Clearly it's "Nostradamus reading the runes" as RPF/RP(T)Ltd. aren't in the game of pre-announcing anything (after they got burned over the "Official" 7" screen).
I don't know any specific dates, but I can see from my crows nest/partition, the team working on future products (including myself). It's no secret there is going to be Pi4. But it's still a while* off.
No. Why put time and money into changing a thing that works? Unless you are developing and growing your application and find it needs more performance, memory or whatever.Should I go on with the hardware upgrade...
No. Waiting is boring....or wait for a RPi 4?
PI2B is 900MHz, not 800MHz. It is not easy to force CPU to run at highest speed. I did and did one problem sample heavy computation single core integer performance investigation. And there PI3B+ is only twice as fast as PI2B. So unless you need that (up to) factor 2 boost there is nothing wrong staying on PI2B (I did buy a PI3B+, but only because the PI2B and PI3B I use are borrowed):
If you upgrade it appears you will need to also upgrade your power supply. No one knows what a Pi 4 will be or when it will arrive.
Not strictly true. Some people have a very good idea what the Pi4 will be*, but not necessarily when it will arrive.
From what the OP said about getting intermittent undervolt indication on his Pi2B, he needs to upgrade his PSU (or his power lead, if separate) even if he doesn't upgrade the Pi. He shouldn't be getting undervolt on a Pi2B with a 2.5A PSU...ever.
I'd rather trust what jamesh says than some guesswork based on random special offers.RichardS wrote: ↑Wed Jun 06, 2018 4:11 pmArrow Electronics just GAVE me a FREE Raspberry Pi 3b+ for ordering only $50 dollars worth of other stuff - not just FREE shipping a TOTALLY FREE $35 RASPBERRY PI 3B+
The $50 dollars didn't count the Pi3b+ itself, but that's still pretty incredible when you consider that they ALSO threw in FREE overnight FEDEX shipping on the whole order.
This worked out well for me, because I had some stuff I wanted anyway, but I would note that last time Arrow did something like this was to heavily discount the plain Pi 3b just a few weeks before the Pi 3b PLUS came out.
So we may be seeing a Pi 4 sooner than you think.
https://www.raspberrypi.org/blog/eben-q-a-2
I agree with that assessment with the caveat that Eben probably already had a pretty fair idea of what was feasible before making that statement and what he was looking for were minor "around the edges" possibilities. While I don't really *expect* a Pi4B around the next anniversay, neither would I rule it out...at this time.hippy wrote: ↑Wed Jun 06, 2018 5:50 pmhttps://www.raspberrypi.org/blog/eben-q-a-2
"With 3B+ out of the way, we’re going to start looking at this now. For the first six months or so we’re going to be figuring out exactly what people want from a Raspberry Pi 4. We’re listening to people’s comments about what they’d like to see in a new Raspberry Pi, and I’m hoping by early autumn we should have an idea of what we want to put in it and a strategy for how we might achieve that".
From that I wouldn't be expecting anything this year. It might fit a schedule for release next Pi Day or launch anniversary but I think that's optimistic.
I'm surprised. I thought that some of the Pi3B+ changes not related to WiFi, like the more powerful CPU, and POE (Power Over Ethernet) capability had already filtered down to the Pi2B, since logically this is the model that MOST needs faster Ethernet, and MOST needs POE, since it has no WiFi.W. H. Heydt wrote: ↑Wed Jun 06, 2018 6:40 pmI agree with that assessment with the caveat that Eben probably already had a pretty fair idea of what was feasible before making that statement and what he was looking for were minor "around the edges" possibilities. While I don't really *expect* a Pi4B around the next anniversay, neither would I rule it out...at this time.
On the other hand, I still think that a Pi3A+ would be a good candidate for late this year or early next year. Another possibility would be a quiet, though rather noticeable, migration of the Pi2B from the A0 to the B0 iteration of the BCM2837. However, either of those will depend largely on sufficient foresight for there to be enough BCM2837B0 SoCs on hand as not to impact Pi3B+ production.
It's the same SoC, just flipped over and mirrored (keeping the contacts in the same places) with a heat spreader and using the PCB as a heat sink. The improved thermal handling permits a higher CPU clock. A "de-rated" (probably chips that a binned differently because they don't test at 1.2GHz) is used on the revised Pi2B, the v1.2 board. In theory, I'm reasonably sure that the RPT could use the B0 stepping as used on the Pi3B+ on the Pi2Bv1.2 board, though it would make more sense to redesign the board for the better thermal conditions and--as you note--the PoE circuitry. In "Off Topic" there is a thread discussing whether or not the Pi3A+ would be a good idea. You might start a similar thread making the case for a Pi2B+. Of course, PoE support would be pointless on a Pi3A+...RichardS wrote: ↑Thu Jun 07, 2018 1:50 amI'm surprised. I thought that some of the Pi3B+ changes not related to WiFi, like the more powerful CPU, and POE (Power Over Ethernet) capability had already filtered down to the Pi2B, since logically this is the model that MOST needs faster Ethernet, and MOST needs POE, since it has no WiFi.
