Raspberry Pi - make Tier 1

weberjn

New Member

Reaction score: 2
Messages: 5

If there was an affordable little computer with over 25 million units sold, that millions of Unix enthusiasts use, and I had a great OS, then I'd make that little computer my first priority and raise lots of new disciples of my OS.

So, why is Raspberry Pi treated as the poor relation, without possibility of updates?

 

ucomp

Active Member

Reaction score: 31
Messages: 186

So, why is Raspberry Pi treated as the poor relation, without possibility of updates?
Can't you read what has already been answered to you to the same topic? :
#27
You should go to a Raspbian-Forum if you want something like Tier1 for you "first priority toy-computer" -
from my personal view the RPI doesn't fill the HW-requirements for a stable professional FreeBSD-Server so why should I expect the next revolution from this toy???
Developers who work day&night on a Tier1-architecture have to be responsible for a stable platform and do not work on platforms which are predestined to fail in some(important) disciplines ....


--- just to add this post which describes some reasons why NOT Tier1 for RPI : ---
#63
 
OP
OP
W

weberjn

New Member

Reaction score: 2
Messages: 5

from my personal view the RPI doesn't fill the HW-requirements for a stable professional FreeBSD-Server so why should I expect the next revolution from this toy???
from my personal view the Raspberry Pi does fill the HW-requirements for a stable toy FreeBSD-Server where lots of hobbyists could learn to administrate a BSD system. But giving them a system that gets no updates is not responsible.

Instead the millions of hobbyists learn Debian and will of course later in their jobs go the Linux way they know and opt for a RHEL server instead of a stable professional FreeBSD-Server they never got to know.
 

ucomp

Active Member

Reaction score: 31
Messages: 186

..........Instead the millions of hobbyists learn Debian and will of course later in their jobs go the Linux way they know and opt for a RHEL server instead of a stable professional FreeBSD-Server they never got to know.
who told you that you need an RPI
for learning FreeBSD-administration ?
please give us a link to that claim
or are you the 1st one worldwide who believes that ?
You can learn FreeBSD on nearly every available computer,
I'm sure in every ("West-World"-) household there is at least 1 computer
which is Tier 1 -supported by FreeBSD.
And if not you can run FreeBSD in a VM on your Computer
or for a few bucks in the cloud (Tier 1-supported , of course).

--edit: --
surely you can make a beginner mistake,
I do not care but maybe it would help you to read this for your future-posts:
 

SirDice

Administrator
Staff member
Administrator
Moderator

Reaction score: 7,562
Messages: 30,279

It's the architecture that's going to be Tier 1, not a specific device. In order for ARM and ARM64 to become Tier 1 many more devices besides the Raspberry Pi need to be supported.

 

mark_j

Member

Reaction score: 10
Messages: 50

If there was an affordable little computer with over 25 million units sold, that millions of Unix enthusiasts use, and I had a great OS, then I'd make that little computer my first priority and raise lots of new disciples of my OS.

So, why is Raspberry Pi treated as the poor relation, without possibility of updates?

Most probably because it's not a 'server'. However, this architecture is moving towards servers, so that's now why FreeBSD seem to be getting serious about it.

See: https://www.freebsdfoundation.org/project/freebsd-armv8-64-bit-arm-port/

So, it seems, should the foundation not have formulated an idea that Arm was going to be mass adopted into the server market, it would still be being treated as a second-rate citizen and languishing where it is now.

Like you I hope the outcome is a better supported Arm architecture and therefore Pis.
 

