There's the possibility to disable fast boot In the "Choose what the power buttons do" menu of the Power Options. That way a shutdown becomes an actual full shutdown.
Go to Control Panel\All Control Panel Items\Power Options\System Settings and untick "Turn on fast startup.." That means no more fake shutdown, slightly slower boot time.
Actually, there's a CMD console command that used to do all that. For instance, to turn off hibernate completely, you would use the command "powercfg -h off", and you would use "on" to re-enable it... Of course, I'm not 100% percent sure if that's something that's still available in windows or not in windows 10 and 11, aside from the fact that it APPEARS to work on my windows 10 system
Dave I have a suggestion: A complete, step by step cold boot of a modern day PC - from PSU power on to login screen. Mobo, intelME, memory check, TPM verification, BIOS, bootloaders, kernel... To really show us what really goes on while we wait for PC startup. I saw an article about Linux booting and it was fascinating to learn how top-level user accessible OS actually is and what is REALLY happening underneath it - and before it can happen.
Seconded on that. PCs have gotten wildly complex now. One of the things that has always frustrated me is the way that graphics on UNIX is such a s***show so it becomes actually quite difficult to do something like a perfectly clean boot with a simple logo all the way from BIOS to finished boot without the graphics flicking in an out. For instance on NetBSD version 7 they brought in DRM graphics from Linux and the way they did it doesn't really play that nice with the existing framework, it just seems to take over so the system would start in text mode then half way through suddenly jump into hi res DRM mode. And you often see a lot of Linux distros jumping around between a boot logo and then crap being spat out from the kernel to the console etc, screen flickering as it switches modes etc. Seamless boot is quite tricky to pull off it seems. Doesn't help that the freedesktop people aren't exactly the most transparent bunch of people and that PC graphics is still such a dark art. I once tried to fix a screen brightness support issue on NetBSD and ended up going through the kernel code for like ACPIVGA, DRM, wscons framework etc and it seemed such a god damn mess, it was like some Kafka nightmare. In the end I found an extremely dirty hack which just involved writing a PCI register that let me control brightness on my particular system but it turned out to be a VERY dangerous method. It worked on one laptop just fine but when I tried it on it on another laptop I couldn't get screen brightness working on it appeared to damage the LED PWM driver and the backlight would flicker badly for a few minutes every time you started it up.. god knows what I did but it did some actual damage!
To the person pushing a button to turn on a magic box, no. To a person writing a startup driver to make your magic box work, yes. It's not really a good reason to stop learning either way.
As a desktop support tech I hated Fast Startup. I'd have to explain to everyone why restart and shutdown were now different, and if I told someone to reboot I now have to specify restart and not shutdown.
A couple of enterprises I work with have all-SSD machines and they globally disable fast startup. The difference in boot time isn't worth the "What do you mean I don't reboot, I shut down every night" support cases.
It didn't take me long to notice that Windows 10 Shutdown was actually hibernating: 1: I had 64 GB RAM. 2: I used DriveSnapshot to backup my system, and it would not backup a hibernated system. Disabling Fast Startup in Power Options is easy enough, but I decided that I didn't want to keep blasting my SSDs with massive writes, so I simply disabled hibernation, and would just use Sleep. If I had a power loss, I wouldn't lose much, as I would make sure any current work was saved before Sleeping. If the power loss corrupted my system, I could just restore my system image if Windows was unable to repair the damage. Why Microsoft didn't simply add a Fast Shutdown option and keep Shutdown to mean what EVERYONE had got used to was just madness!
Fast Start in Windows is something I always disable, in today's age where SSD's are pretty much the standard, the bluescreens it often causes is more a bug than a feature.
Same here, I've always found it boots FASTER on an SSD with Fast Start disabled, it was designed to speed up loading from HDD and is detrimental on SSDs.
Raid arrays boot terribly slow. Fast boot doesn't matter to me because saving seconds on a 30 plus second boot isn't too impactful but I still leave fast boot on. I don't experience bluescreens or errors so there's no problems for me. So people reading this should probably opt to try it before disabling it. I use a lot of systems the newest being a 12900k/samsung 980pro Windows 11 based system and the oldest being a dual core from 2007 running 8.1 that runs media center, and machines in between 2007-2022
@@alexatkin And Fast Start Boot may cause problems when booting into another OS (if you use Windows and Linux on the same machine). As Windows with Fast Boot enabled stores the state of the NTFS partitions in the hybernation file and restores it on the next start, it may interfere with changes to that partitions from Linux (or any other OS). The worst case would be data loss.
