I've been on the "modern standby" hate train for as long as it's existed. Not because it's such a terrible idea per se. You can debate that if you like. I hate it because it fundamentally changes the behavior behind an existing API (lid close, suspend button, or ACPI S3 trigger, take your pick) without warning, and in most cases, without the ability to get the old behavior back.
Regardless of a valid use case for "modern standby", there is DEFINITELY a use case for "old suspend to RAM" and "old hybrid suspend to RAM", which have de facto been blown away by the change.
I have no problem with adding a new sleep state that makes sense for other users. Go ahead and add S22 for that exotic sleep state manufacturing robots use between shifts. S344 for bluetooth devices waiting for reconnection. Whatever, I don't care. But don't replace an existing PRIMARY UI PATTERN with them.
To be fair this is not on microsoft for inventing S0, or even pushing for it for windows devices. It's on device manufacturers like Dell and Lenovo for dropping S3 from UEFI in favor of it.
I learned about this workaround a while back on some throwaway comment on a youtube video, and from what I can tell it really works. I never bothered to really look into system logs or whatever to verify what was going on.
But in a nutshell, unplug your laptop before you close the lid. Windows does not suspend to modern standby when that happens. I always had an overcooked and dead laptop at the other end of my commute, which went away when I started unplugging first. Now my windows machine can go literal days in sleep mode without really impacting remaining battery that much.
which points to a absurdity of s0, it seems to consistently miss or have bugs in things I would expect to be core aspects like
- monitoring if power is unpluged
- monitoring if battery level sings below threshold
- monitor CPU temperature
and take appropriate actions, like transitioning to "classical" S3 or better hybrid S3
it's totally a misterious for me how that doesn't work well even 10 years later
I mean bugs which idk. unexpectedly drain a bit more battery etc. are to be expected but I would expect the UEFI to be like "uh that's too much CPU heat lets stop here", "uh that's draining the batter too fast let's stop", "uh only 40% batter better not even try anything". And sure you want hello to work even with <40% battery but that's if its enabled, which often it isn't.
Yeah you are right, my little "hack" is really something that makes it less terrible, but it does not make it as nice as traditional S3 sleep. The fact that this sleep mess is still so buggy and annoyance prone beggars belief.
This might be an urban myth so read with a pinch of salt, but there was a line of Dells at some point that warned to not close the lid without a proper shutdown, because it'll literally cook the LCD and cause permanent damage. Dell decided this was the right course of action for this problem.
This does work, however it’s really not ideal for Windows machines using a dock (which is usually the power connection) and external monitors. If you unplug first, all your windows get moved to the single laptop display, but if you close the lid first, everything is where it was when you resume.
> It's on device manufacturers like Dell and Lenovo for dropping S3 from UEFI in favor of it.
Note that even CPU manufacturers no longer provide firmware development support for making S3 work on recent series of CPUs. (Apparently some get an exception, as I’ve heard of some laptops from large brands that are sold with Linux and do have S3 support in the firmware.)
Device manufacturers just want windows to run. And why wouldn't they? They sell the devices with windows. If windows doesn't use S3 then S3 never gets tested, whether it's present or not. So it doesn't work, whether it's present or not, and you can't use it, whether it's present or not.
What's the market share of windows on laptops these days, 20% perhaps? I don't know, but I know that Chromebooks and Macos are ahead, far ahead.
Once upon a time, there were fifty times as many Windows laptops as Linux (by which I mean debian, mint, etc) and Linux was ignorable. Now the factor is closer to ten, and I find it difficult to believe that device manufacturers simply ignore that. If they don't support S3, there likely is a better reason than the dominance of Windows.
I don't like using Windows, and I would like them to be a smaller player.
But if I search for "desktop" operating systems share world wide I get numbers that are way way different.
91% of what? Windows has been losing market share to tablets and macs.
Anyway, if you as CTO tell the CEO that testing with Linux will get you access to x% of the market, the CEO answer will obviously depend on x. As Windows loses market share to tablets and macs, x increases, and it becomes less plausible to suggest that vendors are simply ignoring Linux because it's small.
GP talked about the choices of device manufacturers. If device manufacturers consider the sales potential for a new laptop model, are they going to ignore the competition by tablets? How about convertibles or tablets with a keyboard?
Consumers consider them somewhat interchangeable ("I bought a Surface to replace my old laptop"), so they IMO can't be excluded. And if you don't exclude them, then the share of windows laptops is now so low that adding linux support is a noticeable increase in the sales potential of that hardware.
Device manufacturers are quite cynical IMO. Would you personally take on extra testing burden to increase the sales potential for a new model by 0.25%? 0.5%? 1%? 2%? 4%? 8%? I postulate that the device manufacturers wouldn't bother for the numbers at the low end, but if they don't do it now, then they have a better reason than "windows doesn't require it" (quoting from GP).
I'd argue that the deficiencies in Windows as a laptop OS has led to some people choosing iPads and the like when they would've chosen a Windows laptop.
Obviously, a tablet and a laptop have divergent usecases, but there's some overlap. And for those consumers that exist in the overlap, I think many (maybe most?) would rather choose an iPad.
Microsoft kind of saw the writing on the walls with this and that's how we got the whole Windows 8 fiasco.
From the visitors to a large company's web sites. High-traffic sites, general public with a little bit of skew towards high earners, nothing can be named.
Reading your other comments I see where you're getting from.
And if you're indeed adding phones and tables Apple and MS are both at a ~20% (android has an overwhelming +50% by the way) from the same sources I looked at before.
If you just look at "desktop/laptop" Windows still reigns supreme.
Now the question is could be: should you include those for this discussion or not?
Yes because lots of people use them.
No because the conversation was about laptops.
> I know that Chromebooks and Macos are ahead, far ahead [of Windows].
Perhaps in the US, though I doubt it. Chromebooks are not even sold in a lot of places. Apple products are much more expensive (relatively and absolutely) in the rest of the world.
Pretend that you know someone at your bank who can answer the question "what devices do people use for your online banking?" You don't need to really ask, you already know the answer: For a while it was overwhelmingly Windows on PCs/laptops, now Windows has been pushed to a minority position by smartphones, tablets and macbooks.
You might say that those don't count, but I think that for a device manufacturer they do count. I have private and work smartphones and laptops, mine are made by three different manufacturers, all three make both phones, tablets and laptops. AFAICT that's typical. There are exceptions (such as Dell) but overall, the markets overlap so much that most manufacturers need to regard it as one market with different nïches.
They are not one market, in any sense. The vast majority of adults in the country own both a computer and a phone. I'm not sure how they could possibly be treated as competing with one another.
Within the market of personal computers (not the fictional market of personal computers combined with other devices that aren't personal computers), Windows has very dominant market share.
I agree, but there is a venn diagram here. Many use cases are covered by both products. As time goes on, the venn diagram gets closer to a circle. An iPhone of 2008 was a niche device, but nowadays you can complete 99% of tasks completely on an iPhone (yes I made up 99%, just pretend it's some number higher than it once was).
I think, for most adults, their primary personal computer is a smartphone. Many, not sure of the numbers here, don't own a laptop or desktop at all. They may have one for work, though.
I'm curious… if many uses cases are covered by two products, don't those two products compete? If two products compete, how can you think that they're not in the same market in any sense?
To my mind, my ultralight private laptop is more similar to a many tablets and two-in-ones than to the 16" Macbook Pro that I was issued from my employer. Saying that the two laptops are in the same market and the tablets/two-in-ones aren't in that same market… just doesn't make sense to me.
> To be fair this is not on microsoft for inventing S0, or even pushing for it for windows devices. It's on device manufacturers like Dell and Lenovo for dropping S3 from UEFI in favor of it.
It's like they don't even use their products
Oh who am I kidding, if they used they would notice all the bloatware on it and the shoddy build quality, etc
How do you hibernate with zram file? Is a seperate swap partition necessary for hibernating?
Even if I create 200% swap file, I get `Call to Hibernate failed: Not enough suitable swap space for hibernation available on compatible block devices and file systems` on fedora 40.
I've concluded that it's a HW driver bug after trying out creating swap file, adding resume device and offset to grub; even tried setting SELinux to permissive.
It tries to go into hibernate but immediately wakes up.
There are others in Fedora forum who have had successful hibernate with Zram & swapfile.
I didn't find anyone else with my laptop Asus ROG Strix Scar 16 (2023) with successful Hibernate either on our Linux discord.
Modern standby has existed for longer than ACPI did when modern standby was introduced.
Yes I know ACPI was ratified in 1996 but it didn't ACTUALLY exist until 1998 and I would argue it wasn't actually implemented until early 2000 for enterprise users and 2001 for consumers.
Every single criticism, every single one, levied at modern standby was also made against pre-modern standby ACPI.
In addition to that, tech luminaries too numerous to count ranted for years and years and years that ACPI was a secret evil plot by the NSA to I don't know, kidnap your dog or something. That's not an exaggeration. That actually happened 20+ years ago.
> Modern PCs are horrible. ACPI is a complete design disaster in every way. But we're kind of stuck with it. If any Intel people are listening to this and you had anything to do with ACPI, shoot yourself now, before you reproduce.
