types of wishlists. Most wishlists and theories on here are on what the viewers would like to see, but just aren't realistic in demands. This speculation list is more reasonable to the problems above.

1. A CPU just like the Raspberry Pi 3, just at 1.4-1.5 GHz, at 28nm. Small stuff that is pretty necessary.
2. Due to intense demand, probably 2GB RAM. (Somehow).
3. The 3D area for VideoCore V was finished, but not much else. I personally see a VideoCore V/IV hybrid put together as part of the RAM upgrade, that might be faster but without any new codecs or other fancy stuff. Just faster, and that is it. I could also see the same VideoCore IV as on all the old boards, without any improvements but the RAM one.
4. The IO probably won't see any major increases. Maybe the RPF will have time for minor adjustments and minor improvements, but I don't expect USB 3 in it or Gigabit Ethernet.

I know that my view on what the Raspberry Pi 4 will be is much lower than what many people wish list, but this is a more realistic approach to upgrades based on cost-benefit ratio. If you don't like what this looks like, remember, you can get a Mini-ITX computer for $200 or so, and have several times the performance there to do with as you like. Sorry.

The other option from Raspberry Pi would be to build a supercharged Raspberry Pi 4 with all upgrades mentioned above, but at a $70 or so asking price. The Raspberry Pi Foundation would (instead of taking about $1 per board sold as of right now), get $15 or so per "Model C" purchase to help pay off these upgrades. Then again, this could be unlikely because of the talk with Google's CEO, who convinced Mr. Upton to make sure the Pi 2 remained $35 instead of being more expensive (with some more performance).

The only thing the community can do at this point is watch and wait for the day the Pi 4 comes out, and have fun then! :D

W. H. Heydt
Posts: 10981
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: Why the Raspberry Pi 4 wishlists are tough sells

Tue May 09, 2017 3:43 pm

gtechn wrote: 3. USB 3 is in the same boat as the RAM, except even more expensive. If you want faster Ethernet and USB 3, pretty much the entire I/O on the VideoCore needs adjustment, which is (almost certainly) harder and more expensive to fix than the RAM. Several million (again) please, with a few million more for comfort!
I'm not at all sure this is the case. USB3 should--at least in concept--be easier to do than expanding the number of address lines, though it depends on exactly how isolated the USB "block" is and whether or not replacing USB2 with USB3 affects other areas of the SoC. The other fly in the ointment is any licensing charges that will exist for USB3, as there isn't a lot of headroom to increase the cost of the SoC and still fit within the price point of the B-series.
7. The Raspberry Pi Foundation has almost maxed out every upgrade that is not crazy expensive. There are only so many upgrades you can do and keep the price $35. The Raspberry Pi Foundation has done almost all of them. It is into the big, expensive stuff now. Think: The RPF is limited on money. The upgrades are in the multi-million dollar range. How can those upgrades pay off when you are selling the work of those upgrades for $35?
Four million Pi3Bs sold in the first year of release. It is reasonable to expect a putative Pi4B to do the same, so Broadcom would have a pretty much guaranteed multimillion chip sale for whatever the RPF has decided to use next. And that's without considering if there are any other customers that would be happy to buy the same SoC. Other than that, I agree. The "low hanging fruit" has been picked.
1. A CPU just like the Raspberry Pi 3, just at 1.4-1.5 GHz, at 28nm. Small stuff that is pretty necessary.
Or stick with 1.2GHz (at least for a default, at least at first...see what has been done with the BCR2835). This is for two reasons. The first is that Broadcom appears to be a reasonably conservative company in rating their chips. The second is that, at 28nm, a 1.2GHz chips is likely to have a lot less of a thermal issue than the existing 1.2GHz 40nm part.

