Environmentalist

The flaw is assuming that people discussing how to save power on their devices have any interest in your opinion that they're just wasting their time.
Agree. The main goal of the computing industry has been reducing power and increasing speed at the same time, over decades. But the reality has been that the number of systems has been increasing more rapidly than the power consumption has decreased. This is called progress in some places. Another question is, how much the overall happiness (if at all) has increased. But FreeBSD seems to be no culprit in this process at all.
 
Last edited:
how much the overall happiness (if at all) has increased.
In what regard? Happiness of the users with their gadgets, the sellers with their profits, or of mankind in general? But then, who would ask such questions, and, would not that be off topic? O:‑)
 
Apologies for chiming in.
Maybe it might be justified for environmentalist reasons.

I regularly get ridiculed and even attacked on this forum when I talk about the fact that a bug in vesa.ko prevents successful S3 resume on people using nvidia cards and sc console.
As long as the bug is not fixed, the workaround is to build a kernel without vesa.ko.
Having S3 sleep work with the GENERIC kernel on FreeBSD would not be detrimental, at least in financial and environmental regards.

Having extremely high energy prices here, I consider outright wasteful to not have my office PC sleep overnight and when not used for longer intervals.
From the $$$ savings S3 sleep gains me, I could buy at least one new PC every year.
So I hope you can refrain from ridiculing me, as I am not rich enough to consider this savings as financially irrrelevant.

Already in 2017 I PRed the bug.
Even though a fix would need only two lines of code, there didn't follow any action

So I kindly ask you to review my draft of a new PR attempt to increase chances that this bug gets fixed.
Thank you!
 
Having S3 sleep work
Oh, that's a good topic, indeed, I mostly gave up on this as there are many possible side effects - and having some half-written document lost due to the machine not coming up again is worse than just saving it and shutting down fully. So, yes, it's a good topic, one would probably go into standby more often if that could be expected to work reliably.

Be happy if it's only two lines and that solves the matter for You.

I just gave it a try, and the result was, well, impressive... I remember it did work suitably with Rel. 11.2 or such. But now, with 12.2, on a simple IvyBridge with the integrated GPU, no usb except mouse and printer, and not much else out of the ordinary, the behaviour is as follows.

-> Standby:
normal behaviour, PS and fan stops, power led now blinking.
-> Restart:
PS and fan start again, but no effort to activate the display signal.
After about 60 sec. machine powers off entirely, starts again, and still no effort to activate the display.
-> Reset button:
No effect.
After another 60 sec. machine power off entirely, starts again, and does a normal full boot. Obviousely all is crashed, filesystems are uinclean, etc.etc.

Not recommended for practical use.
 
Did you hit the power button to awake after sleep?
I do not recommend this because I consider this powering off directly out of the sleep [*].
I'd recommend to just hit a key (shift or ctrl, to not unintentionally inputting chars).

Then the computer powers on.
It can take up to one minute until all stuff has been initialized and is active, like drive spinup, USB init, video etc.
Thus just be patient a moment if you wonder in case you do not instantly see things happen.

To avoid cluttering this thread, maybe we should make a specialized thread for environmentalists?
For example, "To help the environment, help me get my computer sleep and wake up reliably!"


[*] Edit: Maybe this is also worth investigating, as this can be potentially considered as bug in the ACPI and suspend/resume framework, misinterpreting the power button event as "poweroff" event, even if it should be interpreted as "resume" event, in case the system was in the S3 state when the power button event was initiated.
This would prevent people used to the powerbutton as resume initiator (i.e. people used to non-FreeBSD-OSes) doing an unintentional poweroff...
 
Did you hit the power button to awake after sleep?
Nope. Hit space bar.

I do not recommend this because I consider this powering off directly out of the sleep [*].
Should not (except when pressing >4sec.)

Then the computer powers on.
It can take up to one minute until all stuff has been initialized and is active, like drive spinup, USB init, video etc.
Thus just be patient a moment if you wonder in case you do not instantly see things happen.
It didn't really.
But there is another possibility which may be the reason: for IvyBridge GPU there seem to be an old and a new driver and both should somehow work. So, as this is rather unspecific, I decided to use neither and just stay with the kernel-included drm2 which did work with Rel.11. And it still works with Rel.12 - but maybe some things have changed with the standby, and those would likely not get adapted in that old drm driver. That might be an explanation.