Linus Torvalds, 2003. https://www.linuxjournal.com/article/7279 He's talking here about pre-modern standby ACPI, the standard everyone loves and misses today and writes long articles eulogizing.
I'm going to chalk this up to people hating change for the sake of hating change.
edit: I'm actually gonna chalk this up to hating for the sake of hating. I know, you know, and I know that you know that if old-skool BIOS was the standard, Hacker News would be flooded with 10,000 word Medium articles about how the evil tech companies are conspiring to keep our systems insecure so the NSA can kidnap our dogs and if pre-Modern Standby was the standard there'd be stories hating on that too.
It's like how haxxors reminisce about the days when the CPU bus was exposed on a port or card-edge connector out the back of a machine and everything was "so free and open" but when a modern interface with 1/10,000,000,000th the number of vulnerabilities is introduced those same people hate on it.
Nobody is mad that "modern standby" exists. We're mad that actual standby is no longer supported on modern hardware. It's very annoying to have to hibernate a laptop when you put it in a bag or else it will overheat and possibly crash.
I struggled to get through the top half of this post, which is about various Thinkpad model types that the author had. But the second part, where the author explains the various sleep modes and what it means, was very informative & worth a read!
I've personally always solved this problem by enabling "Hibernate" (not sure if that's a Windows-specific term) which writes the entire RAM contents to a file and then shuts down completely. The downside is that it takes a few seconds to boot (impressively few seconds on a modern machine, actually), but still, the laptop doesn't come on instantly. But I like knowing that there's no chance the laptop somehow turns on and overheats/drains in the backpack, because it's completely off.
This doesn't take away from the author's rant at all, the idea that a sleep mode would need to support notifications or updates (!) is absurd. A good sleep mode would need to only support "minimum power usage" and "fast wake-up" and nothing else. But with MS's usual mix of excellence and absurd stupidity, that appears to not exist, and "hibernate" is to me an acceptable compromise. My understanding is that Macbooks have gotten this right for decades already, pretty nuts that PCs lag behind so much.
I used to be able to just close the lid on my laptop and forget about it for days. Nowadays with the new laptop provided by my employer, if I close the lid wrong, it doesn't even go to sleep and the battery drains completely (which is... understandable, I guess, since it didn't go to sleep) but even if I do all the rituals and it goes to sleep, the battery still drains away within a few hours. I can't full charge my laptop at the end of the workday, keep my VS and stuff open, and just the laptop to sleep. I have to shut it down properly and then start everything back up again the next morning.
I noticed that my HP only really goes to sleep if I close the lid WHEN UNPLUGGED (Windows settings get ignored or whatever reason). In case other laptops work the same, glad to be of help :)
So the device is (supposedly) awake enough to connect to the Internet, check for notifications and updates, but just not quite there to notice that it has been physically unplugged.
Thanks for that link. My ThinkPad X1 Carbon (gen 7) BIOS supported choosing Windows 10 or Linux sleep mode - I switched to "Linux mode" and now I have S3 sleep. We'll see if that fixes the issue.
I've battled this for a long time. The prospect of having my laptop be and stay charged is exciting.
It makes sense, because one of the modes of laptop use is to attach external display, keyboard, etc, along with power (2 or even 1 USB-C cable), close the lid, and work.
My employer disabled sleep entirely for some reason (not even available via the power menu), which means that if I forget to hibernate it before I throw it in my bag it’s concerningly hot by the time I get home.
What’s double weird is that, if I close it without hibernating it, not only does it not lock the laptop but it seems to prevent the “lock after x minutes inactivity” setting kicking in. Which seems like a huge security problem.
That happened to Ubuntu on my personal laptops, and occasionally I'd notice the mouse cursor was on a different position than when I closed the lid - I think the screen was somehow triggering the touchpad and that kept it awake/unlocked.
This is one the primary reasons I use a MacBook (the other being the trackpad). Instant and reliable on/off when I close/open the lid, without ever worrying about it is something I just take for granted.
If I need a mouse and do a controlled shutdown/boot I might as well use a desktop.
The fact that it has worked for a decade+ over hardware revisions and OS updates is the key thing here.
Windows worked for a while, then didn't, then it worked again.
I actually destroyed a Windows laptop with a faulty lid suspend. Closed the lid, put it in a laptop bag and drove 3 hours. Got there and it was 100% dead, cooked itself alive pretty much by overheating in the bag.
Linux sometimes works, sometimes it doesn't. Depends on hardware and software versions. It might work at first, then you update the OS and it breaks again.
While I haven't personally seen it do this (is this on a personal laptop or a work laptop? Corporate management tools do all sorts of silly things), if it's doing that, would recommend just turning it off (it's in settings).
That’s interesting, I have the exact opposite experience. My Intel MBP would keep running or doing something in my backpack such that the battery would be lower and it would be hot. It wasn’t always and the Intel MBP worked better than any windows computer I had used before it but still noticeable. Sometimes after not using it for a day, closed and unplugged, it would be dead.
In contrast, my M-chip MBP never gets hot and the battery is reminiscent of my iPad. I used to stress about my power cord and making sure I had it with me, I still always carry one because I like being prepared but I don’t have to use it hardly ever away from home.
Modern macs also have an irritating behaviour where they will aggressively connect to your bluetooth headphones even when they are asleep. There is absolutely no purpose for this, except annoying you. I've to open the laptop and disable bluetooth otherwise my headphones immediately connect to it and then can't connect to another device.
Huh? I've had this happen to me on two distinct macbooks now, so I assumed it was a mac thing. Don't have any special audio apps, I use the laptop for google meet and slack huddles though.
I’m thankful this is back to being almost flawless with ARM Macs. Back in the day it worked great too (my first Mac was 2009) but in my anecdotal experience a lot of the later Intel Macs would go to sleep just fine but really did not want to wake back up again, on a semi-regular basis.
Yep that's true. It's impressively fast on my brand new 32GB HP Spectre though.
Side-rant: It's nuts how hard it is to find a good laptop that has 64G RAM, let alone "more than 64G" as you cite. I finally thought I found one in a Thinkpad X1 2-in-1, but then it just had terrible build quality, broken speakers (low rumbling sound, unfixable even after a repair and a replacement), badly working components (eg fingerprint reader) etc. I ended up returning it. The HP is a full 1000 euros cheaper (!), and it's better in every way (incl processor speed) except the smaller RAM. Oh how the mighty have fallen.
Hibernate is not just for laptops though. I have a workstation with 128G of memory, and it’s annoying that the file allocates the full 128G even though i may only be using like 32G. I mean SSD’s have become cheaper but still..
There was a dark time in Apple's lineup (2015-2020) when they ran hot, the keyboard was widely considered terrible, battery time was unimpressive, and the lack of ports forced everyone to carry half a dozen dongles everywhere... These times are gone.
> The HP is a full 1000 euros cheaper
Oh true, Macbooks do seem expensive, until you consider actual value. If "an actual computer" is the primary tool of your trade, I'm 100% convinced it's good ROI to invest in your own comfort/productivity/peace of mind.
To date, I have always use 16 GB RAM on laptops, and "hard work" is done on a remote server via SSH.
This also alleviates the problem of sync'ing between multople laptops, as I use both a MacBook Air (lighter + better mobility: wake-up, WiFi, handling media) and a Lenovo X1 (Linux environment for software development and research).
That is just peak level stupidity. I sometimes boot into Windows, and every time Windows Update runs, the fans are on high, my system is taken hostage ("Do not turn off your computer - updates ready 100%" - showing for 15 minutes), and there is nothing I can do about it.
Compared to this, an `apt upgrade` is basically an instant action (only kernel updates take a bit longer). Too bad Ubuntu is trying to break that with snap, but it is still a minor inconvenience.
Its even worse if you have a bitlocker config that requires a pin.
It will reboot during the update but then fail the boot because noone entered the pin in time, leaving you to complete the update actively when you open it back up.
Ironically this happens even if you select "update and shutdown" because the restart happens during the update process.
I wouldn‘t call the always–on feature stupid but deliberate. The purpose of modern Windows being an ad surface and capture device for learning everything the user knows from user interaction. Switching off Windows Updates, even on Windows Pro and Windows 10, is practically impossible, short of isolating it from net access.
I suspect that the real reason for this change is to ensure that Windows gets updated even if the laptop is sleeping, and that this is a change that enterprise customers have asked for.
In the past where a computer would be connected to the corporate network all the time you could usually get Wake-on-LAN to work so you could boot computers at night to apply updates and ensure that users weren't waiting around forever. With laptops and people working remotely you need a new way to ensure that the bulk of the updates happen when they won't interfere with users trying to work - this is the solution.
Microsoft may well tout additional potential benefits, but that's more about selling it to consumers so they don't complain about changes as much.
And, of course, some manufacturers do a better job than others at implementing this. --I haven't had a problem with any of my Lenovo laptops or even the Surface laptop. Dell, on the other hand, doesn't seem to implement it well at all.
> a change that enterprise customers have asked for
"enterprise customers" have proper security policies and tooling to enforce them in place.
My employer has a "small" utility that periodically checks for version of installed software and will warn me if I don't install the necessary updates. If I don't install updates for more then x days I'm cut off every network access to the corporate network, and I must first install the updates and then ask my manager to let me back in the corporate network (again, there's automated tooling for that).