2. Due to intense demand, probably 2GB RAM. (Somehow).
For some values of "demand"... There is also the cost factor. What I would expect is that the initial version of the Pi4B will have 1GB RAM, just as the initial Model B had 256MB (and the original plan for the Model A was 128MB!) with a possible "mid-life kicker" to 2GB when RAM packages and prices permit.
4. The IO probably won't see any major increases. Maybe the RPF will have time for minor adjustments and minor improvements, but I don't expect USB 3 in it or Gigabit Ethernet.
This is the one area where I think there is an actual need for the upgrade. I don't "expect" to see USB3, but the Pi could sure use it. Much more so than more memory or a higher clock speed.
The other option from Raspberry Pi would be to build a supercharged Raspberry Pi 4 with all upgrades mentioned above, but at a $70 or so asking price. The Raspberry Pi Foundation would (instead of taking about $1 per board sold as of right now), get $15 or so per "Model C" purchase to help pay off these upgrades. Then again, this could be unlikely because of the talk with Google's CEO, who convinced Mr. Upton to make sure the Pi 2 remained $35 instead of being more expensive (with some more performance).
I believe that you have misread or misunderstood the nature of the remarks made to Dr. Upton. He mused about making a more expensive ("server") model of the Pi and was told that a better path was to get as close to free as possible, and that resulted in the Pi0. In some respects, the Pi0 has been a less than sterling addition and a major distraction, fun as the Pi0 and Pi0W are.
The only thing the community can do at this point is watch and wait for the day the Pi 4 comes out, and have fun then! :D
Well...yes.

Bear in mind that whatever the SoC is for the Pi4, the *design* work was probably completed a couple of years ago and there has probably been prototype Silicon in existence for at least a year. If--as is possible--the Pi4B is launched next year--then by now there would have to be production ready Silicon in existence. The remaining major tasks will be getting the board design finalized, getting the software ready for it and ramping up chip production to make the pre-launch supplies (at a guess, the RPT should have about 200K boards on hand on launch day...). If the Pi4B doesn't launch until 2019 you can push some of those dates back by 6 to 12 months.

hippy
Posts: 6068
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: Why the Raspberry Pi 4 wishlists are tough sells

Tue May 09, 2017 3:56 pm

It all sounds reasonable but I am not so sure about the 'millions of dollars' costs. That may well be right but it seems the RPT/RPF have taken the original BCM2835 and created the BCM2836 and the BCM2837 without incurring such huge costs. More fundamental changes than 'just slapping cores together' will have a higher cost but I am suspicious of how accurate any estimates of costs are. Is "several million dollars" an accurate estimate or just a handy synonym for "a lot of money" ?

And while the RPT/RPF is not the richest business in the world it does seem to have revenues which could support "several million dollars" of engineering costs or investment. The 'Raspberry Pi' may be a charity but it is far from penniless. It is a multi-million dollar business which could choose to spend "a lot of money" if it wanted to or had to.

While not wanting to spend money, not wanting to break backwards compatibility, not wanting to lose investment already put in, supports continuing along the same path, it doesn't mean things can't or won't change course. It is perhaps not likely to - and I think 'milking it for all it's worth' with some 'additional tweaks' along the way will most likely be the path - if there's one thing the RPT/RPF does well; it is to surprise people.