Maybe this is also worth investigating, as this can be potentially considered as bug in the ACPI and suspend/resume framework, misinterpreting the power button event as "poweroff" event, even if it should be interpreted as "resume" event, in case the system was in the S3 state when the power button event was initiated.
If that actually happens, I would say it's a bug (but can easily be workarounded as soon as one knows about it).
 
Hmm that's interesting...
would be curious to know what actually happens/goes wrong on your config.

There are some drivers which do not correctly resume, but these are usually old exotic stuff, as the suspend/resume framework was added way before Ivy Bridge iirc.
I have a laptop with HD4000, maybe I should test 12.x on it. On earlier releases, suspend/resume just worked fine.

Regarding the reliability of resume on my nvidia-equipped PCs, I can only say that I had only two resume failures, since I started to use FreeBSD as desktop again in December, doing zzz multiple times every day.

The first fail was when I was distracted and accidentally pressed the power button to resume, instead of a key, just to see it power down like I described in my last post. (Need to test this... might Kitty have walked over the keyboard in the night unnoticedly?)
The second fail was when I did freebsd-update and forgot to build and install my custom kernel directly afterwards (without vesa.ko, as workaround to make resume work).

So my personal impression is that the suspend/resume framework itself is very reliable, as I yet have to experience a "real fail" resuming after sleep, after hundreds of resumes.

What I miss is an up-to-date overview/list of resume-blockers and fixes/workarounds.
Such might save people a lot of time of narrowing down what prevents the successful resume.

Some can be fixed or worked around.
At least if known workaround(s) exist...
Now I am thinking about how a thread could be named to collect all this information regarding solutions or workarounds spread around between few users, who did deep research to find one or the other resume-blocker...
Maybe such a thread could result in thousands of saved kilowatt-hours, actually helping the environment...
 
Hmm that's interesting...
would be curious to know what actually happens/goes wrong on your config.
Sadly, thats not so easy to figure out. I tried it two more times, with the same behaviour. Nothing makes it to the logfiles, and not even the keyboard gets initialized (numlock no reaction) and that's not usb. Then I tried GENERIC kernel, with no X, and no drm loaded, and again the same behaviour.
So this may be any strange parameter I might have anywhere in the bios, loader.conf or sysctl. What is definitely interesting is the reset button not being honored immediately. I have never seen that before, but that seems like the routines are not even getting properly through the mainboard-internal firmware code.
(The mainboard has slightly changed - it was ASUS P8B75-V with Rel.11, and now it is P8H77-V, the only difference should be this one has DisplayPort and so it can use WQHD).
 
Yes, this thing with the reset button not being honored immediately is definitely strange.
I guess the same, there are some things like the boards' firmware, AMD has their equivalent to Intels system management thing. Or it is some UEFI thing... (of which I have zero idea)
Also this self-turning off and on, and rebooting. Honestly, I don't associate such with FreeBSD.
Strange, too, that you get no acpi messages in the log. Will have check with GENERIC kernel whether it is the same here if I try broken sleep (i.e. with vesa.ko loaded).
I think we should really make a thread to collectively find out the best approaches what to do to narrow down the problem when resume fails, as the wiki only says "TBD" there...
(...and because I feel guilty for OT...)

I do not know whether vesa.ko (options VESA) still is needed on UEFI systems after that patch in 2019, which apparently transferred from vesa to vt the task to hand over control to xorg after resume.
If it is no longer needed, one could try to check whether a kernel without vesa.ko would work.

A reason to try this could be that vesa.ko relies on some esoteric antique VESA BIOS functions to work correctly, of which the outcome becomes less and less predictable, the more modern the hardware is.
I would not even rule out that the strange poweroff/poweron-rebooting could be caused by such.

Intel recently dropped a lot of support for obsolete things like Windows XP.
DOS drivers seem to have been completely purged from the Intel site now.
So there is reason to presume that DOS crud like VESA might be cut down in VGA BIOS too, or already have been in their newer reference BIOS.
 
Did you two try to set the suspend_bounce sysctrl for a test?
Back then when I investigated the problem on my nvidia-equipped computers, I tried all what I could find in the wiki, in the forums and in the mailing lists.
I remember there was some effect from some settings, resulting in apparently no action following the suspend command, so this didn't really help much.