So this push for "always on" bullshit is not what enterprises asked for.
Microsoft wants mac/ios/android-like experience, but just can't deliver such thing due to hardware fragmentation.
I never understood why microsoft doesn't go full apple and only allows certain top-tier features on surface laptops or other hardware they own from top to bottom.
If you're serious about software you can't underrate the role of hardware: Apple has clearly got this incredibly right, and it's a fundamental part of their value proposition.
While I don’t disagree with you, if Microsoft started gating premium Windows features to their own hardware, both the US and EU would serve them the antitrust suit to end all antitrust suits.
My company laptop has sleep _disabled_, and all the caches, cookies, browsing history etc. are cleared when I log off or shut down. I guess this is due to security reasons. No updates while "sleeping" in my case.
Cellebrite is built around that reason: USB/Thunderbolt stack vulnerabilities, etc., that allow the full-disk encryption key to be extracted out of the RAM of a running system.
I believe iOS is capable of flushing most cleartext data on suspend until the passcode is reentered, so suspend should be as good as shutdown; I don’t know if macOS can do it. On desktop Linux it’s theoretically possible with systemd-homed, but in practice that needs desktop-environment support that’s as far as I know does not exist, so shutting down is more secure. I can’t remember anything about Android either way except for some features for disabling the USB port in some alternative firmware.
On Windows with a typical TPM-only setup, powering off is probably just as bad as suspending, because just powering the machine back on is enough for it to helpfully unlock everything. If you do need to enter a BitLocker passphrase at boot, then you’re in the same situation as typical desktop Linux, so powering off is more secure.
> I suspect that the real reason for this change is to ensure that Windows gets updated even if the laptop is sleeping, and that this is a change that enterprise customers have asked for.
But you don't want that to happen when your laptop is in your backpack.
It is bonkers that Microsoft and manufacturers both would remove the ability for the users to decide it is not a good idea to have an OS running. It effectively forces anyone to shutdown completely their laptops.
Correct me if I'm wrong, but trying to show ads when the laptop lid is closed and the machine is in "sleep mode" (with quotes) is not exactly smart, is it. But again, if MS can still charge their advertisers for such silliness, maybe it makes economical sense (for MS).
Having the microphone on to capture useful information from the environment would be helpful for ad targeting, though I don't know offhand where on the spectrum from paranoid to expected behaviour that would be these days.
That's CYA nonsense, up there with "Q-tips aren't for cleaning ears". (Yes, they are.) Nobody actually shuts down his laptop before chucking it in his bag.
> Nobody actually shuts down his laptop before chucking it in his bag.
I do. I had a few poor experiences with laptops accidentally turning on while in the bag between 1995 and 2010.
After the last time, in 2010, when the thing started beeping due to overheating, I started the habit of shutting down/hibernating completely instead of just closing the lid and popping it into the bag.
Laptops still occasionally do that. I'm kind of impressed that they manage to keep chugging even when the lack of airflow keeps them at >90°C for the entire time it takes for either me to notice that they turned back on, or drain their battery.
They don't, but the problem is the laptop manufacturers and/or Microsoft have fucked up the implementation sufficiently that it's no longer a safe operation.
I have opened my backpack to find it very hot because my laptop accumulated a bunch of heat in its cushioned pouch and subsequently drained all it's battery
I know more than one person who ended in the ER with a perforated eardrum because of qtips. I honestly don't understand why we haven't outlawed them yet (in countries with universal healthcare, in the US you could just have a more expensive health insurance if you choose to use them).
It is not nonsense. I had my laptop turning on from sleep mode inside my bag and getting really hot, because of idiotic Windows 7 settings back then, which turned the device on, when networks changed. It is just ridiculous. Probably still in there that default. Since that day, I always shutdown my device completely.
Don't conclude from your own carelessness to others. Even just one person shutting down their laptop will instantly disprove your claim, that "nobody does XYZ". So you will basically almost always be wrong with such a statement/claim.
The user can choose to use operating systems and applications that let him configure which apps get to run in the background and at which times. You don't want email updates while asleep? Configure your OS to implement this policy. There's no reason that hardware makers should continue to support multiple and legacy sleep modes like S3 just because OS vendors don't give users appropriate policy knobs in software.
Why is S3 "legacy"? Just because Microsoft says so?
When it is working, S3 sleep covers my needs perfectly. And it is not about email updates. I don't want my laptop to do anything while in sleep. It's that simple.
You're right. Why should hardware vendors target operating systems that exist?
Sure, S0 on Windows makes your laptop turn on in your bag for impossible-to-disable updates, causing it to overheat or the battery to run down. And no laptops on the market will S0 on Linux without running down the battery in less than 24 hours.
Why should hardware vendors mollycoddle users who make the foolish decision to use either Windows or Linux?
It also violates the core idea of sleep mode: Keep the current state of my computer intact until I continue working. If a windows update forces a reboot, then that computer state will be lost.
Indeed, I need to remember to unplug the USB-C cable between my macbook pro and my screen, otherwise it flickers on and off every couple of seconds during the night
wired controllers on nintendo switch docks do this, and yes, it is really annoying. It is about every 3 minutes. and it's for the same reason as the macs and windows machines - the switch wakes up and checks for updates then goes back to sleep.
IIRC you can’t run updates while the laptop is asleep though, as that typically requires a user password to initiate. Even with MDM you can’t force updates during sleep
That problem is solved by the Bootstap Token in modern macOS and MDMs which support it. The token gets escrowed with the MDM during enrolment and can be used to unseal the volume for updates without prompting for a password. Good for unattended updates to kiosks or whatever.
if it would work correctly it's acctually a grate UX boon for the user
basically you don't block any bandwidth while a user is using the computer and instead when they are not
also you download slower (less bandwidth) in the background because you have much more time available to download it (like e.g. the whole night)
this feature is also only about downloads, that windows tends to force apply updates once they are downloaded is a different unrelated issue altogether
so theoretically especially for laptops which are only used idk. 1-2h max a day for people with not so fast internet this could be grate
could because it isn't, due to implementation details
e.g. you would only want to run this if you are: plugged in (power), the internet is reliable (at least as reliable as "normally" for any given network) etc.
but AFIK it doesn't really detect changes in power status, doesn't detect bad natwork conditions (which can lead to increased power use), and probably doesn't use "intentional slow" download either
you also can't opt out
so yeah it seems shit with how it's implemented -- but the idea by itself isn't bad
No one has to call Torvalds, they can patch the stuff themselves if they're in such a hurry that they can't wait the few hours/days for a contributor to push a patch.
You are completely missing the point of the original comment. Why is my computer unusable for X minutes when I didn't expect that and perhaps, needed some critical work done?
Me too - to those using acronyms and other abbreviations in their writing, even "obvious ones": it is a good practice to introduce what each stands for the first time it gets used.
ACL = Access Control List / Association for Computational Linguistics
AAA = Authentication Authorization, Accounting / American Automobile Association / Triple-A Battery Size
...
S3 = AWS Simple Storage Service / ACPI Sleep States: S3 is one of them, they range from S0 - S5) / Suspend-to-Idle (S0ix) "Windows 10" mode selected in system firmware, "s2idle" selected in /sys/power/mem_sleep, and Suspend-to-RAM (S3)
Same! It's unfortunate that the names collide. I was also totally confused.
From the article,
> Traditional Sleep requires all system hardware and software components to work together. The operating system must support Sleep, as well as the hardware (e.g. CPU) and the BIOS/UEFI. According to the UEFI to Hardware Interface Standard (ACPI), this usual form of sleep is referred to as S3. S3 is a Sleep State in which all system components, except for the RAM and CPU Cache, are powered off.
Is there any application of "Modern Standby" that is actually delivering significant value to customers somewhere? A lot of the theoretical benefits of it, IMO, have been relatively theoretical. Perhaps Windows updates, but the computers I've used with Modern Standby still have to spend time installing updates when awake.
I cannot imagine what the benefit of receiving notifications and other information like this on a Windows system would be. As far as I know, very, very few apps are programmed for the native "App" development flow (or even to use the notifications api) as opposed to "desktop" apps, and the number only gets smaller as you go into long tail enterprise apps. Perhaps this was designed for Outlook, but I think the use case is dubious.
Of course, there are millions of laptops in the wild, many of which are undoubtedly used by the people who work on Windows. Every time I find the laptop closed and running its fans, or find a computer in a bag that is slightly warm near the CPU (expected behaviour of working Modern Standby), I wonder whether it's just my computer or if it's every single one of them out there. Do people just accept that their computer has to be either shut down or will have an unknown amount of remaining power? It seems like there would have been a huge push to get this fixed if it was really broken, but I rarely hear users talking about it.
Of course I do wonder whether with the advent of the new Qualcomm ARM CPUs, whether they have finally managed to get Modern Standby to be a good clone of what Android devices do...
> Is there any application of "Modern Standby" that is actually delivering significant value to customers somewhere?
Why do you think it was created to deliver value to the customer?
It is merely a reason for Microsoft to spy^Hextract value from a customer even when the PC is sleeping. Like humans, laptop spend a third of their life in sleep mode and it would be nice to monetise that time, multiplied by a couple billion users.