gtechn
Posts: 156
Joined: Thu Jan 07, 2016 5:32 pm

Re: Why the Raspberry Pi 4 wishlists are tough sells

Tue May 09, 2017 4:09 pm

It all sounds reasonable but I am not so sure about the 'millions of dollars' costs. That may well be right but it seems the RPT/RPF have taken the original BCM2835 and created the BCM2836 and the BCM2837 without incurring such huge costs. More fundamental changes than 'just slapping cores together' will have a higher cost but I am suspicious of how accurate any estimates of costs are. Is "several million dollars" an accurate estimate or just a handy synonym for "a lot of money" ?
Actually, according to the RPF, this IS a good estimate. They said that making the BCM2836 and BCM2837 actually cost them millions, and that was a simple "cut and shut" where they just cut, copy, and paste existing parts together. Now think the costs of that plus the cost of upgrades.
And while the RPT/RPF is not the richest business in the world it does seem to have revenues which could support "several million dollars" of engineering costs or investment. The 'Raspberry Pi' may be a charity but it is far from penniless. It is a multi-million dollar business which could choose to spend "a lot of money" if it wanted to or had to.
They could, of course, but they have repeatedly said that making these changes just won't pay off because the RPF/RPT isn't big enough. They would be taking financial losses on some of these upgrades for a long time.
I'm not at all sure this is the case. USB3 should--at least in concept--be easier to do than expanding the number of address lines, though it depends on exactly how isolated the USB "block" is and whether or not replacing USB2 with USB3 affects other areas of the SoC. The other fly in the ointment is any licensing charges that will exist for USB3, as there isn't a lot of headroom to increase the cost of the SoC and still fit within the price point of the B-series.
USB 3 would affect other parts of the SoC, because even though USB 3 might be easy to add, making it fast enough people would care (and call it USB 3) would require adjustments to the processing, SD card reader, and various "pipeline" adjustments. Also, as far as I know, there are no licensing fees with USB 3 standard itself, just on the implementation (if another company built silicon for it, and you rented it from them). If Broadcom added USB 3 themselves, there would be no licensing fee, just fees for designing the chips.
Or stick with 1.2GHz (at least for a default, at least at first...see what has been done with the BCR2835). This is for two reasons. The first is that Broadcom appears to be a reasonably conservative company in rating their chips. The second is that, at 28nm, a 1.2GHz chips is likely to have a lot less of a thermal issue than the existing 1.2GHz 40nm part.
The Raspberry Pi 3 overheating is a myth according to the Raspberry Pi Foundation. If you don't overclock, you will not have problems. It will get warm, yes, but it will not cause problems. It would need to be 1.4-1.5GHz at least, otherwise, it would appear to be as fast as the old version in processing speed. It kills a major incentive for upgrading.
For some values of "demand"... There is also the cost factor. What I would expect is that the initial version of the Pi4B will have 1GB RAM, just as the initial Model B had 256MB (and the original plan for the Model A was 128MB!) with a possible "mid-life kicker" to 2GB when RAM packages and prices permit.
I doubt it. People weren't happy when the Pi 3 came out with 1GB RAM. A Pi 4 with 1GB of RAM? It would be better for the Raspberry Pi Foundation to simply not release the Pi 4 until prices permit for 2GB. Besides, that gives them more time to work on the processor anyway.
I believe that you have misread or misunderstood the nature of the remarks made to Dr. Upton. He mused about making a more expensive ("server") model of the Pi and was told that a better path was to get as close to free as possible, and that resulted in the Pi0. In some respects, the Pi0 has been a less than sterling addition and a major distraction, fun as the Pi0 and Pi0W are.
Sadly yes. A "server" version would be a great idea. However, Pi Zero W might not be a distraction soon. 250,000 Pi Zero Ws in the wild now, in only 2 months!

hippy
Posts: 6068
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: Why the Raspberry Pi 4 wishlists are tough sells

Tue May 09, 2017 4:33 pm

gtechn wrote:
I am suspicious of how accurate any estimates of costs are. Is "several million dollars" an accurate estimate or just a handy synonym for "a lot of money" ?
Actually, according to the RPF, this IS a good estimate. They said that making the BCM2836 and BCM2837 actually cost them millions, and that was a simple "cut and shut" where they just cut, copy, and paste existing parts together. Now think the costs of that plus the cost of upgrades.
If that is so then the encouraging message is the RPT/RPF have been able to support engineering costs of millions and they have higher revenues now than in the past.

$10 million sounds a lot but as that could be recuperated through a $5 unit increase over 2 million sales it seems entirely feasible.

It perhaps depends on where costs are incurred; in design or manufacturing. One might be able to do a lot more for little extra expenditure.

I am not convinced cost is such an obstacle but without a more detailed breakdown it's impossible to prove either way.

W. H. Heydt
Posts: 10981
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: Why the Raspberry Pi 4 wishlists are tough sells

