W. H. Heydt wrote:
Point 1: The SoC for the Pi4 really can't be used as--effectively--a drop in replacement for the BCM2837 from the Pi3. Even if you just consider the number of external memory address lines, that's obvious.
Point 2: On the RAM pricing... Prior to the Pi4B, all Pis used--if I understand it correctly--LPDDR2 while the Pi4B use LPDDR4. Did you take that into account in your pricing study?
Point 1: I agree. The board would need some redesigning (RAM moved to front, LPDDR4). I don't know what you mean by memory address lanes, unless you thought I meant keeping the 512MB of LPDDR2, which I don't.
Point 2: 4B uses much faster LPDDR4, whereas the earlier models use older, slower LPDDR2. There are several reasons I don't think this will be a problem.
1. In a statement I believe to the MagPi (though it could have been an engineer on here) a few years ago, RPT said they might have to upgrade RAM someday not just because of the speed, but as LPDDR2 becomes more niche, it actually becomes more expensive per chip, while as LPDDR4 becomes more widely used, it becomes cheaper. The difference between LPDDR2 and LPDDR4, in 2019, might
be extremely close.
2. The 3A+ and 3B+ uses one hub chip for Ethernet and USB, whereas the 4B uses two chips (one for Ethernet, one for USB over PCIe). Thus, I would assume that having two hub chips on the 4B costs more than the old one hub chip on the 3A+ and 3B+ and earlier. The cost savings of removing the two hub chips, then, is much greater - and should help offset any added RAM costs. I don't know the official BOM pricing, I just think it's reasonable to assume that two hub chips is roughly twice the cost of one (even though they are different models).
A Pi4A would be the same as the 3A+ except with the 2711 and faster DDR4.
Would there be a requirement for a faster 3A+?
Vision, robotics apps?
Only those bumping into the 3A+ limitations would know.
The USB-C is needed on the Pi4B because that needs extra current out the USB3 ports.
Not needed on a 4A.
So does the 4A need two HDMI?
Wireless dual monitors?
The Model A is known heavily for its use in digital signage. A few years ago, there was one December where a "digital signage" company bought out the entire inventory - you couldn't buy an A for a few months after that. Eben Upton has described the dual-HDMI support as a much-requested feature by digital signage companies. I think it is very reasonable that they would try to satisfy the demands of a huge market for the A by adding two HDMI.
And if they are adopting two HDMI, they might as well adopt the USB-C power with it, even if it isn't quite necessary. It doesn't make sense to have a third I/O configuration combining microUSB + two HDMI. That would be just confusing and unnecessary. microUSB+1HDMI and USB-C+2HDMI is enough.
But then I would want 4GB ram to use it as desktop
Another 4 model SKU A's?
I would seriously doubt it, as adding multiple SKUs of anything is a logistics challenge. If you want 4GB of RAM, I would seriously not expect it to appear on an A model. Buy a 4B+.
Also the uHDMI is excellent too, best move (Zero should have used it ages ago at release)
Agreed. I would have hoped that the 4B+ used miniHDMI because it is more stable and the Zero uses it, but what fits is what makes the decision. In this case, I would wish
(my only wishlist here, the rest is just theories) that the RPF would switch the Zero from miniHDMI to microHDMI, for consistency's sake. But that might cost too much which would break the $5 price, in which case I say keep it. Two $5 dongles is nothing to cry about.
Add the Raspberry stamp back to the WiFi module, I really liked that little touch now missing from the Pi4
If one of the Engineers knows why it is gone, I would love to know. I will miss that.
Many thanks to the Pi Foundation for all the wonderful products that give us all kinds of options for what ever we are building.
Yes, the RPF deserves our heartiest thanks!