I think it is clear that the goal of the Windows division is to milk every user for all their worth. By the time they are done, laptops will be an obsolete product and Microsoft doesn't really want to support Windows 20 more years.
>Is there any application of "Modern Standby" that is actually delivering significant value to customers somewhere?
Why is this important, or even something to consider? The only thing that's important with Windows and features in it is whether something delivers value to Microsoft.
But does it? Nobody asked for it, it's not even meaningfully used by any Microsoft product and can't drive sales for them, its endless litany of bugs just increase support workloads.
You have a good point, but there's probably some department manager who got a big bonus from being able to spin this "feature" as a great thing and implement it.
> Is there any application of "Modern Standby" that is actually delivering significant value to customers somewhere?
I guess the value is for enterprise users because the ecosystem is so bad.
When I return from sleep on my laptop, my VPN is disconnected and can't reconnect itself automatically as I need to go through the auth + 2FA phase. I am also logged of all Azure AD authenticated web apps and Microsoft web apps like Teams and Outlook are so bad they don't automatically sign in back when you have signed on on other Azure AD resources.
I haven't started my corporate laptop on the original windows 10 for years so I can't compare but I imagine a "modern standby" would have some keepalive on the VPN connections and apps like teams so they don't sign you out?
But in the end this is kind of useless on those Dell laptops because it drains the battery so fast and they have such a low battery life that you just want to shut them down if they are not connected to their docking statino. I guess it serves the people who are on-prem and close the lid to go from one meeting room to another.
But when I was working on-prem I would just configure it to just lock the laptop and would use the sleep shortcut to put the laptop to a normal sleep. Apparently most people were too dumb to figure that out and would walk between their desk and the meeting room awkwardly with the lid up, I can't believe they have managers roles or are even allowed to work in the IT industry ¯\_(ツ)_/¯
Keeping VPN connections alive could be beneficial. Most current laptops do not have cellular modem though so they will have no internet connection while being transported. VPN will therefore stop working anyway. In fact the entire laptop will stop working when it overheats in the bag.
I guess this was pitched by some product person as the feature that brings the laptop on and connects to wifi instantly when you open the lid. All the other uses are post facto justifications. This was all dreamt up when Apple was still on x86. Obviously, now that Apple has superior battery life, these gimmicks don't make any sense.
What I don't understand is, if this issue is so pervasive, why aren't these laptop models getting terrible reviews by both professional and non-professional reviewers ?
Modern Standby enables much faster return from Standby. You open the lid of your device and it's there. That's not the case with S3. I find it adds a lot of convenience.
My 10 year old Thinkpad with S3 wakes up in the time it takes to open the lid, with almost as much battery as it had when I put it to sleep days ago.
My shiny new work laptop without S3 wakes up in the same time, and then says "Battery critically low, shutting down."
This is not a convenience, it's insubordination. I told it to sleep, and it didn't.
But you know what's really insane? When the new work laptop is just sitting there with the screen locked and I walk up and hit a key, it takes longer to just display the password box than the 10 year old Thinkpad takes to wake up from S3 and display the password box. (Ok, it's an unfair comparison, Windows vs Linux, but still...)
How is this the case? Suspend to/from RAM on Linux wakes up in less than a second in my experience. Which makes sense, because everything is still in RAM and systemd doesn't need to do a bunch of work to get everything back online. Maybe Windows has a weird implementation of S3?
Funny, one of the reason why I started heavily using S4 rather than S3 was because with SSDs the time to restart from swap was so fast that it was barely more an inconvenience.
> Perhaps this was designed for Outlook, but I think the use case is dubious.
Oh, sweet summer child. I sometimes use Windows on my work machine, complete with the "new" Teams and "new" Outlook. For some reason, these need a while to update when I wake my machine from sleep, even though it usually sleeps connected to power, with a network cord plugged in and in range of Wi-Fi. Teams, in particular, will say it's offline for a few minutes, even though I can browse the web as soon as the box actually becomes interactive.
The computer actually does something while asleep, judging by how hot it gets. But, as you say, it's really not clear what. It will do this even though Defender is deactivated (we have some other handbrake at work which doesn't do a full system scan on its own) and the system is fully up-to-date.
One of the top reasons I'm staying with MacBooks. It's not unusual for my laptop to have hundreds of days of uptime. Going back to shutting down everything every time I close the lid is unusable for me at this point.
Whoever fixes this first will make a lot of money.
Edit: SteamDeck is also very good at this and that's why I'm sticking to it for handheld gaming.
MacBooks aren't immune to sleep issues either though. Mine used to forget the external display about a fifth of the time, and forget it has any display at all once every 20 or 30 lid close->open cycle.
The solution happened to be unplug and replug for the first issue, and force reboot for the second. I've enough people having issues with external displays that I don't think it's an isolated problem.
For comparison a Surface Pro failing to go to sleep or getting hot in the bag hasn't happened to me in 2 years of owning these machines. It might surely happen, but it's far from a regular occurrence I think.
It depends what your risk profile is. There are no absolutes.
Numerous minor updates have screwed up things for me, particularly with some applications I rely on. I will usually lag 2-3 patches behind and wait for other people to suffer first and confirm what I use is stable. I am still on Sonoma / 14.5 at the moment.
Incidentally I actually have a Mac Pro here which is running Monterey which hasn't been patched for about 9 months. The universe has not and will not be imploding any time soon.
Everyone forgets the "availability" bit of security.
I find it a bit surprising that this is still a thing. Why is there ever a time when I _must_ reboot? Is it just that mainstream kernels were designed at a time when people had lower expectations around this sort of thing, and now it's too hard to evolve their designs toward something that would allow for zero-downtime patching in ~all cases?
Other examples that make me wonder if it's mostly because people haven't demanded better:
- Enormous updates for all kinds of things (gigabytes for a bug fix release) because differential updates aren't pervasive.
- Windows updates where a huge amount of work is done during the "rebooting" phase (why can't most of this be done before reboot?)
- Absolutely atrocious power management on pretty much anything that's not a MacBook, and even not perfect on those.
I never thought we'd have flying cars by now, but if you asked me a decade ago to predict the future of operating systems... it wouldn't be this.
Is there a specific piece of software or driver you got installed on all of them? I've owned 2 of 3 of your MacBook model years and 'sleep on lid close' was very reliable on all of them.
Apple Silicon works even better, resume from sleep is instanteneous instead of taking 1-2 seconds.
A few were work laptops, so they came with the suite of poorly-written corporate spyware agents which I suspect was the culprit.
On my personal laptops the cause could be all the accumulated junk I installed. Apple's refusal to support legacy software may help out since it seems the old software doesn't want to run on new hardware.
If I leave it on sleep for a long time it dies also. But it's perfect for playing games in short breaks, especially with kids. I can immediately drop it when one wakes up crying and return back to it a few hours later and pick up where I've left off.
Closing the lid and putting it into the bag has the same effect for me.
I've my own ritual for preparing myself for the "I am done" phase. Doing the last commits, writing down a few sentences on what should I do when I go back to work and closing the lid.
I have to restart my MacBook every few days unfortunately because sometimes it just has weird issues (like usb devices or networking stopping working).
The issue is not for one person or company to fix. Just like it's impossible to convince all the OSS desktop folks to follow one displayserver/toolkit/UX paradigm it's equally impossible to enforce bug free hardware and drivers from all the PC hardware manufacturers and whatever Microsoft is doing at the moment it seems fixing this is very low on the priority list...
I only use mine occasionally, so I turn it off an put in a drawer. When I power it on a few months later the battery is flat, and the machines date in far in the past.
This have never happened with any other laptop I've had.
Good you do not hold a company-controlled macbook where they force you (after annoying you to death) to do OS upgrades that may break software they themselves provide and require you to use. And then tell you to wait the next of apple's update or some fix because there is nothing they can do. At least windows updates usually do not break software. Unless something gets corrupted or wrong on the way and it does not boot to the OS at all, in which case good luck.
Note that security updates are independent of the OS updates (or should be). There is no real excuse of forcing a user upgrade an OS version.
Otherwise under personal use/control macbooks are great, though, in not forcing you do updates of annoy you to death about them.
Windows updates have lately been causing lots of problems, from relatively benign things like getting stuck in background failure loops to more problematic things like silently removing drivers to outright nasty things like constant BSODs. Using both macOS and Windows regularly, I'd say the two companies are neck and neck for how cavalierly they've been breaking things with major OS updates.
Laptop manufacturers on their own clearly didn't hurt this (dying) category enough over the years prioritizing shallow form over functionality and constant design regressions, Microsoft had to push some more of their deranged "vision" too. S0ix was a joke when introduced years ago and remains one, but it's not funny to laugh at it anymore because its existence eventually killed the actual sleep state an user expects. There is, without an ounce of doubt, no benefit for an user to have their laptop device continue operating when it's expected to be "sleeping" without intentional, conscious instruction to resume work. Some of the proclaimed advantages to it I distinctly remember were "receiving emails" and "automatic updates when the device is not in use", and while the former is mostly useless because it's not a smartphone, the latter is actively harmful to an user's interests when unprompted. Hibernation on linux and windows (where it's secretly kept away for whatever reason since 8 or so) is far more predictable with any modern sabotaged devices missing S3 support, and modern drives are performant and large enough that it's not going to make much of a difference, at least for pauses that one expects to last more than an hour or two. Some more of the endless compromises to put up with because of unrestrained "courage".