Tue May 09, 2017 4:53 pm

gtechn wrote:
I'm not at all sure this is the case. USB3 should--at least in concept--be easier to do than expanding the number of address lines, though it depends on exactly how isolated the USB "block" is and whether or not replacing USB2 with USB3 affects other areas of the SoC. The other fly in the ointment is any licensing charges that will exist for USB3, as there isn't a lot of headroom to increase the cost of the SoC and still fit within the price point of the B-series.
USB 3 would affect other parts of the SoC, because even though USB 3 might be easy to add, making it fast enough people would care (and call it USB 3) would require adjustments to the processing, SD card reader, and various "pipeline" adjustments. Also, as far as I know, there are no licensing fees with USB 3 standard itself, just on the implementation (if another company built silicon for it, and you rented it from them). If Broadcom added USB 3 themselves, there would be no licensing fee, just fees for designing the chips.
Should not affect the SD card interface. That's on the SDIO, not the UB bus. Though a faster SDIO would probably be an appreciated upgrade, I don't recall seeing anyone jump up and down asking for it. As for the rest of the USB3 issues...maybe. There is one other unaddressed issue, though. If the SoC moves to USB3, then a new LAN chip will be needed and unless the on chip boot ROM is changed, it would have to have a compatible NIC or the latest feature--network booting--won't work. (This is, I think, a good example of "unintended consequences" to requested changes...people forget what *else* may be affected.)
Or stick with 1.2GHz (at least for a default, at least at first...see what has been done with the BCR2835). This is for two reasons. The first is that Broadcom appears to be a reasonably conservative company in rating their chips. The second is that, at 28nm, a 1.2GHz chips is likely to have a lot less of a thermal issue than the existing 1.2GHz 40nm part.
The Raspberry Pi 3 overheating is a myth according to the Raspberry Pi Foundation. If you don't overclock, you will not have problems. It will get warm, yes, but it will not cause problems. It would need to be 1.4-1.5GHz at least, otherwise, it would appear to be as fast as the old version in processing speed. It kills a major incentive for upgrading.
*If* the Pi4 has faster I/O (i.e. USB3), then I'd switch some Pi3 uses to it in a hot minute even with a 1.2GHz clock and 1GB RAM. I will grant that, for one of my major uses, I would feel compelled to wait for a CM4L, but it's not the SoC feature set that would hold me back. And again, even if the Pi4 defaults to 1.2GHz, if it is a 28nm part (very likely, I think), then there will likely be "overhead" for overclocking, which the Pi3 doesn't have, making the Pi4 a more attractive board. And, of course, there would be the potential for a Pi2Bv1.3....
For some values of "demand"... There is also the cost factor. What I would expect is that the initial version of the Pi4B will have 1GB RAM, just as the initial Model B had 256MB (and the original plan for the Model A was 128MB!) with a possible "mid-life kicker" to 2GB when RAM packages and prices permit.
I doubt it. People weren't happy when the Pi 3 came out with 1GB RAM. A Pi 4 with 1GB of RAM? It would be better for the Raspberry Pi Foundation to simply not release the Pi 4 until prices permit for 2GB. Besides, that gives them more time to work on the processor anyway.
Maybe. It might depend on what else is there and/or what is said about it. Pretty obviously, if there is a 1GB Pi4B and anything at all is said about upgrades to the VC, the first question on many minds will be: How much RAM can it address?
I believe that you have misread or misunderstood the nature of the remarks made to Dr. Upton. He mused about making a more expensive ("server") model of the Pi and was told that a better path was to get as close to free as possible, and that resulted in the Pi0. In some respects, the Pi0 has been a less than sterling addition and a major distraction, fun as the Pi0 and Pi0W are.
Sadly yes. A "server" version would be a great idea. However, Pi Zero W might not be a distraction soon. 250,000 Pi Zero Ws in the wild now, in only 2 months!
Even with 250K Pi0Ws sold in 9 weeks, orders are still restricted to one per order. Until that limit is removed, I think the Pi0/Pi0W will be a distraction. Much as I'd like to see a better balance between supply and demand, I don't really care to guess how long that will take, and in the mean time, that has to be a major headache for Eben. (It's really very odd... Original stretch goal: 20K Model B Pis in the first year. Four *million* Pi3Bs in the first year. 250K Pi0Ws in 9 weeks...is a problem. Bet there are a lot of small SBC companies out there that would *love* to have that problem.)

User avatar
scruss
Posts: 2542
Joined: Sat Jun 09, 2012 12:25 pm
Location: Toronto, ON
Contact: Website

Re: Why the Raspberry Pi 4 wishlists are tough sells

Tue May 09, 2017 5:12 pm

Will there even be a 4?

Once you start getting into more RAM and faster processors, there's already commercial competition. The Zero is pretty much the cheap, simple, educational computer that the Foundation set out to build, but it needed the commercial success of the B/2/3 to raise capital and community to get there.

I'd be expecting any next Raspberry Pi to be less powerful than the 3.
‘Remember the Golden Rule of Selling: “Do not resort to violence.”’ — McGlashan.

gtechn
Posts: 156
Joined: Thu Jan 07, 2016 5:32 pm

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 12:55 am

Thankfully, the faster they get polluted the faster they get locked by the mods - which is the perfect relief from this total dross.
Actually, right now, you seem to be the only one polluting this thread right now. Just saying.