The Pi2Bv1.2 runs at the same 900MHz that the Pi2Bv1.1 did, even though it has the same SoC design that the Pi3B does. So it's not *just* a Pi3B without WiFi. It's also slower, and therefore has fewer thermal issues under load. That does mean, though, those chips that don't test as good at full speed can be used to make Pis thereby increasing the effective manufacturing yield and reducing per-chip cost through fewer that have to be thrown away.I suppose you could use a distributed network of Pi3B+ models as "access points" in an application using both powered Ethernet and WiFi, but I have always felt that the whole point of the Pi2B was to offer equivalent performance MINUS the WiFi, specifically for those corporate applications where wired networking is used in preference to WiFi because WiFi networking is not permissible due to security concerns - and given this situation, it would seem that it's actually the Pi2B model that would have most benefited from GigE and POE.
Power over Ethernet is a triviality with 100BT because there are two unused twisted pairs. Upgrading to gigabit buys you at most a 3x performance increase, in the case of my setup only 2x, and a whole slew of flow control requirements because the gigabit adapter can not actually receive at gigabit speeds
The original 3B barely works at 1.2Ghz, not because of thermal cooling which can be easily supplemented, but because of the power regulation circuitry. This is the invisible upgrade that makes 1.4 Ghz reliable on the 3B+.W. H. Heydt wrote: ↑Thu Jun 07, 2018 4:19 amI'm reasonably sure that the RPT could use the B0 stepping as used on the Pi3B+ on the Pi2Bv1.2 board, though it would make more sense to redesign the board for the better thermal conditions and--as you note--the PoE circuitry.
I would not expect a third revision of the Pi2B (using the B0 SoC) to default to 1.4GHz. More likely would be something around 1GHz or 1.1GHz. Moving to GbE would require using the hybrid LAN chip, which would not necessarily be the choice on a new Pi2B. Basically, I see the the Pi2B as a modestly slower, lower power, non-WiFi (for when that is desirable) variant at this point. The Pi2B (both versions) are, in some respects the answer to those that don't think a Pi between the Pi0/Pi0W, like a Pi3A+, and the Pi3B/Pi3B+ makes sense. The Pi2B is already occupying the upper part of the gap. A Pi3A+ would occupy the middle or lower portion of the gap.
That was pretty much my understanding of things. Some Industrial users have designed things around the 2B (potentially including regulatory testing) so the change in SoC for the 2B was to allow 2B boards to be made using a part (the BCM2837) that was available rather than the BCM2836 which presumably isn't required in the numbers needed to make future fabrication runs viable.
The B+ at least benefits from being the same SoC (BCM2835) as the Zero/ZeroW, A+, and original CM so whilst they're in production (CM had a fairly long availability date) the B+ can presumably be fairly easily manufactured as well.
I'm sure there have already been comments elsewhere about the SoC design having been thought about (and potentially some thought already happening for a Pi5) so I suspect there's some good ideas of what could be on the sillicon with potentially some developmental runs done already. I'm wondering if the SoC is mostly finalised potentially with lots of goodies on it but decisions about what parts of that will actually get used in the Pi4 release with other things waiting for a future update (as prices make things viable)W. H. Heydt wrote: ↑Wed Jun 06, 2018 6:40 pmI agree with that assessment with the caveat that Eben probably already had a pretty fair idea of what was feasible before making that statement and what he was looking for were minor "around the edges" possibilities. While I don't really *expect* a Pi4B around the next anniversay, neither would I rule it out...at this time.
I'm sure that's the case. I will refrain from repeating my remarks on what I think are reasonable expectations for features for future Pis as this thread has drifted far enough already and speculative posts tend to bring out people who object to them and the threads rapidly go down hill from there.mfa298 wrote: ↑Fri Jun 08, 2018 12:21 pmI'm sure there have already been comments elsewhere about the SoC design having been thought about (and potentially some thought already happening for a Pi5) so I suspect there's some good ideas of what could be on the sillicon with potentially some developmental runs done already. I'm wondering if the SoC is mostly finalised potentially with lots of goodies on it but decisions about what parts of that will actually get used in the Pi4 release with other things waiting for a future update (as prices make things viable)