UnivProc

New Member

Reaction score: 2
Messages: 3

who told you that you need an RPI
for learning FreeBSD-administration ?
please give us a link to that claim
or are you the 1st one worldwide who believes that ?
You can learn FreeBSD on nearly every available computer,
I'm sure in every ("West-World"-) household there is at least 1 computer
which is Tier 1 -supported by FreeBSD.
And if not you can run FreeBSD in a VM on your Computer
or for a few bucks in the cloud (Tier 1-supported , of course).
While your response is an accurate response to a limited problem. I believe that the value of an RP4 as an educational tool is understated by weberjn. An RP4 has sufficient capability to be the standard processing node within a university's CS and engineering education programs. It is less expensive than a text book and it addresses a number of fundamental problems with instructional computational systems:
(1) Commonality - telling a student to type A to generate result B (having a specifically common platform addresses this issue) is very useful. Having all students load a specific version of FreeBSD on an RP4 would be an excellent solution (if updates are performed, they are scheduled for an entire class). This type of hand-holding may seem excessive to the pros that frequent these forums. I can assure you that for instructional purposes it is not. I have a reasonable level of expertise and I am routinely stumped by student questions relative to systems with which I have limited experience.
(2) Computational Physicality - the I/O pins on a RP are invaluable to to making computing "real". When extending software education beyond CS and EEs, it is very useful to be able to demonstrate usefulness, like directly controlling a motor in the student's hands, meaduring distances using sonar, decerning visible flicker on an LED, etc. The daunting task of impressing on chemical, civil, or mechanical engineering student the importance of being computer literate is made somewhat easier by the ability to physical-ize computation.
(3) Oops - Allowing a student to play within the bowels of a system can have untoward results. Using a "toy" has the advantage that it is difficult to break and easily repaired (insert a new microSD card). Instead of avoiding tasks that risk system damage, we can welcome them. Loosing a RP4 teaches the same lesson that loosing their MAC Pro would teach, without the irate calls from parents.
(4) Networks with lots of nodes that can experience oops - if students are doing security work, it is best if a compromised system is easily speculatively restored (I do not believe you compromised my system, but I am going to re-flash it just in case).

There are more, but I want to avoid TLDR (I know, I have already failed).

I believe that weberjn makes a very valid point. IBM's success in the 70s and 80s is, in part, attributed to their aggressive academic model that made it easy for students to have access to IBM computers. Unix, C, Python, etc. have similar linkages to their early academic penetration. Ignoring the short-term market driver criteria and vigorously supporting RP4-FreeBSD could result in substantial long-term value to the FreeBSD community. Waiting to see if this market fully develops before attempting to address will irretrievably surrender it to Linux. A reasonably complete (C, C++, Java, Python, SQL, network (Ethernet, USB, WiFi), web, graphics (HDMI), and pin header - basically all the RP4's physical external interfaces) software and board support in a standard downloadable image is the starting point.

Just a thought. I apologize profusely, in advance, if my opinion at all insults, disrespects, discomforts, disagrees, or otherwise negatively thesaurus-izes you.
 

mark_j

Member

Reaction score: 10
Messages: 50

You are correct UnivProc, I would also add, that contrary to what ucomp said (https://forums.freebsd.org/threads/raspberry-pi-make-tier-1.71378/#post-432116) about using a VM, this is not enough.

When using a real physical machine, especially a low-cost ARM device like the Pi, you can leave it running all day/night. You can practice using cron to schedule jobs, perform backups, host a database, host a web based bulletin board and so on. This allows students to learn more than just the nuts and bolts of an OS but also all the ancillary bits that make day to day administration both hum-drum and exciting. All the while, costing you maybe $20 PA to run. Compare this to even a modest PC consuming around 90W per hour, which has thermal considerations as well as noise and I think the Pi is an invaluable tool.

Of course, it's also hobbyist and extremely versatile. It needs more respect from FreeBSD! ;)

We had noisy sun workstations in our classes back in the days. The fan whine was excruciating. You left with a headache! :(
 

ucomp

Active Member

Reaction score: 31
Messages: 186

Hi UnivProc , Hi mark_j ,

1.: thanks for your postings but you confuse what we really were talking about:
I didn't talk about FreeBSD on the RPI itself ,
I own an RPI and , more important :
The RPI IS SUPPORTED BY FreeBSD
even the FreeBSD foundation fell in love with the RPI , they even got to bed with the PI and give it a good-night-kiss every day 😂 just look here :

... It needs more respect from FreeBSD! ...
even more respect ? perhaps rename FreeBSD to rpiBSD ;-) .. lets discuss this :)

What I/we really discussed about is the Tier-status of aarch64 .

the initial reference platform for Arm64 is NOT THE RPI , here's one( of a lot) hint again (from developers view) why that could be the case : #63

So from the developers - view the Tier-status of a platform is not an emotional question,
but to stay in emotional terms ;-) :

...VM, this is not enough.