Also, I really like these threads when they answer technical questions like my original post. It helps others learn more about the technology behind the Raspberry Pi and understand what they are asking for better, thus making more intelligent wishlists. Furthermore, attempting to disprove a wish list item helps me (and hopefully others), again, learn more about VideoCore and why said feature would or would not work.

So, bring these intelligent wish list threads on!

Bad Wish List:
3.7GHz 6-core ARM which can run x86
4K Graphics Core
USB 3.1 Gen 2 and 10-Class SD cards (why not?)

Good Wish List:
1.5GHz Quad-Core ARM, because the Raspberry Pis in the past like smaller upgrades like this
2 GB RAM after RAM upgrade above, and price stabilization.

gtechn
Posts: 156
Joined: Thu Jan 07, 2016 5:32 pm

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 12:58 am

Forum Moderator mahjongg clearly says at viewtopic.php?f=63&t=182744&start=50 :
It is NO problem that people want to talk about their dreams, as long as they ask about somewhat realistic things, (some people may have heard about some innovations, and may want it, but do not have enough technical know-how to realize what they really are asking, so try to inform yourself before going off to ask for impossible things, like x86 compatibility).

But I also know some people are getting sick of this discussion, and may want to derail it.

To all of you I only say keep it nice and on topic, and I will try to keep the peace, which means I will delete posts that do not add to the discussion, but are simply inflammatory, or misplaced "humour".
So, yeah. Those of you who don't like speculation threads, just ignore them.

Zebu
Posts: 29
Joined: Tue Aug 11, 2015 1:57 am
Location: Australia

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 2:08 am

I would be happy with the pi4 being exactly the same specs as the pi3 but with the connectors all along one edge rather than all around

even a pi3a with the the usb connector rotated around to be on the same edge as the hdmi connector

'tis but a dream ;)

W. H. Heydt
Posts: 10981
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 2:27 am

Zebu wrote:I would be happy with the pi4 being exactly the same specs as the pi3 but with the connectors all along one edge rather than all around

even a pi3a with the the usb connector rotated around to be on the same edge as the hdmi connector

'tis but a dream ;)
That idea has been made before. Have you considered trying to find a M-F Right Angled USB adapter so that, while the USB is along a different edge, the cable still goes in the sam direction as the others?

Or, alternatively, take a look at the WD SATA Adapter. Apart from the SATA connector, all the connectors are on the same edge.

User avatar
Gavinmc42
Posts: 3874
Joined: Wed Aug 28, 2013 3:31 am

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 2:59 am

Perhaps we should have a scoring system based on what is possible?

Sata - 0 out of 10
USB3 - 0/10
2GB - 1/10
1.5GHz - 5/19
Pink PCB with unicorn picture - 9/10

Then we can add up the scores when the Pi4 comes out and see who guessed closest and they win a big Raspberry :lol:
Must be having an off day, even I am getting tired of speculating.
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

Zebu
Posts: 29
Joined: Tue Aug 11, 2015 1:57 am
Location: Australia

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 3:21 am

W. H. Heydt wrote: That idea has been made before. Have you considered trying to find a M-F Right Angled USB adapter so that, while the USB is along a different edge, the cable still goes in the sam direction as the others?
this is my current solution
W. H. Heydt wrote: Or, alternatively, take a look at the WD SATA Adapter. Apart from the SATA connector, all the connectors are on the same edge.
WD seem to go out of their way to make pi things hard to buy in Australia, though i think pihut may ship it here. I'm leaning towards designing a carrier board with enough connectors to keep me happy and getting a couple made up at seed or ohspark

mfa298
Posts: 1387
Joined: Tue Apr 22, 2014 11:18 am

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 7:08 am

gtechn wrote: 1. The Raspberry Pi 4 will, almost certainly, use the VideoCore. Otherwise, the Raspberry Pi risks falling apart because it will no longer be backwards-compatible and a new chip (like Mali) would have employees in the RPF tied up for months trying to learn how to use the new chip, causing updates and repairs to slow to a crawl. It would also open the door to Raspberry Pi Clones and seriously damage the Raspberry Pi's reputation.
We know that the VideoCore IV is now pretty much at it's limits which might imply we won't see that on the Pi4. As to whether it's replacement would be a newer VideoCore or something else might be an interesting debate. I would expect RPF to choose the best option available for a few years worth of upgrades.