Only building custom kernels, trying to narrow down potentially bad kernel modules made me find out that something is wrong with vesa.ko.
Doing this directly led to finding the culprit, as my first step was to remove from the kernel all vt and vesa-related stuff, as they were natural prime suspects.

Investigating that, I found that skipping this invalid BIOS call in vesa.c fixes the problem.
On the forums, I advised people to try the same and this usually fixed the issue for them, too.
So I believe it might be not exclusively my "insane" "vivid imagination" that makes me suspect a problem there.

And setting the define for a verbose ACPI?
Ohhh which one is that? *wannaknow* :)
 
Did you two try to set the suspend_bounce sysctrl for a test?
Ah, thats interesting. But it doesn't suspend at all then. It just switches to text console, resets the usb, then activates the usb again and switches back to X screen. No problem with that, apparently.

It seems the problem is rather the board itself not getting out of suspend state. A short glance thru the cmos config does not show anything too evil, so this seems to need a more elaborate endeavor (minimal config etc.), and I am seriousely not in the mood for such currently - there are other things I need to hack...
 
The flaw is assuming that people discussing how to save power on their devices have any interest in your opinion that they're just wasting their time.
The flaw in that logic is they are subject to it anyway, like it or not. That's their opinion.
 
Luckily with COVID and the reduced aeroplane presence and fuel usage, it has basically made up for my use of slightly inefficient i386 machines in my house for the next couple of million years XD
I gifted my i386 to a guy with ADHD I've known all my life that lives across the hall from me. He's never used a computer in his life, but once he gets the hang of typing and can enter his password correctly he's going to learn FreeBSD 11.2-RELEASE offline.

I have it running for him and told him to leave it on 24/7 so he can fiddle with Gimp, listen to music and learn to type on leafpad. I got root and plenty of patience with. If he shows me he is trying I'll rebuild it so he can go online and Admin it.

I thought the guy with a Degree in Computer Science - Communications upstairs must have to know more about computers than me, but only when it comes to splicing cable to make an Ethernet wire. He thought I didn't know a keyboard from a surfboard. Working from the terminal to build a boxen from ground up something he couldn't begin to do, much less believe I could. He can use the Tutorial.
 
Found the topic fun to engage with. So why not have some wise words on the subject from someone that see the bigger picture.

Not sure why out of all thing one can do, one try and save on the environment on a personal computer. Like honestly any savings done here will be ruined by the countless of stuff happening around you.

Like as said above by others. First thing to check is the screen your using. (for most people) Then the computer itself. Wasted energy is HEAT. And in some ways I like some heating in my room thank you! So even that is not very wasteful here in winter time. But a computer (laptop from say 2008) not used heavily will not be very power hungry. But put a big work load on it, IT WILL be wasteful compared to a modern machine. Since trying to cope with the non intended use of the machine ruins the power usage. A modern machine evolved with the need of the heavier workload. (not true but close enough)

But monitors are always wasteful, and that is part of the design. They are converting electricity into something we humans can interact with (light) so it is very wasteful when it comes to power. And really if your compiling ports then stop doing that to XD Someone has already done that for you. Or rather spent energy on it.

It should be quite obvious. But CPU's are in essences TINY TINY mechanical things moving fast. And the waste of energy is really low due to the small size of it. But the less you use it the better. And if you need to use it then a better and more efficient CPU give more bang for the power use. As they become smaller and smaller as we progress in the tech. Less movement the less HEAT created.

Even old hardware is not bad. The power savings of getting a new board is going to be nothing when comparing the material and energy needed to create and get the new PC, and then get rid of the old computer is doing harm too.

It is just a matter of how much computing you need. The more you need the better a modern computer gets. At least in theory. But being happily using a slow but highly efficient modern ARM thing is best of both words. But do not think the OS or the Hardware is a big factor in saving the environment. It simply is better to go back to a more basic usage of tech. Like 70s and 80s. Nothing fancy but lean and fast.

Solar power made from your own panels. Now THAT will blow the environment friendliness sky high. Well compared to running on money electricity that is. Since the electricity likely is produced in a environment negative way. In the name of PROFIT. So that is a very recommended approach to feel better about computer usage.