... $20 PA..
my 2 $/month CloudVM at full throttle - data center connection on Tier 1 FreeBSD will totally outperform this snore box of RPI 😂
 

mark_j

Member

Reaction score: 10
Messages: 50

You missed my point. I was arguing your point of running FreeBSD in a VM for all the reasons I stated.
I should add, I did a calculation:
Pi @ 5W peak * 24 * 365 * .19 KW/h = $8.32 PA (just roughly, of course as I have no idea what KW/h costs for you or others.)

I'm also not comparing some cloud or VPS system to the Pi, just stating its use in classroom education. Plus you can tinker with it (if only FreeBSD would fix SPI, but that's another issue).

Now, show me a cloud solution like that... 😵
 

ucomp

Active Member

Reaction score: 31
Messages: 186

...
. Plus you can tinker with it (if only FreeBSD would fix SPI, but that's another issue)....
Who or what keeps you from tinkering, there's even a report in the FreeBSD journal about it( #9 ).
Who or what prevents you from visiting the mailing lists and discuss/fix SPI-related things?
FreeBSD is a community-project, it supports the RPI and you`re welcome, I really do not understand your problem. The discussion was about the completely unfounded demand (see thread title) to give the RPI Tier 1 status, (by the way , Tier- status is not assigned to any single device : #5 ).
 

ucomp

Active Member

Reaction score: 31
Messages: 186

TLDR (I know, I have already failed)
no, no.. you didn't completely ;-)
A reasonably complete (C, C++, Java, Python, SQL, network (Ethernet, USB, WiFi), web, graphics (HDMI), and pin header - basically all the RP4's physical external interfaces) software and board support in a standard downloadable image is the starting point.
afaik onboard-Wifi is not supported yet.
as said before a standard downloadable image is available.
C++(clang) is included in the "standard"- image.
Java :

According to my first tests, it could be made available to FreeBSD aarch64 (no guarantee, because aarch64 is NOT Tier1). If you have special interest in it, I will give you more information. The problem with this discussion is a bit that Tier1-demands for complete solutions are always made by people who do not know what's going on in the background.
As far as java is concerned, there is a development in the background which takes ALL BSDs into account(openbsd, netbsd, freebsd). Of course, Tier1 will be processed first.
I invite you, as an experienced man, to visit the java-BSD-project and, if you can, join the project, you will work with the absolutely best hackers worldwide :) .
Just a thought. I apologize profusely, in advance, if my opinion at all insults, disrespects, discomforts, disagrees, or otherwise negatively thesaurus-izes you.
on my part ;-)
I was so aggressive against the fu***** RPI, not nice of me 😁

Even the CoreTeam loves the RPI so much...
Mr Reuschling @theFreeBSD-journal, page 17 ( #9 ) :

The Raspberry Pi ist the perfect tool .... just read on there ..
 

msplsh

Member

Reaction score: 5
Messages: 34

Anecdote: I installed a 3B+ as a UniFi controller this week. Downloaded 12.0-RELEASE and used pkg install unifi5. Worked great, even if it was kind of slow to start up. Got a bunch of reboots until I realized that it shouldn't be powered off a PC's 500mA USB port.

Sooo, Java: 👍🤷‍♀️
 

rigoletto@

Daemon
Developer

Reaction score: 928
Messages: 1,902

The candidates to move to 'Tier 1' are ARMv8 (aka Cavium and Ampere) and POWER9. ARMv8 is quite stable already and POWER9 is not that bad.
 