As for the clones I think the Pi is fairly safe from most of them. The area the Pi really wins out is having good ties into the hardware side of things (they've got the datasheets and staff to actually make things happen) and the support. Most of the other board manufacturers are using AllWinner chips and there seems to be little interest from AllWinner and the (mostly Chinese) designers and manufacturers to provide any real, ongoing support.

Whilst people have come to expect backwards compatibility across the range history has shown that always trying to maintain backwards compatibility between many versions may lead to more issues than it solves. It could be argued that we're already seeing that with trying to have the same image for the three SoCs currently in use. If we're looking at a major SoC re-design then this might also be the time for breaking some backwards compatibility with an aim to keeping good compatibility between the Pi4 and a few of it's successors).

I shall leave you with an old Windows Joke (I was reminded of it thinking about backwards compatibility)
Windows: A 32 Bit Shell For A 16 Bit Operating System, Originally Written For An 8 Bit Processor On A 4 Bit Bus By A 2 Bit Company That Can't Stand 1 Bit Of Competition!

I'm sure the original version I saw described some of the levels as Bodge and Hack, but I can't find that version.

User avatar
RaTTuS
Posts: 10481
Joined: Tue Nov 29, 2011 11:12 am
Location: North West UK

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 7:21 am

http://apps.charitycommission.gov.uk/Sh ... ryNumber=0 is what the Foudation has to play with
How To ask Questions :- http://www.catb.org/esr/faqs/smart-questions.html
WARNING - some parts of this post may be erroneous YMMV

1QC43qbL5FySu2Pi51vGqKqxy3UiJgukSX
Covfefe

BMS Doug
Posts: 3824
Joined: Thu Mar 27, 2014 2:42 pm
Location: London, UK

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 7:34 am

RaTTuS wrote:http://apps.charitycommission.gov.uk/Sh ... ryNumber=0 is what the Foudation has to play with
As I understand the structure It is Raspberry Pi Trading that undertakes product development, once the money has been passed to the Foundation it must be spent on the charitable goals of the foundation.
This is what Raspberry Pi trading has to play with



extract from Raspberry Pi Foundation Strategy
Raspberry Pi Foundation was established in 2008 as a UK-based charity with the purpose “to further the advancement of education of adults and children, particularly in the field of computers, computer science and related subjects”.
Through our trading subsidiary, (Raspberry Pi Trading Limited), we invent and sell low-cost, high-performance computers that people use to learn, to solve problems, and have fun. Between launching our first product in February 2012 and our fourth birthday in February 2016, we sold over eight million Raspberry Pi computers and helped to establish a global community of digital makers and educators.
In October 2015 the Foundation merged with Code Club, a network of volunteer-led after-school coding clubs for 9-11 year olds.
In November 2015 we launched the world’s first $5 computer, the Raspberry Pi Zero.
We use profits generated from our commercial activities to pursue our educational goals; we also receive funding and in-kind support from generous partners and donors that share our mission.
I can see that someone might say that creating a new generation of Pi would advance the goals but it would be difficult to justify it as being as cost effective at advancing those goals as spending the money directly on trainers / Raspberry pi giveaways.
Doug.
Building Management Systems Engineer.

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 23867
Joined: Sat Jul 30, 2011 7:41 pm

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 8:41 am

Had a bit of a thread cleanup. Keep it polite people.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I think it’s wrong that only one company makes the game Monopoly.” – Steven Wright

jamesh
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 23867
Joined: Sat Jul 30, 2011 7:41 pm

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 8:44 am

Just as an aside, Pi4 wishlists are almost pointless. You should be thinking about Pi5 and above at least. The RPF/T has a roadmap (and it has some very surprising bits in it!), and it takes a LONG time to develop new stuff.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Contrary to popular belief, humorous signatures are allowed. Here's an example...
“I think it’s wrong that only one company makes the game Monopoly.” – Steven Wright

hippy
Posts: 6068
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 10:59 am

RaTTuS wrote:http://apps.charitycommission.gov.uk/Sh ... ryNumber=0 is what the Foudation has to play with
As noted repeatedly in the past; Raspberry Pi (Trading) Ltd is a commercial venture which is not bound by any charitable obligations of the Raspberry Pi Foundation. So long as the goals of the Foundation are pursued by the Foundation there are few restrictions on how the Foundation achieves its goals. It can engage in almost any commercial venture, expenditure or investment in pursuit of those goals. Raspberry Pi (Trading) Ltd as a commercial venture has even fewer restrictions on what they can do.

