ejolson
Posts: 1885
Joined: Tue Mar 18, 2014 11:47 am

Re: The Case for the Pi3A+

Mon Jun 11, 2018 6:32 pm

W. H. Heydt wrote:
Mon Jun 11, 2018 6:10 pm
If an updated A Series Pi is "a long way away", then an updated Pi0/Pi0W is a much, much farther away than that.
One would have thought the same with the original Zero, except version 1.2 and then the Zero W came along.

A Zero+ with 64-bit capable SOC, even if only one core, would help prepare for the switch to a 64-bit Raspbian when and if it happens. I don't have a solution to the expense of putting the memory on the other side of the circuit board. Maybe the Zero+ would be slightly bigger or more expensive than current models.

W. H. Heydt
Posts: 8866
Joined: Fri Mar 09, 2012 7:36 pm
Location: Vallejo, CA (US)

Re: The Case for the Pi3A+

Mon Jun 11, 2018 8:58 pm

ejolson wrote:
Mon Jun 11, 2018 6:32 pm
W. H. Heydt wrote:
Mon Jun 11, 2018 6:10 pm
If an updated A Series Pi is "a long way away", then an updated Pi0/Pi0W is a much, much farther away than that.
One would have thought the same with the original Zero, except version 1.2 and then the Zero W came along.
Yes...at least regards to adding the CSI mini-connector. Even there, though, it was a "known quantity" since it is used on the CMIO boards. Also, the Pi0W uses the same SOC as the Pi0. Thus, the Pi0v1.2 and the Pi0W are incremental improvements, rather than being wholly new designs.
A Zero+ with 64-bit capable SOC, even if only one core, would help prepare for the switch to a 64-bit Raspbian when and if it happens. I don't have a solution to the expense of putting the memory on the other side of the circuit board. Maybe the Zero+ would be slightly bigger or more expensive than current models.
A 64-bit Pi0 series board would be entirely new, so very nearly the full engineering effort. I have pointed to a solution (and the inherent problem) to keeping the RAM chip on the same side as the SoC, and that is to use--as is done on all BCM2835 boards--PoP RAM. The issue then becomes one of TDP. Using a single 64-bit core would help there. Shifting to 28nm would help there. Reducing clock speed would help there.

The biggest issue then becomes...where do you get the SoC that meets those criteria? There isn't enough profit (so far as is known outside of Pi Towers) to finance a new chip from what is made on Pi0/Pi0W sales. Unlees and until someone comes along and requests a chip from Broadcom that happens--incidentally--to met the specified criteria, or someone finds an incredibly cheap way to develop a new SoC (say, by doing to a Pi4 SoC sort of the reverse of what was done to make the Pi2 SoC), it simply isn't going to happen.

I think this is part of why this thread has survived. Nothing here is proposing that the RPT/RPF do anything *new*, per se. The idea of a Pi3A+ is to use things the RPT has already done and then cut it back to an A+ format, which should--as such things go--be a comparatively minor effort. It's really (I think) a matter of taking a Pi3B+ design and then removing unwanted/unneeded stuff, specifically the LAN chip and the PoE, and even there, the PoE header holes could be left in place and just not have anything connected to them, save for the one line there used for the "reset" function.

ejolson
Posts: 1885
Joined: Tue Mar 18, 2014 11:47 am

Re: The Case for the Pi3A+

Mon Jun 11, 2018 11:46 pm

W. H. Heydt wrote:
Mon Jun 11, 2018 8:58 pm
where do you get the SoC that meets those criteria?
I would suggest using possibly defective BCM2837s with 3 of the 4 cores disabled or whatever new SOC is designed for the Pi 4 again with all but one core disabled.

User avatar
Gavinmc42
Posts: 2107
Joined: Wed Aug 28, 2013 3:31 am

Re: The Case for the Pi3A+

Tue Jun 12, 2018 2:18 am

Could the 3A+ use the older BCM2837?
As the 3B+ becomes the top of the line, does that free up usage of the older part?
64 bit 4 cores is more important than max speed in my opinion.
Does it need the new WiFi combo chip?

Call it a 3A and the 3A+ has the faster WiFi/CPU.
Actually would that make it a Pi2AW?

RPT has a bunch of engineers but not enough to do multiple spins at the same time?
The hint is there that they are very busy, makes me wonder what on ;)

There is still a hole for smaller DSI displays and the Zero does not have a DSI but the A does.
Who ever got 100k A's may be interested in faster version too?

But good software can make up for a lack of speed.
I don't use Linux anymore for gadgets and the single core 2835 is plenty fast enough, somewhat of an overkill :D
I am still learning AI/CV/ML for smart gadgets where the 4cores, 64bit/NEON is more useful but that is at least 2years away.

The Zero has filled the space where the A is for me and there is in effect three versions of Zero's.
Would a lower power Zero be of more use? For battery powered gadgets?
There is still plenty of life left in the VC4, lots more tricks to be done with it in the embedded market.

Everyone says the IoT market is going to be much bigger than the PC market.
100k A's is not a small number, but it is tiny when compared to the possible total IoT market.

Have not checked, is it 20Million Pi's yet?
Hope those RPT guys are training the next gen of engineers, er whoops, I nearly forgot the mission :lol:
Does anyone predict any slow down in the next 5-10years?
Wonder how many models there will be then?
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

Return to “Off topic discussion”

Who is online

Users browsing this forum: No registered users and 5 guests