UnivProc

New Member

Reaction score: 2
Messages: 3

FreeBSD is best as a heavyweight server OS so no, but I think when ARM based servers become popular ARM will be moved to tier1.
I apologize profusely for my lack of understanding and insight. What is a heavyweight server? Or, more exactly, what is there about process management, memory management, I/O management, and/or file management that is impacted in a non-linear fashion (relative to resource overhead) by a server's memory footprint, core count, or I/O subsystem density? User interfaces come in highly varying weights (e.g. XFCE/LXDE/LXQt < Gnome/KDE) and can be easily selected relative to the computational characteristics of the target platform. An RP4 is hundreds of timers more powerful than the PDPs and VAXs that birthed BSD. An RP4 is roughly as powerful as a 12 year old heavyweight server. Has FreeBSD somehow been made less efficient as it matured over the last dozen years? I apologize because I realize that this sounds snarky, which is not my intention. I understand that it is not a microkernel model, but are there restrictions on FreeBSD, that Linux does not share, in terms of stripping our unnecessary subsystems (there are some relatively lightweight Linux kernels)? Also, when a system has gig-E (I know it is not 10 or 100, but it is easily as fast as a 10 year old server), USB 3.0 (the pair represents more I/O than many older servers), dual 4k HDMI (with admittedly very under-powered graphics capability), 4gB of DRAM, and a quartet of 64-bit cores it needs most of the "big person" operating system features to function. I am clearly missing something. SORRY! (I am not asking for an academic explanation of OS functionality, I understand that much.)

It can be argued that many of computing's interesting future directions are towards lightweight systems (IoT, most embedded systems - who wouldn't prefer a FreeBSD managed toaster to one using Linux?). FreeBSD is, in part, interesting because it has a different (I believe better) licensing model. I believe the future is lighter weight, even the virtual machine future. I mean no disrespect, but this makes no sense to me (conversely, many things make no sense to me (many of them political, religious, and economic in nature and hence very much outside forum decorum)). Your statement would lead me to believe that the the superior licensed FreeBSD will become somewhat irrelevant (more accurately never become relevant) in these exploding future markets. Years ago one of my professors suggested that IBM would have an ever expanding share of an ever contracting market, with the implication that this would not be a good thing for IBM (this is not an attack on IBM or otherwise a criticism of IBM or is business and.or technological decisions - it is a simple statement that owning the mainframe market would become relatively less valuable as a percentage of the global computational market).
 

acheron

Aspiring Daemon

Reaction score: 212
Messages: 535

C++(clang) is included in the "standard"- image.
Java :

According to my first tests, it could be made available to FreeBSD aarch64 (no guarantee, because aarch64 is NOT Tier1). If you have special interest in it, I will give you more information. The problem with this discussion is a bit that Tier1-demands for complete solutions are always made by people who do not know what's going on in the background.
As far as java is concerned, there is a development in the background which takes ALL BSDs into account(openbsd, netbsd, freebsd). Of course, Tier1 will be processed first.
I invite you, as an experienced man, to visit the java-BSD-project and, if you can, join the project, you will work with the absolutely best hackers worldwide :)
I have some bootstrap and patches for java12 in mixed mode on aarch64 if you want to play with them.
 

mark_j

Member

Reaction score: 10
Messages: 50

Who or what keeps you from tinkering, there's even a report in the FreeBSD journal about it( #9 ).
Who or what prevents you from visiting the mailing lists and discuss/fix SPI-related things?
FreeBSD is a community-project, it supports the RPI and you`re welcome, I really do not understand your problem. The discussion was about the completely unfounded demand (see thread title) to give the RPI Tier 1 status, (by the way , Tier- status is not assigned to any single device : #5 ).
You probably don't understand my problem because I don't have one.
 

ucomp

Active Member

Reaction score: 31
Messages: 186

I have some bootstrap and patches for java12 in mixed mode on aarch64 if you want to play with them.
Hi Mikael,
on your Website? I‘ll find it ;-) thanks in advance!
your 11 bootstrap worked fine on FreeBSD aarch64,. as you know we have to target all_bsds and the bootstrap ELF- binaries are not compatible with openbsd/netbsd, I haven‘t worked on it the last weeks but will take a look on it when I find the time.
Regards
Klaus
 

acheron

Aspiring Daemon

Reaction score: 212
Messages: 535

Hi Mikael,
on your Website? I‘ll find it ;-) thanks in advance!
your 11 bootstrap worked fine on FreeBSD aarch64,. as you know we have to target all_bsds and the bootstrap ELF- binaries are not compatible with openbsd/netbsd, I haven‘t worked on it the last weeks but will take a look on it when I find the time.
Regards
Klaus
Yes. java11 is built in 'zero' mode, very slow compared to the 'mixed' mode.
 