TL;DR The Foundation may be a charity but that places very few restrictions on what it can do.

I am surprised how often this has to be stated, and why people keep bringing up the Foundation's status as a charity as if it is somehow a major restriction in what they can do. It is not.

fruitoftheloom
Posts: 20900
Joined: Tue Mar 25, 2014 12:40 pm
Location: Delightful Dorset

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 11:09 am

hippy wrote:
RaTTuS wrote:http://apps.charitycommission.gov.uk/Sh ... ryNumber=0 is what the Foudation has to play with
As noted repeatedly in the past; Raspberry Pi (Trading) Ltd is a commercial venture which is not bound by any charitable obligations of the Raspberry Pi Foundation. So long as the goals of the Foundation are pursued by the Foundation there are few restrictions on how the Foundation achieves its goals. It can engage in almost any commercial venture, expenditure or investment in pursuit of those goals. Raspberry Pi (Trading) Ltd as a commercial venture has even fewer restrictions on what they can do.

TL;DR The Foundation may be a charity but that places very few restrictions on what it can do.

I am surprised how often this has to be stated, and why people keep bringing up the Foundation's status as a charity as if it is somehow a major restriction in what they can do. It is not.
Point of note, Raspberry Pi Trading must give all its profits to the Raspberry Pi Foundation

Point of note, the Raspberry Pi Foundation is accountable to the UK Charities Commission and therefore does have restrictions which make it a Charity.
Retired disgracefully.....

ShiftPlusOne
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 6020
Joined: Fri Jul 29, 2011 5:36 pm
Location: The unfashionable end of the western spiral arm of the Galaxy

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 11:16 am

RPTL is owned by RPF. It's not going to spend money on things which don't benefit RPF.

fruitoftheloom
Posts: 20900
Joined: Tue Mar 25, 2014 12:40 pm
Location: Delightful Dorset

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 11:20 am

ShiftPlusOne wrote:RPTL is owned by RPF. It's not going to spend money on things which don't benefit RPF.
That is the point which most people fail to grasp about UK Charities.

A Charity has.a Trading Arm to generate income for the Charity to undertake its stated remit....
Retired disgracefully.....

hippy
Posts: 6068
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 11:35 am

ShiftPlusOne wrote:RPTL is owned by RPF. It's not going to spend money on things which don't benefit RPF.
But everything RPT does to benefit the RPF is of benefit to the RPF so that doesn't really restrict what RPT can do or how they choose to do it.

Whilst engineering investment may take away what RPT can give the RPF in the short term, so long as it recuperates that or it generates more in the long term, that is entirely acceptable. If money is 'thrown away' but it generates more sales then that is equally acceptable.

It's that 'they can't do that because they're a charity' which is most often wrong.

hippy
Posts: 6068
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 11:45 am

fruitoftheloom wrote:Point of note, Raspberry Pi Trading must give all its profits to the Raspberry Pi Foundation
Point of note; the key word is "profits". That doesn't dictate how they spend or invest their money. There's not even any requirement to generate profits though it's in everyone's interests that they ultimately do.
fruitoftheloom wrote:Point of note, the Raspberry Pi Foundation is accountable to the UK Charities Commission and therefore does have restrictions which make it a Charity.
Point of note; those restrictions do not apply to Raspberry Pi (Trading) Ltd and the restrictions on the charity are far less onerous than some people like to make out.

A charity must indeed pursue its goals but there are very few restrictions on how a charity goes about pursuing those goals.

hippy
Posts: 6068
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: Why the Raspberry Pi 4 wishlists are tough sells

Wed May 10, 2017 12:00 pm

fruitoftheloom wrote:That is the point which most people fail to grasp about UK Charities.
It does seem there are quite a few people who do not understand how UK charities or the charity commission work, nor understand the frameworks and legislation they work within. Many seem to have a vision of how things are or should be which does not reflect the reality of the actual situation.

Can I ask what your particular expertise is in this area ?

For my part I have only had associations with charities rather than running them. I have never been a trustee of a charity though I have been a member of committees running non-profit organisations and have had association with others.

Return to “General discussion”