I always find your videos informative about subtle things "under the hood" with Windows that I never knew before. Oh, you're of course entertaining as well and have an excellent delivery on the info. Great channel.
Love your channel. I was a pc consultant from 83 to 2004 then retired. Trying to get back into the field now with studying for my A+. the background you provide is priceless. Thank you.
In the control panel under System and security->Power Options->Change what the power buttons do; there is a checkbox for turn on fast startup that is checked by default. Unchecking it makes shutdown a proper shutdown again.
@@davybloggs1564 Near the top of that window there should be a UAC icon next to the sentence "Change settings that are currently unavailable" clicking on that should unlock the checkbox, but you may need administrator access for it to work. I am using win10.
Changes what the power buttons do; there is a checkbox for turn ON There is a difference in meaning between ON OFF and on off that people do not realize and are not taught by the corrupt education system Neither do the people actually notice - even when it is in plain sight - as the people 'have been turned into ZOMBIES ON OFF ( in the ALL CAPS ) relates to switching - - such as Turn the Light ON = or the Computer refuses to turn OFF { IT is also marked that way on switches - - so pay attention to detail ) on off is used in a different context: Such as - I start work on Monday and I am off on the week-end So learn it and reach it to others who are ignorant of this difference And remember - pay attention to detail - Male sure the brain sees exactly what the eyes are looking at Such as HOLLYWOOD and not Hollywood It is written on a Huge sign yet ZOMBIES cannot see it. ZOMBIES have eyes but cannot see - ears and cannot hear
Ironically by the time they started making these changes, SSDs were almost already ubiquitous, making the changes not only meaningless, but decreased the user's understanding of their system at the same time. Good job MS.
Nonono...they know this, but they still wanted faster boots! The goals were (as I recall) initially to get boot times down under 30 seconds, and then down under 10 seconds. So, full BIOS POST was ditched via "fast-boot" to reduce that initial 10-15 seconds of hardware verification by BIOS. Some of the shenanigans that were being pulled in this race to the bottom were ridiculous. And the canned response any time something Went Wrong was, just do a full boot. It was enough to make those of us in OS Test just want to pull out what little hair we had left.
@@Ughmahedhurtz Good thing the memory training times of DDR5 completely undid most of that effort spent on reducing boot times, haha. Well, on AM5 systems, anyway. The situation with Intel systems may be different, idk.
@@aaronjenkins9573 Memory (re)training is only one part of this, though. Granted, if it's taking stupid amounts of time on some systems, that's a problem. Funny thing is, my current systems where I run dual NVME SSDs in RAID-0 take much, MUCH less time to load the OS from scratch than the BIOS does to boot/detect/handoff. I haven't seen any of this retraining time BS on DDR4. Hopefully by the time I upgrade my 3900X to a 8950X, they'll have all that figured out.
Reading data from the SSD isn't the only thing that happens on boot. Your computer still has to process a lot of things before it can get into a usable state. And depending on how many things you want to have started up, this can still create a significant boot time. By packing away things that don't really need to be restarted from scratch and only need their state saved, you can still shave off a lot of time that way. You just load it back into memory and hit go.
@@Ughmahedhurtz which is very ironic when you have a 1.5GB/s M2 SSD capable of booting the system from cold boot in 5sec, which is faster than all that complex code running and doing a lot of shite . "Fast-Boot", more like complex-boot. I basically always disable it, with hibernation too, and I disable all the power saving extra states that only serve to throttle the throughput of the CPU when you need it. You know when you have a car and hit the pedal to the bottom, and it gasps, that's what I feel when I use stupid power saving states and speed-step or whatever Intel call that misfeature. I fucking don't want to save power, I want to have faster computer, if it uses 800W I literally don't care, I paid for it, and I want to use 100% of it all the time forever, until the planet becomes a death ball of global warming effects or until the heat death of the universe. To be fair, the system only uses 70W when doing nothing, which is very acceptable, its like leaving a incandescent lamp bulb on. It uses less power than my A/C unit which is always on and uses much more power than that.