This has destroyed one of my older laptops, it decided on its own to patch itself while in a laptop bag in sleep mode. It was a "workstation" laptop that can draw something like 60W from battery, and it promptly cooked itself. The plastic melted and smoke was coming out of the bag. On a moving train.
Here's the thing: this self-serving behaviour from Microsoft will eventually kill someone, or quite possibly a large number of people. It's just a matter of time until a high-powered laptop does this in someone's luggage on a plane, catches fire, and then there will be congressional investigations like we had with the Boeing 737 MCAS system.
In both cases near-monopoly companies are gambling with peoples' lives to meet some internal KPI that in the grand scheme of things just doesn't matter... but it does to some manager's bonus, so you see... people are going to have to die.
Thing is.... My gigantic desktop computer has thermal cutoffs, it will shut down if the CPU gets too hot. I remember desktop PCs having that for ages. I've had overheating laptops do the same. Are laptop manufacturers just getting lazy and skimping on failsafe hardware temperature cut-offs!?? Like, as a hardware manufacturer I would always add that factor-of-5 "the software side WILL fuck up" safety measures. How is the HARDWARE allowing software to let it melt itself in this day and age!?
It's probably because the hardware was not designed to be operating at full power surrounded by insulation (i.e. in a bag). In that scenario the parts that the hardware monitors may still be at 'acceptable' temperatures (which can be ~100C), while other (unmonitored) parts that would not normally get that hot get way too hot.
Having accountants and not engineers architecting software and hardware gets you to that point. Crappy engineers don’t help too, there are too many unskilled people in „IT”
„old” and „legacy” hardware 8-10 years ago was fighting with the tasks and you could still put laptop on your legs. Funny enough yesterday I went into this sleep stuff, because my 5yo Dell battery drained while being „shut down” after 1.5 weeks and while charging I tried to run 2 tabs in browser to look for fix and my legs burned - so I had to put it diagonally on the table to cool down.
We have order of magnitude faster hardware capable of doing order of magnitude less things.
I don't really understand what the fuzz about fast boot is. Why can't most people just wait a few seconds? It's not like boot still takes minutes.
I almost always shut down my laptop. This serves me two purposes:
1) It saves power
2) I always start with a fresh session, keeping distractions at a minimum and having habits to "finish" the current task I am working at. (I do the same with browsers for example. Here I even less frequently save my session).
There are a few exceptions where I know I'll just take a quick break and resume work shortly, but they are rare. No, you don't have to have everything always available within milliseconds. No, I don't need notifications always on (so I guess I agree with the author here). But maybe just different tastes, different habits.
I don't know if this is true, but I talked about a broader topic along those lines with my therapist recently and she was like: 'We invent all these smart things to save time and have more for ourselves, but in reality people have less time, because then they just want to put more tasks into the same timespan'. I have absolutely no idea if there is any truth to that, but it feels true on my empirical observations. Embrace slowness!
Shutting down and booting are intensive operations, it's likely that for short enough periods of time, sleep is actually more power-efficient if it works well.
I don't have numbers though and with SDDs and fast boot times, maybe booting is way less power hungry than before in the end.
2, I shutdown my personal laptop because I don't care having a blankslate when booting up.
But regarding the work laptop, Sometimes it's nice to have stuff ready when you're oncall.
I have my applications/windows pre-prepared so that I don't have to go trough a thousands hoops in the middle of the night to resolved that priority 1 incident.
I understand your point, but I also can see what's the fuzz about.
If the fuzz is legit is not for me to say.
After hibernating or shutting down my laptop, I need to enter the password for the disk encryption again. It's a long, secure password, and I hate entering it. Now that my new laptop does not support S3 anymore and the battery does not last one night in S0ix, I have to enter it every morning, and I constantly mistype because I don't see what I'm typing. I'm considering changing it to something less secure because of this. TPM without PIN is not allowed by my employer, and TMP with PIN is not supported on Linux AFAIK.
TPM with pin is definitely supported[1] by systemd-cryptenroll on linux:
> It is possible to require a PIN to be entered in addition to the TPM state being correct. Simply add the option --tpm2-with-pin=yes to the command above and enter the PIN when prompted.
Thanks, maybe I was just too stupid then. I messed around with dracut and gave up when the system stopped booting. Fixed it while cursing all this complexity and got back to actual work.
The main pushback we get is from our floor planners who work offsite in customer's homes - that extra 2 minutes or so while it boots from scratch apparently seems like an eternity when you're in a dirty and hazardous environment.
>Why do we accept being spat upon and treated like basement bums when we put our money on the table???
Tell me you aren't in the target market without telling me you aren't in the target market.
Microsoft will never, ever, ever be able to make everyone happy. I do not agree with the direction they are going either, because I (as a long-time developer on Windows) am not in their target market.
The term "target market" is itself misleading. The end user of Windows is, generally speaking, not buying anything from Microsoft (an Office 365 subscription notwithstanding). It doesn't even matter what any segment of the end users want anymore, so much as what the largest segment of them will tolerate.
This is the target market. It's not power users, it's not computing professionals, it's not me, it's likely not you. It's my mom. It's your mom. It's people that don't really know what their computer is doing.
The end user doesn't pay for the software and so is not even a part of the market at all anymore. The target market of Windows is OEMs and Microsoft's "business partners", with a shift lately from the former to the latter.
Windows is no more sold to your mom or mine than it is to either of us. It's sold to people who aren't even users.
I use an IdeaPad with a 4000 series Ryzen. It works fine for the time being.
Articles like these make me afraid that the damn thing will break eventually and I'll have to spend exorbitants amounts of money on something like a Framework to get a usable Linux experience.
Not to mention that having a fake-off state like S0ix is a golden ticket for covert surveillance (like our phones don't do enough for the gestapo). I don't think anybody cares about Cortana, even less so on a suspended laptop. The notifications and update angle is really suspicious as well.
My intel 12th gen works find in s2idle or deep(s3) running Fedora. deep is lasts a little longer but that might be because of configuration, which is a bigger part of s2idle.
The community.frame.work forum has reports from people running newer boards so you don't have to guess before buying.
Steve Jobs had an insight that a computer was just another household appliance
But it was built by engineers who came from the homebrew world
This is a great example of how our appliance has gotten worse, even though it’s on hardware that is 1000x the throughput, and we’ve lost the ability to configure it too
It's gotten better but it's very dependent on chipsets and driver support for those. I have one Samsung laptop with an Intel chipset that more or less goes standby reliably. Sometimes it still messes up and it's dead when you open it. But mostly it's fine and it's gotten more rare with recent kernel versions. The fingerprint reader is pretty much the only thing that doesn't work on that laptop and probably never will because it's a cheap Chinese thing with some proprietary firmware.
The core issue is that the firmware in most laptops is just ancient garbage and nobody seems to really care about it. Apple has full ownership over their own firmware so it's less of an issue for them (though it still has issues). But most chip manufacturers servicing the wintel laptop market treat software as an afterthought. They aren't very good at it and they just put in the minimum effort that they can get away with. So, it's all proprietary, poorly designed, decades old, etc. If it breaks they bang on it until it sort of works again (windows only).
Laptop makers then make things worse by just buying a lot of cheap components. Synaptic seems popular for touch pads, for example. My Samsung has one. It's not good. I've actually never come across anything made by them that wasn't garbage.
With Linux slowly creeping up in market share on desktops, it's something that could start changing. Most Linux users are very selective in what they buy and there are certain brands that are just so hopeless that people actively avoid them. Lenovo bought a good reputation from IBM when they acquired their laptop division. That's ages ago but they sort of nurtured that and it's probably helped a lot. It's not rocket science to just make sure things work on Linux.
My work laptop is a macbook. The hardware is just miles ahead of the competition. It's just not even close. Everything else is a major downgrade. I bought that Samsung to replace a mac because I was waiting for the M1 to come out when my old laptop died. I worked on that thing for half a year. Going back to a mac was just such major upgrade in every way possible. You forget how great they are. Software wise I'm actually fine with Linux. Everything I use is open source or available for Linux. I don't use any of the Apple software (the i* stuff). I'm there for the hardware only. If Linux support catches up, I might switch actually.
My new ThinkPad, which I absolutely love otherwise, keeps the CPU fans running if I put it to "sleep". Which left me completely flabbergasted the first time it happened! I've had to force-enable hibernation and use that for any kind of backpack travel. Which will shorten the lifespan of the SSD, but what else am I supposed to do?
Is it running Windows? I've found that on Linux this specific issue doesn't happen, it will turn off the fans, and they will stay off. But the battery is still drained, even though, on average, I tend to get better battery life than with Windows on the same machine.
But Windows tends to wake it up randomly and do God knows what, since the PC will actually be hot to the touch.
Bonus points for attempting to install updates, then reboot, only to end up stuck waiting for the BitLocker PIN.
The BitLocker thing is definitely a problem. On Apple systems which have a similar FDE system, the preboot environment will shut the machine down if the page is left open for too long. The CPU/GPU (if present) are in a pre-power-management state during this time too, so the laptop will really heat up.