Sell green power to buy cheap bad power. No joke this is what happens around the world. Do not trust environment nonsense from companies. They make a green good power source "for free" but opt to make money out of it instead. Since that is what they do. Big industry next to me do just this exact thing. Buy Coal power and sell Green power. Make them allot of money. They waste energy since they are being PAID for creating GREEN power. But they do not use it but exchanging it for money. There power bill is being paid off and getting a profit from it. Instead of you know.. Using the power they created for there own production on the spot.

Really that is the messed up world we live in. Waste energy and getting paid good for it too. At that place they run everything on electric, all lamps are on all year around. Not trying to conserve power in any way. They are Environment Friendly company? Yes. Just happen to create green power without them even trying. A bioproduct at it's finest. Think about that before buying a electric car running on "green" power. It is just as "green" as YOU yourself make it. Same with computers or tech in general.

Really unless you need a computer the "Environmentalist" thing to do is ditching technologies for the most part. Really only thing to use it for is to avoid traveling. And so then if that is the reason to use a computer. Well... It is more about networking that you should be concerned about and not the computer your using. Well with a intent in Environmentalist actions being the number 1 thing.

Do not just focus on what happens on your computer. Think the hole process you embark on when being online. As with everything in life. Think!

It is better being on a computer VS driving to a friend to talk with them and such. It's better watching YouTube then going on a holyday in a car or airplane. But the energy and hardware (and therefor pollution) from being online is no joke. But it dose not have to be. But anything fun or whatever draws you to being on the web is BAD for the environment. It is to avoid needing to travel that is GOOD.

But just 1 other human being on the planet not taking "Environmentalist" actions will ruin anything you do anyways. And to bring the point home fully think about this one. ANY criminal* money making scheme is going to pollute and ruin the earth so much more then a million ordinary people running a computer. At least if they are just sending email's and peer to peer data transfers.

But that's the problem. We are not peer to peer data sharing. We are feeding big servers data. That's no good for multiple reasons. As why they do not charge us for this is simple and open facts. They use our data and run computer intensive (power intensive) things on it. For as you might know already, profit. Training AI's or selling it to put ads on us. (good use of electricity ye?)

I like to give a idea of how pointless the focus on a OS or hardware really is in the big picture. So that is why I bring this stuff up. There is so much more productive things to focus on ;) Like not letting servers suck up power from you watching cats on the web. Do something IRL like grow your own food. Use the web to learn about that stuff. That is what I love about the internet. Knowledge.

Now instead of peer to peer data transfers (a good way to moved data around) we use Facebook, YouTube etc with a middle man server. Witch if you think about it means one extra set of power draining appliances for data transfers as a minimum. They spend more electricity on stuff on there servers then what the home computers uses do. They simply are doing so much in the background compared to if you just had talked directly with however you where interacting with online. (peer to peer computing) FreeBSD server we download files from are doing the least amount of efforts to deliver the files to us. But other servers do all kinds of crazy stuff. For what reason? Profit? Most likely.

Literally ANY peer to peer data transfers are seen as criminal actions today. (more or less) You are going to be challenged why your sending encrypted or non encrypted messages peer to peer. Since it looks suspicious. Even now there are legit reason for it.

And it also is simply a hard thing to do. That is probably the big reason really. Much harder to learn or care about it when you can just go online on a forum like this one. Or Instagram. That is why the changes are not happening. Convenience. To really do anything worthwhile from a Environmentalist standpoint the changes to how we do things need to change. (did not see that one coming did ya all?)
A forum like this are grate! But what else do you do on the web? Might you rethink having them get more data to process?

Not just change the hardware or OS we run will do anything. It is the software and services we run on top of it. Trying to be smarter about what we do in our lives.
The big picture is difficult to grasp. But the point is to limit what actions you take and how much you rely on people with there motives being always about money.

Same with vehicle transport, food, ANYTHING really. Doing as much as you can on your own is the key. You find out what is good or bad for the Environment simply by what you can accomplice on your own or in your community close by. What dose nature support basically is the key to saving the Environment. And working together as small communities this called "internet" can still exist but for real practical reasons. Educations etc. Not Netflix...

