Re: Raspberry Pi 3B and pitiful network speeds

From: Mark Millard via freebsd-arm <>
Date: Mon, 21 Jun 2021 05:31:39 UTC
On 2021-Jun-20, at 20:27, MJ <mafsys1234 at> wrote:

> On 21/06/2021 1:04 pm, Mark Millard wrote:
>> On 2021-Jun-20, at 19:16, MJ <mafsys1234 at> wrote:
>>> On 21/06/2021 7:29 am, Denis Ovsienko wrote:
>>>> . . .
>>>> As a separate note, in my setup ifconfig tells "100baseTX
>>>> <full-duplex>", which is expected for RPI3B. In the original message it
>>>> tells "1000baseT <full-duplex>", which means the board is RPI3B+ or
>>> Hi Denis.
>>> Yes, you're correct. This is the 3B+. I apologize for misrepresenting it. It is also the earlier version board,
>>> Built in 2017. I think it's version 1.1 (it's in a case, I don't feel like dismantling it).
>> I do not have access to such for any testing, just a 3B. (Not
>> a reasonable comparison.) In checking on that, I ran into the
>> following information:
>> only shows one revision for 3B+:
>> Code	Model	Revision	RAM	Manufacturer
>> . . .
>> a020d3	3B+	1.3	1GB	Sony UK
>> . . .
>> Under RaspiOS the revision code shows up via: cat /proc/cpuinfo
>> It was released in 2018. (See .)
> Here's the specifications:
> This is the exact model. It was, I believe v1.1. Later release was v1.2
> If you do need convincing, I'm prepared to pull it apart and photograph it... :-)

I do not expect that it will show any v1.? at all on
the PCB, unfortunately.

It is common for boards released towards the start of year Y
to have PCBs that reference year Y-1. So I do not expect
that a 2017 helps identify much either.

The only thing I know of that might well identify the v1.?
vintage is to find the revision code via software, and see
if it is a020d3 vs. something else. If it is a020d3, then
the table at:

should apply and it would be v1.3 . Otherwise, the table is useless
for the issue, having no other documented 3B+ codes. (They document
that "there may be codes in use in the field that are not in this

It does not look like pulling it apart is likely to be worth it.
If a check is made, the software check of the revision code looks
more reasonable to me. But it is no big deal if you do not bother
with such.

Mark Millard
marklmi at
( went
away in early 2018-Mar)