I wonder how limited the preboot BitLocker screen environment really is. Could there could be some flag set by the UEFI that says "I'm a laptop and the lid is closed without power connected, shut down after a timeout".
In my case, my computers will end up on that screen in two cases.
The first is my desktop, without a TPM, which will ask for a BitLocker password. In this case, if I'm not fast enough, it will shut down the computer.
The second is my work laptop, which I usually expect to reboot or otherwise am around when it ends up there, in which case I'll usually type in the PIN quickly enough. However, I expect it to have a similar behavior as the first case, although I haven't tried.
But the reason I know this happens, is that I once left this PC attached to my 32" screen next to my bed, and it woke me up at night. I just went and forcibly shut off the thing.
So apparently this is actually configurable, but it's either on or off, and it's set to one minute by default (either 1 minute or no shutdown, no inbetween).
The command in question is ```bcdedit /set {bootmgr} bootshutdowndisabled 1```.
>But Windows tends to wake it up randomly and do God knows what, since the PC will actually be hot to the touch.
For the best part of a decade the first thing I do on any mobile device I touch is disable wake calls in the power plan settings. Learned it the hard way after the T420 woke me up multiple times per night with loud beeps...
I think Windows will shorten the lifespan of your SSD even if you use "shut down", at least if you don't disable "fast startup". "Fast startup" basically replaces "shut down" with a light form of hiberation, which makes shutting the machine down take a long while and turning it on about a second faster. Not worth the tradeoff in my opinion. I also got some bugs when using fast startup, sometimes the sound wouldn't work until a reboot or my WiFi took weirdly long to connect. This is on a Lenovo Yoga/IdeaPad.
Unfortunately a lot of Laptop manufacturers were forced to leave the fans on all the time due to overheating which would cause battery issues if they didn't run the fans.
I remember getting a BIOS update on my Dell a few years ago that permanently turned the fan on too.
There shouldn't be any overheating when in S3, from what power usage ??
And they can wait to switch to S3 (and shut down the fans) once the temperature got low enough, a bit like how with S3+swap, the computer won't enter suspend until the RAM has been copied to swap.
If your notebook has 16 GB of RAM that needs to be written out every time it hibernates, and you do that three times a day, five days a week, you’d accumulate about 63 TB of writes over five years. That’s roughly a third of the TBW rating for an average 256 GB SSD. A typical 1 TB SSD would have three times the endurance or more.
I'd say 1/3rd is a rather large chunk of an SSDs TBW if its just for hibernating. In practice, for reasons you outlined, it would be a smaller amount but I would only find 1-2% tolerable.
Chances are, the computer will break or be replaced in five years or less. It doesn’t really matter if you reach 2%, 33%, or even 90% of the TBW if the system or component is replaced before it uses up all its writes.
> Chances are, the computer will break or be replaced in five years or less.
I think this is less and less true as time goes on. I have a laptop from 5 ish years ago I still use and it's actually pretty fast.
And, even so, it's not like the ONLY wear on your SSD is the hibernation. So this 5 year figure, I'm not sure it's right. I think, maybe if you're someone who stresses your hard drives a lot (maybe you work with photography or videography?) hibernation might make an impactful difference.
The five-year estimate is based solely on hibernation writes, which is a very conservative figure. In reality, most users will use closer to 2% of the TBW for hibernation, not 33%.
For those with heavy I/O workloads, they likely have higher-grade or larger SSDs with a higher TBW rating.
Additionally, TBW isn’t a hard limit and can often be exceeded, with some drives lasting beyond 200%.
For most users, TBW won’t be an issue, and other components are more likely to fail before the drive wears out.
Their aim with S0ix is relatable and can somewhat be compared to Apples move to force USB-C on everyone - which is great and a necessity to be done by a market leader.
So, wait for the EU to mandate companies use the same standard? Even then, there were rumors that Apple was going to cripple non-Apple brand chargers.
Apple: We're gonna only offer thunderbolt on all of our devices, so you can always rely on our super special premium connection!
EU: Ahem.
Apple: We're gonna only offer USB-C for great interoperability with competitors' systems, aren't we generous and great? Also this was our idea this whole time, stop asking questions!
The "fun" part is that smartphones/tablets implement those features with S3 (CPU off, RAM in self-refresh).
Updates? Well here is an RTC to wake-up. Voice control? Well here is an always on audio dsp. Remote wake-up? Well here is a WiFi frame filter to trigger IRQ.
I think it isn't tractable on most laptops because of all the hardware that has been onloaded to software, and the hardware isn't capable of those hardware filtering, and partly OS and drivers are probably to blame too. An embesded OS takes under 1s to wakeup, so even managing TCP KeepAlives wih resume/suspend is realistic on those devices. Not so much when wakeup takes 10s.
The fact that they are pushing S3 as "Linux Sleep" or "Legacy Sleep" and S0ix as "modern sleep" is absolutely not innocent from a marketing perspective.
We ought to reverse Uno this non-sense. Perhaps the Linux Foundation and all the distro development teams out there should start calling S3 _True_Sleep_™ in all their official communication.
I swear I remember Microsoft adding "sleep" and calling it such in some version of Windows because "it's new and different from 'suspend'", then later other systems started using the term "sleep" instead of "suspend" without actually changing anything.
As someone who used to work at Microsoft, this is spot on. The sheer amount of arrogance coming from (some of) my peers who worked there was incredible.
Microsoft decides for you how you would like to use your computer. If you don't agree, use Linux. Or Mac, but Apple is really no better than Microsoft with regard to arrogance.
I'd say Apple is even more arrogant, but at least their software is arrogantly functional.
Maybe I could look past this sleep nonsense if their super cool new sleep actually worked. But it just doesn't, instead it cooks your laptop 10% of the time and it you open it up and it's dead.
S0 standby sucks. The laptop inevitably turns into a fan heater in my backpack and when you get it out to use it the battery is dead. You can spend ages on google finding obscure registry keys to tweak to turn off network in S0 and tweak hibernate settings which partly helps but it's like we've gone a big step backwards here.
I'm always amazed by the tales of people who seemingly use sleep mode on the laptops without issues. In my experience every time I try this trick the laptop wakes up on its own in my backpack and turns into a heatbomb. This reliably reproduces on all half a dozen laptops I've used over past decade or more. So I was kinda trained to always turning off laptops for transporting.
Is sleep really fixed and finally can be used on Windows? Do you tweak power settings or configs for that in any way?
Or get a Macbook Air. Got the M1 in a few minutes in John Lewis and still good. No sleep problems, no fan to whirr.
The previous Macbook Air 11 also was great for sleep/wake. The Windows 7 Thinkpad before that was ok but took like 30 seconds to restart from sleep and would crash sometimes.
Without getting technical about it... my previous work laptop (2019 vintage) when suspended, would blink a light until unsuspended - hours or days later - with still usable battery capacity left.
The current one. Fold it shut, stick it in my backpack. No noticeable powering-up or overheating issues, but if it's more than a few hours, the thing has a drained battery and is fully shut down. Luckily Windows "fast startup" mitigates the pain - it doesn't take very long for it to unsuspend-from-disk and reach a usable state again, with apps still open etc. No idea what Linux would do here. I actually have Linux on my previous work laptop (Elitebook 840 G6) and "proper" suspend works just fine on it.
LTT had a rant about this [modern sleep] and I really hoped it would draw some attention given his size and reach. It is truly mind-blowing that not even Apple got it right.
From personal experience, it's still safe to never shut down an apple laptop, just throw it in the backpack/suitcase and it will be cool and with the same battery at the destination.
Tbh all my Apple laptops, including the piece of shit one with the emoji keyboard, have had no problem sleeping.
Ofc that's on mac os, not sure what they offer if you install linux or windows.
macOS has its own fair share of sleep bugs, particularly stuff like Parallels, IntelliJ or MS Teams for whatever reason routinely manage to prevent sleep or to randomly activate and hog the dGPU, turning a "lap top" to a "testicle airfryer".
More than once I went on a train ride to the office, having clicked "Suspend" in the main menu and watching the screen go black before placing the laptop in my bag, just to find out the battery was completely drained and the 2019 MBP so hot that it not just required cooling down before it would power up again but also hot enough to fry an egg - an IR thermometer showed 70 °C!
If anyone knows or writes an application that could place a red warning icon in the menu bar warning me that "suspending" will not actually work, I'll be glad to exchange it for a crate of beer.
> particularly stuff like Parallels, IntelliJ or MS Teams for whatever reason routinely manage to prevent sleep or to randomly activate and hog the dGPU
mac os is nice enough to allow applications to prevent sleep. Check "pmset -g assertions" to point out the culprit.
A couple years ago it was Chrome for me. It kept an assertion saying "WebRTC has active peer connections" on all the time. This has only lead me to have a pathological hatred for WebRTC and I turn it off in all my browsers now.
As for the dGPU, i've never had one but I do have a feeling that spells trouble, especially with these electron apps and their Helper (GPU) processes. I'm old fashioned enough to do most of my work on desktops that are plugged in and only buy lightweight laptops (13-14 inch mbpros wit no discrete GPU).
I know who the culprits are, Activity Monitor shows it in the Energy tab.