OP
OP
W

weberjn

New Member

Reaction score: 2
Messages: 5

The discussion was about the completely unfounded demand (see thread title) to give the RPI Tier 1 status
My original post was about the lack of updates.

I think it would be good for students, hobbyists and the FreeBSD community, if many Pi users ran their device on FreeBSD.

But an internet-connected device and an OS, that is used by so many users, has to be secure. No need to have botnets or worms on zillion devices.

For an OS to be secure, there have to be updates.

One of the first things one should run on Raspbian is

Code:
sudo apt-get update
sudo apt-get upgrade
And with FreeBSD on Raspberry, this is currently not possible, is it?

Code:
freebsd-update fetch
freebsd-update install
So, I think FreeBSD should support the Raspberry. And it should be upgradeable.
 

SirDice

Administrator
Staff member
Administrator
Moderator

Reaction score: 7,562
Messages: 30,279

And with FreeBSD on Raspberry, this is currently not possible, is it?
Last time I tried it it worked, but because it's still Tier 2 there's no guarantee.

In any case, a source update is always possible but it's a painfully slow process on a Pi.
 

ralphbsz

Daemon

Reaction score: 1,136
Messages: 1,803

To begin with, people who claim that with the RPi4, the Arm architecture has moved into viable servers, have a somewhat unusual definition of "server". Certainly very small machines can be used as toy examples of servers, for example for instructional purposes. But the bulk of installed servers are a completely different class of machines, typically with lots of memory (server memory in data centers is today measured in TB, not in GB), with much higher IO and networking horse power (typical ethernet interfaces being deployed today are dual 100gig, and the number of PCIe Gen4 or Gen5 in the next generation of servers to be deployed next year or 2021 is counted in dozens). The RPi4 is roughly a factor of 100 behind these metrics. It remains an educational computer, or a toy for amateurs.

By the way, that doesn't mean that I'm knocking the concept of the Pi. I think I own 3 or 4 of them (two of which are sort of in production), and I greatly enjoy using them at home. But that doesn't mean that I leap to the conclusion that they are the solution to all the world's problems.

Instead the millions of hobbyists learn Debian and will of course later in their jobs go the Linux way they know and opt for a RHEL server instead of a stable professional FreeBSD-Server they never got to know.
You sound like using Linux is a bad thing. Your goal seems to be to reduce the number of Linux users, and the tool you want to use is to increase the number of FreeBSD users. I disagree: Using Linux is a good thing. As is using Windows, or a Mac. Or not using computers at all and instead doing crochet or woodworking. I have a really big problem with basing one's support of FreeBSD on one's dislike or hatred of Linux.

There is another deep question: Should it be the goal of the FreeBSD core cabal (the people who make decisions about the future of FreeBSD) to increase the number of users of FreeBSD? Or should they instead try to create a better OS for those people who are using it, even if that number is and remains small? Or maybe should they do something yet different? I don't claim to know what the correct answer to these questions is; I'm just happy that for me FreeBSD is a convenient tool for some of my tasks.

However, this architecture is moving towards servers, so that's now why FreeBSD seem to be getting serious about it.
Yes, I've been hearing about the Arm architecture moving to servers for about 10 years now. I no longer believe that it will happen, because it seems to always be forecast for "next year". In the last 10 years, all non-x86 server architectures have died one by one, and today even PowerPC is barely hanging on. Instead, we now have a healthy competition between AMD and Intel (with AMD making significant inroads), which is causing their chips to suddenly get much better. If one looks at the roadmaps for server CPUs, there has been a significant acceleration in the last 5-7 years, and it is simply because (a) AMD is trying to make an end-run around Intel, and (b) Intel is actually paying attention to improvements, because for the first time they have competition. This leap forward makes it harder for alternate server architectures (Arm, PowerPC) to make headway.