So really to be environmentally good, limiting networking and sending and retrieving as little as possible is more important then you might expect. Meaning your basically not going to be using a computer in the way of 2020. But rather just for emails and stuff that really is just sending small files around. That is less impactful then sending snail mails or driving around in a car. So it works out as a positive thing even now it dose pollute. Agene using the computer to avoid traveling with vehicles are A GOOD THING. But that's about it. It is wasteful since it is not necessary or vital in life to be on a computer as we do now.

Avoiding internet and computers plays well in with a low powered ARM based devices. Since in essence if you only do need a low range PC, well you then have changed the way you do thing too. Still you can be online and watch a movie or YouTube. But it is so much more valuable to think about what impact it has outside of your eye sight. Same as how coal burning has such a damaging impact. It is outside of eye sight the problem occurs, always. It is not "free" being online. At least not for the Environment. The cost are huge but better then a airplane trip for "fun".

Literally a tiny ARM or RISC board will draw nearly nothing. The monitor will be the big power consumer. Then the network transfers and anything happening out there on the web. That should be the concern. What dose it take to create this tech, and keep the internet running. That is the real pollution maker.

So to be frank there are better things to look into before looking at whatever power saving measures can be done on OS level. It is more about whatever that OS is doing and what it is communicating with. (and how much)

Like look at that Pi keyboard with a built in computer. You can still watch YouTube on it. But the problem is YouTube. That is the power hungry thing. As is basically anything but the most old school internet usages. Like when we used phones to communicate peer to peer. That is the kind of data transfers that are ethical to be using. And they where SLOW. But did something useful with little impact to anything.

Like I do not even think FreeBSD is the best OS for this task of power saving. (nor Android) Efficient use of electronics are the key for a Environmentalist caring person. Just as avoiding running unnecessary lights in your home (LED ones ofc).

FreeBSD I think is to fancy. XD But I mean sure it is not that bad. You just need to learn so much about computers to get anywhere with it. I think Linux or Pi OS is better suited. Simply more support for the goals of a OS for basic web stuff. FreeBSD is a hobby or a server OS. Or someone that likes to be more in tune with the computer and like them to be powerful and awesome.

If the question was about saving battery life and extend usage time. Then yea looking into hardware and OS is the main thing to do. Since any reason to use the battery is to get something done. For fun or work.
But that was not the question. So happy Environment saving. ? I recommend something like a Pine phone. Mobile and can be charged by a solar panel easily. Can connect to a monitor making it a grate computer also. I really dig the idea of getting one of them. Simply for the reasons I have said in the post. It is a good device since the goals line up with so much of or mere mortals needs.

 
Yes, the cost-benefit calculation is twisted by incorrectly eco-balancing any behavior.
But this is characteristic for the Economy of Death.

For example, I just do not understand how little people talk about power consumption when they "upgrade" their networks.
I mean, what benefit does a SOHO user gain from a <3% used Gbit ethernet, of which every PHY consumes, say, 5W, when one could use "obsolete, uncool" 100Mbit ethernet instead, which only uses, say, ~0.5W, without noticeable performance loss.

So, I wish drivers had their options to set the device to a particular speed setting, and driver power.
Imho, Economically viewed, it is wasteful to power for 100m when in most installations no cable segment is longer than, say, 5m.
 
For example, I just do not understand how little people talk about power consumption when they "upgrade" their networks.
I mean, what benefit does a SOHO user gain from a <3% used Gbit ethernet, of which every PHY consumes, say, 5W, when one could use "obsolete, uncool" 100Mbit ethernet instead, which only uses, say, ~0.5W, without noticeable performance loss.
You don't get them anymore. Given I get my intended haswell-board, it will not have any PCI/PCI-x slots, so I cannot use my fxp 100Mbit adapters anymore. So I already bought a few used ones for PCI-e, and indeed these are running remarkably hot. But, that also differs on specifically how old they are.
Also, on the quad card it would be desireable to power down those that aren't used, but that doesn't seem to work.

And in any case, you're oldfashioned: nowadays we talk 10GBit for SOHO *veg*

So, I wish drivers had their options to set the device to a particular speed setting, and driver power.
Imho, Economically viewed, it is wasteful to power for 100m when in most installations no cable segment is longer than, say, 5m.
There are some switches in that device, but I didn't yet figure out how they would work. em(4):
Code:
     hw.em.eee_setting
             Disable or enable Energy Efficient Ethernet.
     hw.em.smart_pwr_down
             Enable or disable smart power down features on newer adapters.
 