But since it's sporadic, I sometimes forget to look into Activity Monitor if I'm in a rush, and for fucks sake this is not a problem that should exist in the first place.
I mean, yes, I've had more issues back when I still used Windows, but my Apple gear costs thousands of dollars, I expect better quality.
I expect that if the system knows that my intent will not be successful (because some application locks suspend) that the option is either not available/greyed out or that I at least get a goddamn prompt "XYZ is locking suspend. <Abort attempt/Force suspend>".
As for Electron, yes, Teams is cursed in that regard (as it is in locking the dGPU), but for me it's usually Parallels, closely followed by IntelliJ that randomly messes up suspend. Neither of them are Electron.
> I mean, yes, I've had more issues back when I still used Windows
Tbh I don't understand why an OS would allow applications to prevent sleep either, but you mentioned MS Teams and I mentioned Google :)
I don't think IntelliJ ever prevented sleep for me, but I mostly used the Android Studio flavor. I've caught it using up all my ram though if i left it on for weeks.
As for Parallels, I've never tried to close a laptop with a running VM on it. I don't know why, but I always suspend the VMs manually (from the Parallels menu for example) if i use them.
Kinda telling how Google switched to S0(s2idle) on Android since Pixel 3 or 4, fixed up most of their own drivers, submitted fixes to mainline (and probably to firmware developers too) and a few years later every Android device uses S0 and sleeps flawlessly.
Meanwhile MS have been trying to do the same since 2012 and there are still numerous issues.
This sounds totally nuts. I guess I’m happy to be running a laptop where when I close the lid, it will sleep and have battery when I open the lid, in an hour or in 2 days. I thought that was the basic requirement for being a laptop? The essential feature. It’s so surprising to hear things have gone backwards in the Windows world.
My suggestion to anyone dealing with annoying power management is to get a MacBook Pro (you can get tons of ram, we are ordering 96gb models for our engineers) and run your preferred OS in a full screen VM if you don’t like MacOS. I did this for years because I don’t trust Linux to manage battery or reliably pair with Bluetooth devices (especially Bluetooth audio), but I much preferred Linux window management and access to Docker.
Parallels runs Windows quite well, and either that, Virtualbox, or the various apps that are thin layers over the system Hypervisor.framework all do the trick.
Asahi Linux is totally unready as a daily driver / production OS in my opinion. They are doing great technical work and I love reading the progress updates, but there's too many limitations as of today: No microphone, Thunderbolt/USB4, USB-C displays, or fingerprint reader; and much worse sleep/suspend compared to macOS. I mean, if you're happy with <48hr closed lid battery life, then you can get any general market "laptop" with all the issues described in the original article.
One of the big reasons that I moved to MacOS in ~2004 was that, at that time, for reasonable money you couldn't get a laptop that had any sort of reliably sensible sleep behaviour in Linux, and even Windows was iffy. The iBook had no such issues.
It's somewhat surprising that this is still the case, 20 years later.
> Therefore Consumers have the option to state their disapproval with the current state of S0ix by buying Laptops that still support S3.
I’d like to do this, as I find a laptop that cannot last in sleep for at least a week ill-suited for my needs. Where do I go to figure out if my next laptop supports proper sleep?
My old ASUS laptop that triggers complete poweroff when a sleep command is sent to the motherboard, making sure you lose any work you had open as punishment for assuming it had a sleep mode: "I'm four parallel universes ahead of you."
I don't understand how a laptop manufacturer can ship products with such a fatal flaw and still be in business. Is the windows laptop market so dreadful it does not stand out?
Or do people do not care because they do not use laptops on the go anymore?
Not up to the manufacturer any more. Windows does what Windows wants. If Windows doesn't use S3 then it never gets tested, and what isn't tested is broken. On the negatives of S0ix (your battery running out), all laptops will have this same flaw because it's a Microsoft-induced flaw.
Oh this hits deep. I recently installed Windows on a laptop after a long time of only using it on a desktop. I thought I was going crazy - where had proper sleep gone ?!
'Luckily' hibernate only takes about 4x as long as sleep while still being just quick enough not to be too annoying.
I agree with the author that it totally defies the purpose of a laptop. Another terrible move by Microsoft
It is why I will fork out 2-3x for a Macbook, because apart from the screens being amazing, when I close the lid, I know it does exactly what 99.99999% of the population expects it to do, and I can safely put it in my bag.
> ...it does exactly what 99.99999% of the population expects it to do...
Nope, it does not, but Apple is miles ahead of MS in this implementation, so it works almost as intended (not really sleeping while not preheating your lunch in your backpack).
My solution to this was to setup my laptop to boot in less than ten seconds. I could generally go from shut down to logged in with Obsidian, Firefox, and Anki open in about 30 seconds.
The article is qualified as a rant so I don’t expect to read proposals there but how can we pick a laptop with a working sleep mode? Is there a database like proton db somewhere?
Having owned countless Mac notebooks over the last 30 years — starting from the PowerBook 140 up through the M1 MacBook Pro — I can tell you with confidence that Apple has made the same sleep change shenanigans mentioned in OP’s article.
It used to be that you could count on a sleeping Mac to stay asleep until you explicitly woke it up by opening the lid, pressing a key on the keyboard, or pressing the trackpad (or separate trackpad buttons — yes, those used to exist). Perhaps more importantly, you could count on the sleep function to have barely any effect on the Mac battery.
I don’t recall when, but at some point over the aforementioned three decades, Apple started changing the terms of this sleep contract.
It seems Apple decided that some functions should still be available when the Mac is “sleeping”, with no way to restore the previous behavior. As a result, random wake-ups are the new normal, replete with unexpected battery drainage.
I have seen modern MacBooks go to “sleep” with an 80% charge at night, only to be rendered dead with a 0% battery level by morning. Does something this extreme happen often? No. But it never happened before. And moderate battery loss while sleeping? That happens very often on today’s MacBooks.
Your mileage may very. And Apple probably continues to implement better computer sleep behavior than competing vendors. But I would argue that people who think MacBook sleep behavior is excellent… have never experienced how it behaved in the past. That behavior was superb.
But sadly, I imagine most people have never experienced that excellence.
> But in Contrast, Apple kept the legacy Lightning and even brought back Magsafe (and if it only were to milk the cashcow).
I don’t quite understand this line. Are they talking about Lightning or MagSafe with the parenthetical? MagSafe specifically is something users were begging them to bring back and they did it without removing USB-C charging (best of both worlds).
I see S3 as another victim of the enforced convergence of PC and mobile that MS did get into with Windows 8.
On paper modernizing S3 made even sense since getting all the system components to work nicely with S3 is not an easy task. What wasn't thought off is that handling the same task in modern standby generally isn't easier.
I don't get why the author blames bad drivers on the standby mode.
I remember similar issues with S3 back then. You had to jump through many hoops on FreeBSD and Linux to make it work without draining your battery while still being able to resume from standby successfully. Disabling some drivers, but not others, etc.
Thus, a faulty implementation of a device or driver is not to blame on the standby mode. FWIW, Modern Standby works fine on my Thinkpad X1 Titanium Yoga with Windows 11. The only device that drains more battery than necessary if not disabled is the Qualcomm WWAN modem. That stops once you disable it, like in the "good old days". So for me, it's nothing negative, but actually adds convenience because with Modern Standby, my device goes to sleep and resumes much faster than the devices ever did with S3.
I can totally relate. There are some relatively modern processors though that still support the old but gold S3 state. The i5-1240P and i7-1260P processors support these, but not all manufacturers integrate UEFI support for this sleep state. I found that Dasharo offers the ability to switch to S3: https://docs.dasharo.com/dasharo-menu-docs/dasharo-system-fe...
We've had a have a few issues related to the move to modern sleep - in that "shutdown" is now hibernate (with fast boot enabled) - so any issues that would be fixed with a restart remain.
With some machines, the modern sleep where it stays connected to the network can sometimes cause connection issues that aren't as obvious to the end user.
In a lot of cases we disable fast boot, and disable modern sleep and the problem goes away - but having the ability to select which parts should be enabled and disabled, like how you get the option with PS5's rest mode, would be very useful - especially if controllable by GPO / Intune.
I didn't know S3 vs S0 but this may be the reason why I never keep my laptop in sleep mode and I prefer to put it in hibernate. What about hibernate? Couldn't it be an alternative to the "legacy sleep"?
Hibernate mode takes a long time to enter and exit, as it involves writing a lot of data to disk and then reading it back upon resuming. The whole point of Sleep mode is that it's fast: you just close the lid and you're in sleep mode, saving power. Then if you want to check something quickly, or resume your work, you just open the lid and in a few seconds or less, your computer is fully operational again.
Well, you're right but of Sleep is impractical with S0 mode as explained, hibernate with slightly longer transition time may be acceptable. Hibernate has really became fast on modern computers.
You're in luck, Windows shutdown is now actually hibernate (and has been for a few years). So if you just shutdown your laptop, you've been using this hibernate for a while. Turns out all those people saying "shut down your laptop before you close the lid!" were actually right.
I remember the same kind of annoyances already in 2009 with Windows (Vista!). It's one of the reasons that made me switch to Mac without any regret. Nothing seems to have changed in 15 years.