Another great video. Thank you Dave. I got to 162 days on my MacBook several years ago. I bragged on some forum where a Windows guy (a friend) was complaining about having to reboot every week on average. Later that day, karma hit me. Xcode suddenly refused to compile my project. I tried all sorts. Reverted changes. Shut down Xcode etc etc. The errors made no sense whatsoever. Until I found a post online about Xcode having a memory leak. The humanity of it! I had to reboot. Almost on topic - I once took a call from an elderly member of the public who started with "I believe I have information which will lead to the arrest of theee international hacker" (if you're trying to imagine the accent, think of Tucker out of something about Mary). Temptation was to hangup on the premise it must be a colleague winding me up. After a short chat, it became obvious his computer was "waking up" every time his phone line rang, which his modem was connected to.
If Xcode has a memory leak closing it and reopening should fix it unless there's something I'm missing. Worst case logging off and logging back in would probably help if there are background components you can't normally close. If rebooting was the only fix for a memory leak it _must_ be an OS fault.
In Windows, _Shift + Shutdown_ (or _Shift + Ok_ in the _Alt + F4_ dialogue, with “Shutdown” selected in the select-box) overrides Fast-boot. The OS will shutdown with Fast-boot _temporarily_ disabled. _Shift + Restart_ restarts into a boot-options (recovery) menu.
@@davybloggs1564 - I have not upgraded to W11. I rely on desktop Linux as my primary OS. I keep a W10 partition only as a test platform. btw- _MorningStar_ is what I intended with my YT_ID.
@@davybloggs1564 - Actually, I am quite sleep deprived. Alt+F4 will close any _application_ (i.e. active window) in Windows 10, also. I do not have W10 loaded at the moment, but I believe that selecting the _desktop_ (i.e. left-click on desktop wallpaper) will place focus on the operating system for Alt+F4; else, closing all windows and then pressing Alt+F4. Not certain --- I always close all apps _before_ I attempt to shutdown/restart.
I prefer Win+X to get the power context menu, then getting the shutdown/restart options from there, a bit cleaner than getting focus on the desktop then Alt+F4ing. I suppose you could Win+D to get the desktop first but the Win+X menu seems more reliable to me.
Came into this video thinking I already knew all the things end users should know about shutdown/restart/sleep/hibernation... But this video still taught me the new knowledge that hibernations, which I use daily on my laptop, still consume a tiny bit of power to facilitate WOLs. Since I have no utility for WOL, it's now disabled on all my network adapters. Thanks Dave!
No mention of modern standby (in newer laptops) which breaks all the normal power management settings. What a pain it was trying to figure out why my system would go to sleep even if I had power management set to never sleep.
Dave. What can I say? I hate watching TH-cam videos that are more than 5 minutes long. Your videos are an exception. I listen to them more than once. A veritable wellspring of sagacity and practical knowledge. Well done. I gotta get back to my second listening. I like to vary the playback speeds as well. Both faster and slower than the default.
Thanks, Dave. As a true geek who hales all the way back to the preDOS days, I truly enjoy your videos… I invariably learn something from each one and love the ‘inside baseball’ commentary.
Always a fan of your videos... The bit of history always brings me back to some period of tinkering around... This time adding hibernate to my win98 desktop PC back in the day and feeling cool. 😎
I KNEW IT!!!!!!!!!!! Thank you!!! I thought there was something odd about Rebooting. When we do it to help an end user, some programs are sometimes still running like Excel or Word. Being good at troubleshooting and BS I tell people I think it may only be going to hibernate before starting again. I just now shared this video with my coworkers to confirm what I have been telling them the past year.
I learned from my experienced computer tutors in the eighties, the term ‚bootstrap‘ was related to a short piece of punched paper tape, about the length of a bootstrap, that was inserted into the reader to start up a computer and enable it to start the programs fed in from the upcoming spools of punched tape / cards or similar storage devices. So ‚boot‘ the machine was likely an abbreviated way of saying ‚let the run through the machine‘.