Cool... must take a look into this.
"EEE" energy efficient ethernet reduces the consumption considerably by resting the hardware in sort of sleep mode when not used, definitely worth investigation.

And yes, I know that I am old-fashioned. The energy consumption decreases as technology matures.
aquantiachart.png

(Source)
And this is two times for every patch cable.

What annoys me is the lack of sort of "opt-out" for this kind of "progress".
 
I remember people telling me to get rid of my old TDI and buy a shiny new car, "'For the environment'. When driven carefully, I got it down to 3.7l/100km. Normally, 4.5 and when I was in a hurry still <9.
Producing a new one would consume enough energy to drive it a lot more than 150.000km. And before you ask, it had all the trimmings. The greenest thing to do was drive it as long as possible. Had to sell it at 300.000km. :(

So, it should be best to use old HW for as long as possible. The energy cost at your home should not be considered but the whole 'ecological backpack'.
 
So, it should be best to use old HW for as long as possible. The energy cost at your home should not be considered but the whole 'ecological backpack'.
Yes, and this really depends. A 1980 car, well stored and taken care when not in use, which is used only once a fortnight for getting groceries in the village 5km away, probably is in good technical state. If it were not for the political will ("reduce pollution") that gets old cars banned, it would probably be more ecological as economical to continue using that car for some years more, instead of trashing a perfectly good item for a newly made one.

Another thing - I don't know whether it is true, but I consider it credible - I read M$ dropped support for hardware that is >10 years old.
So this is a real driver for "technological and conomical support" for the "Third World". Not sure what I should think of that...
 
And the waste of energy is really low due to the small size of it.
OK I had to rebuff this.
Most of the wasted energy is because of leaky gates (transistors).
True that the gates are very very small and this contributes to the problem.
Most die shrinks the TDP goes down not up.
 
... so I cannot use my fxp 100Mbit adapters anymore. So I already bought a few used ones for PCI-e, and indeed these are running remarkably hot. But, that also differs on specifically how old they are.
Also, on the quad card it would be desireable to power down those that aren't used, but that doesn't seem to work.
Most die shrinks the TDP goes down not up.
Yes that is one thing one usually does not get shown in the fancy marketing papers.
It can really matter whether you use an old chip or a less-old chip of the same series.
With the very old 100Mbit cards ~1995 you can almost burn your fingertip.
With the last ones, already with board shrunk that much that the slot part was 1/4 of the whole board, barely enough to hold the case slot metal, there is barely some warmth recognizable. And this seems to be the same game every ethernet incarnation.

What I really miss is data with actual energy consumption information, minimum, average, maximum not only for CPUs, but the other components, too...
So one could plan a computer or network configuration according to minimal consumption.
Does there exist any hardware database/market overview, which includes such data?
 
OK I had to rebuff this.
Most of the wasted energy is because of leaky gate (transistors).
True that the gates are very very small.
Most die shrinks the TDP goes down not up.
Yes, but the truth is that neither we (as consumers) nor the manufacturer pays the "true" (fair) price for electronic goods, i.e. incl. costs for disposal and/or recycling. That means for us consumers in the "1st world", it is often cheaper to buy new, modern & more energy efficient articles than to use existing ones until the end of their life. The fact is that others pay the price... because the costs for waste disposal & recycling metals & rare earth elements are exorbitantly low, because these processes are built on exploiting people in the 3rd world. Another calculation is what Crivens wrote above: we do not add the energy costs to manufacture a new article into our calculation; that's impliciltely in the article's price because the producer & vendor pay it. But very often, these are subsidized, i.e. the true costs are hidden for political reasons.
EDIT Not to mention the conditions of mining rare earth elements in central Africa & elsewhere. Child labour & slavery is mundane there, and the civil war lasts longer than I can think. It seems one major reason is that we (1st world) are ok with that as long it guarantees cheap primary materials for us, because the western mining companies have arrangements with the War Lords. Newsworthy example: why are the military leaders in Myanmar so powerful? Because they're incredibly rich. Why are they so rich? Drugs & raw materials... Of course that doesn't mean no russian & chinese companies are involved, too./EDIT
Sytem Error (IMDB)
system Error (trailer):
 
Back
Top