I got halfway through the article and the font suddenly switched to an unreadable monospace! (Sorry for bikeshedding but it was a uniquely bemusing experience.)
Yeah, I have the same. Usually, they do it when I'm on the train and they're in their neoprene cover, in my backpack, cooking themselves until they're too hot to touch. Windows Event Viewer says that it's Windows Update waking them up to install patches, so it's a cherry on top that invariably, the update installations have not started, either. The laptops will literally just power up, stay on the login screen, and do bugger all until the battery is dead. (Observed on Lenovo and HP, with both W10 and W11).
Can very much relate have an Linux XPS 9570 myself and sleep states are a mess especially in conjunction with power saving tools. Did as much as I could to decrease boot time and shut down of most of the times.
Now I have an M2 Apple and it’s like a day and night difference. Close the lid works exactly how it should work. I maybe did shutdown the Mac onces.
I’m hopeful that ARM based laptops will have better power management on Linux. Totally agree with what you have written. Ironically though, Microsoft standby sucks on Linux and Windows. Happy to be running a Mac.
I’m curious how well Asahi linux works with standby on a Apple Silicon Mac.
IME experience, at least on HP G8 EliteBooks, both AMD and Intel, Linux actually keeps the machine quiet and cool during "modern standby". With Windows, it absolutely will turn on the fans at some point, and the machine absolutely will get hot for whatever reason.
This seems to have been fixed for a while now (say a year or so), but when the Intel machine was new, Windows would sometimes not wake up from sleep, or if did, the screen would be garbled.
The AMD still has an issue where Windows will "unexpectedly shutdown" (according to the event log) during modern standby. This happens every single time I leave it on standby during the night. This is an almost four-year-old machine by now, not some brand-spanking-new thing, which works otherwise perfectly, especially on Linux.
It’s not so much the heat that bothers most people, but the fact that when you close the lid of a windows laptop and throw it in your backpack, it will be totally out of battery, be running and you have to actually physically turn it off. Simply close the lid ane leave it in your backpack on a Mac and it will still retain most of it’s charge.
I think most people intend their laptop to retain it’s charge when they close the lid and standby used to do this, but on windows machines, it really doesn’t anymore.
It should really write all the ram to hd after a short period of time in your backpack instead of quickly depleting the battery.
> I’m curious how well Asahi linux works with standby on a Apple Silicon Mac.
I’ve not tried it long but from my short experience, it works well.
Btw, Asahi is actually really easy to install if you are curious. Everything is done/installed from macOS including shrinking your macOS partition. And uninstalling it is as simple as removing the partition from Disk Manager and growing back your macOS partition.
All you need is some disk space and, one command line to copy paste and about 20 minutes.
And it works really well overall. It have everything to be a daily driver (including graphics acceleration) except external display support and TouchID. But both of those issues are being worked on.
I played with Asahi on my personal use MacBook Air. It is great for the things it does. But, I’m waiting on the USB-C implementation and the external monitor support before thinking about switching to it for proper use. For now, it is just checking in every month or so to see what all works and what doesn’t.
But overall, I was pretty impressed by Fedora Asahi. I was expecting a clunky experience with hacks and bugs everywhere but I'm now actually expecting to use it as my daily driver as soon as possible.
Like, there are those things (USB-C, external display, Touch ID) that doesn't work at all but everything else works wonderfully. Even the audio is as good as on macOS (thanks to the specially built audio driver) which iirc, have never been the case on Intel macs (yes, audio works on intel but the sound is better on macOS because Apple have a special driver that is able to shortly and safely overdrive the speakers to get better sound - and asahi replicated this driver).
Oh yeah, I've spent many hours on DELLs trying to solve S0ix sleep states... What a pain in the butt that is. There are few major pain points:
1) Every component has to be in the right "state of mind" in order for Si0x sleep to work properly. Do you have shitty NVMe that has buggy PCI powersaving? Too bad, battery will drain 10%/hour. Your yubikey keeps USB bus by constantly sending interrupts? Ha, enjoy your new backpack heater and watch out for fires...
2) It is extremely hard to debug these issues. I've figure out the yubikey and few things, but some of them are esoterical! ifconfig eth0 down = some part of Intel AMT stops sleeping and is now fully on all the time. Why? Because fck you, that's why. If you want to get some debug info, you can get some very cryptic state-flags for (some) components and you really have to guess what these really are - there is no documentation unfortunately.
3) UEFI is a mess - one would say that you want critical component, such as BIOS/boot chain to be as simple as possible and not* to be extensible in nature - because if some malicious code manages to get in the UEFI modules, you are simply hosed. But the other problem is that UEFI provides quite a lot of interfaces through ACPI and ACPI tables are buggy as hell. I had to patch DELL's DSDT tables in order to get back "deep" sleep - unfortunately it stopped working some time ago as they must have changed something somewehere and the computer would not wake up anymore... It is impossible to debug (or rather it is a royal pain to debug).
Anyone have a suggestion, how it would be possible to schedule an overnight backup on a Windows 11 laptop with Modern Standby? This used to be possible using the task scheduler, but that doesn't seem to work anymore.
Tangential but I found it amusing: today Windows rebooted while encrypting my disk with BitLocker, and decided that now would be a great time to retry those updates that bricked the laptop last time...
One aspect I've yet to see covered is how existing hardware doesn't seem to be designed for s0ix to work as intended anyway.
For example, on mobile phones it used to be the case that the application processor wasn't actually the main processor of the phone, it was actually the baseband, and relatedly, this meant that the application processor could be shutdown and woken up when the phone received a text message or phone call. Similarly, modern arm processors have a big/little configuration where you can switch to the efficiency cores when the system is asleep.
While I'm aware that at some point intel started offering efficiency cores, the vast majority of devices that run windows, AFAIK, don't have any way of being awake in sleep mode other that just leaving the CPU running with the screen off.
Maybe s0ix will make more sense as big.LITTLE and other secondary CPU schemes become common, but today seems like it's just upsetting existing users for no apparent gain
1. I'm even surprised that in the 21st century, personal computers (even in iPhone form!) still need to "boot" their operating system (not so: coffee maker, washing machine, TV).
This is bad UX, loses productivity (users waiting for the machines to fire up), and appears to be a relic - ancient machines needed it and then it was perhaps never questions.
2. For desktop PCs, it's not a problem if wake-up time is a bit slow, but laptops should be up instantly (<4s).
It's just futile to to get any kind of sleep working on Windows laptop. An alternative strategy is to just shut it down for long breaks, and keep it working for short breaks ("closing lid does nothing" setting), and that's it. Maybe you will be loosing a few minutes per YEAR waiting for boot, which is nothing compared to hours trying to get sleep working.
i have debloated my windows 10 system. i only use it for gaming and some work stuff. there is youtuber named chris titus tech who explains in detail how to have a debloated windows and also how to debload the graphics card.
I have never opened up my laptop thinking: oh I wish my new emails were already there.
Annoyingly when I enable S3 on my P1 Gen 5 Windows 11 give me a whole bunch of warnings in the security center.
Seems like S3 itself is incompatible with some modern firmware protections.
But frankly, I don't care whether it's S0 or S3. I just want it to sleep, and I don't want the OS to NOT offer me the options to say 'don't do stuff in sleep mode'
Oh, come on. S0ix is fine. If we don't want things to happen during this kind of sleep, we should arrange at the OS level for them not to happen. Nothing about the underlying mechanism forces your OS, at a policy level, to go download cat pictures and software updates while your machine is asleep.
This post reminds me of the people who complained about the transition from APM to ACPI. Despite much wailing and gnashing of teeth about "Why can't my laptop just support nice simple APM? Why do I bother with ACPI?", ACPI won, and we're all better off for it.
I get where you are coming from, but from a user's perspective I really struggle to see the benefits of S0ix.
More parts are kept on, for the sake of updates and notifications. Notifications that you will get on your phone anyway (and that the laptop won't be able to get if it's not connected to a wifi anyway), and updates that will likely cause too much excess heat to be safe to perform inside of a bag.
If you see the benefits more clearly, perhaps you could elaborate on what they are?
So tell me how to tell Windows to stop using modern standby. I've never seen any benefit to it, but I do see Windows misusing it all the time and I see my battery life be wasted due to S0ix. What the fuck is the point of sleep if it's just going to keep wasting my battery life by doing shit I don't need it to do. I want a sleep mode that works as advertised, not "I'll turn the screen off but do whatever the fuck I want when you're not looking".
I heard that S4 («Hibernate») is hidden on Windows, but you can still dig deep in the settings to configure it to happen when you press the Suspend button / close the lid ?
Regardless of a valid use case for "modern standby", there is DEFINITELY a use case for "old suspend to RAM" and "old hybrid suspend to RAM", which have de facto been blown away by the change.
I have no problem with adding a new sleep state that makes sense for other users. Go ahead and add S22 for that exotic sleep state manufacturing robots use between shifts. S344 for bluetooth devices waiting for reconnection. Whatever, I don't care. But don't replace an existing PRIMARY UI PATTERN with them.
To be fair this is not on microsoft for inventing S0, or even pushing for it for windows devices. It's on device manufacturers like Dell and Lenovo for dropping S3 from UEFI in favor of it.