The other factor is that a significant fraction of server compute power is moving off CPUs, and into graphics processors and other exotics. This is definitely true for HPC (where a large fraction of all compute power is already on coprocessors), and it quickly becoming true of the cloud infrastructure. Compared to that seismic change, the question of CPU instruction set is somewhat secondary.

It needs more respect from FreeBSD!
I don't care at all about respect.

who wouldn't prefer a FreeBSD managed toaster to one using Linux?
I want my toaster to work, meaning make a piece of bread that is the correct color and crispiness, as efficiently as possible. I don't care at all what CPU or OS is running inside the toaster. The same applies to my dishwasher and my car. As I said above, I don't see "FreeBSD versus Linux" as a religious war.

Now, in those places where I actually need to select the OS for a computer, I will select the one that is the best tool for the job. For my small amateur home server, that happens to be FreeBSD right now. Although my home server is as much a toy for me as it is a useful appliance (I enjoy configuring and maintaining it).
It can be argued that many of computing's interesting future directions are towards lightweight systems (IoT, most embedded systems - ).
True to some extent. Traditional computing is fragmenting into multiple markets. On one side there are servers with general-purpose OSes, which are getting larger and larger, are de-facto all clustered today, and getting centralized in giant data centers run by fewer and fewer providers. We may be getting to the point where soon there is no logical reason for an individual or a corporation to own a server for their own use any longer. On the other extreme, you are right that we are getting networked and intelligent systems that are much smaller and much more common (every appliance, every thermostat, every door lock and car). But for those, it is not at all clear that general-purpose OSes are the right answer; the reliability and efficiency needs of IoT are probably better served by custom OSes, which to a large extent come from the embedded tradition. The computing devices that humans actually interact with are more and more mobile, tablets, and non-programmable devices such as Chromebooks, with tightly controlled systems for software distribution (such as app stores). This model is better for providing interfaces to humans, with less hassle than traditional "home computers" with their OS and networking hassles.

Your statement would lead me to believe that the the superior licensed FreeBSD will become somewhat irrelevant (more accurately never become relevant) in these exploding future markets.
That is quite likely. And I have no problem with it, matter-of-fact, I enjoy it. The various *BSD projects are largely driven by hobbyists and volunteers, who are doing this for their enjoyment. I really don't see why being relevant matters to hobbyists and volunteers. I would compare developing and supporting FreeBSD to crochet or woodworking. It's fun, you can exchange techniques and results with other people that have the same hobby, and it gives a sense of purpose. But unlike RedHat or IBM, it doesn't need to make money, nor does it need to achieve world domination. Just like my hand-knitted tea cozies are not intended to achieve world domination. And just like I see people with other hobbies (like gardening) as the evil competition that we need to win over.

I think it would be good for students, hobbyists and the FreeBSD community, if many Pi users ran their device on FreeBSD.
That's nice for the students, hobbyists and community members. If they really want it to happen, they are free to volunteer. I just looked at the ARMTier1 Wiki page, and there are lots of boxes where stuff needs to be done, but no people are available.

But would it be good for FreeBSD itself? I really don't know why or whether having lots of Pi users running FreeBSD should be a goal of FreeBSD.
 
OP
OP
W

weberjn

New Member

Reaction score: 2
Messages: 5

Remember AiX and Solaris? Great OSes.

Gone. Not because they're worse than Linux, but because Linux is good enough and the kids coming from school and university only know Linux, Windows and Mac. Not because Linux is free. No serious company runs an OS without support, and Redhat support is not cheap.

With the Raspberry, FreeBSD would have a good chance to introduce herself to 25 Mio kids and hobbyists.

But an OS for 25 Mio devices has to be secure and updateable.
 
Top