That story may be a “back etymology,” since not all computers in the “old days” used paper tape. The Digital PDP-5 used papertape but the “boot” was entered using the panel switches (IIRC, 7 12-bit words), which then allowed a full papertape program to be loaded. Most IBM computers before the 360 had a separate LOAD button on the card reader, and/or each magnetic tape drive. The 360 allowed booting from any device, usually an installed operating system, by dialing the device address into a set of switches and pressing the LOAD button next to them (but IBM always called it Initial Program Load, or IPL, from the 360 onward). So I would go with the “pull yourself up” theory, especially since an operating system had to open files before the full file open/close software was loaded, read JCL to set up the JCL reader and initiator tasks, etc, requiring partial function duplication in different programs.
In a comp sci class in the 70s I was told it was from "pulling yourself up by your bootstraps". There were also some old cartoons that played on this trope of being able to pull yourself off the ground by grabbing your shoe laces. It's an old joke.
"Bootstrapping" is the term for a self-starting process in multiple fields, it's the self-starting part that is the common root, not anything computer-related. That comes from the pulling yourself up by your own bootstraps saying. (Which itself is derived from other earlier sayings.)
Back in the day, I had spliced together a dual-boot Windows 98 and Loop-Linux machine. The two sides could comfortably work on the FAT32 partition. I had to compile the source for the Linux network driver or the Linksys card I had in this unit. It worked great-- except when it didn't. At first, it seemed very random. Sometimes when I booted into Linux, the network card just wouldn't work. After a day or so, I realized-- when I cold-booted to the Linux side, network card always worked. But when I rebooted to Linux, after running Windows, network card wouldn't work. Back then machines turned all the way off, or you had to turn them off physically with a switch or button after "shutting down". Then I realized what was happening: the NIC wasn't completely refreshing after being initialized by Windows. It was remembering whatever Windows had told it between reboots-- i.e. It's internal memory wasn't being cleared. This prevented Linux from communicating with it properly. Oddly, the reverse was not true-- I could reboot from Linux to Windows and it would work fine. So, either Linux was being nice and clearing the NIC's memory when going down on a warm boot, or Windows was being a jerk and not clearing the NIC's memory on warm boot, or the opposite: Windows may have been smart enough to clear the NIC's memory when starting up on warm boot, and Linux was not. Hard to say. Anyway-- it is true, memory may not be cleared between shutdown and startup on a warm boot. Today, thankfully, things are generally simpler. However, there is that: As long as power is supplied to RAM of any sort, it only clears it's contents if those contents are explicitly over-ridden. The only other way to clear content's of RAM is to completely remove power.
I usually turn off fast startup for two reasons: 1. This does not work correctly on some of my (slightly outdated) hardware. 2. As far as I understand, this does not record information that the NTFS partition was correctly unmounted. Which causes problems in other OSes. They assume that the filesystem has not been correctly unmounted, it may contain some errors, and, as a precaution, do not allow write-mounts.
Regarding point 2, I believe it _isn't_ properly unmounted even though it's fully consistent. The FS driver in Windows probably has some state stored in "RAM" and if the actual state is changed under it bad things could happen. Having said that it _is_ very annoying.
Reminds me something about the 1982 Sinclair Spectrum design: its CPU didn't utilize the Z80's NMI vector so asserting the NMI pin on CPU just caused a hard reset. So the trick was to wire a press button to the NMI and patch the ROM with code to do something really neat like run an executive user routine with options to save or load a snapshot 'hibernate' image to/from FDD or print the screen etc.
The hibernation file is a great resource for anyone who wants to know what you were doing at the time the PC last went into hibernation. If you were accessing encrypted data, it will all be there in its unencrypted state in the hibernation file, as will all the encryption keys, passwords etc that were active at the time the PC went into hibernation, and any documents you had open at the time.
false - unless you using windows wrong, things are encrypted whenever written to disk, no matter what. I guess it depends on your settings and windows version, though.
Even while I knew most of this, I still learned quite a good bit! Thanks Dave for the knowledge. May I ask your take on the CPU Management Engine features - which sort of steps into some of the key points discussed on this video,?
That's what I liked about it. I'm very technical, so I know what the different options are. But this video casually gets into where these options make the most sense, and what to expect (boot time, power consumption, hard drive grinding, etc) with each of the optons. Pretty thorough presentation.
My father was an experienced MS-DOS user and solved most issues with “format c:”, not simply a reboot. When he got old it became a problem because he started to apply his DOS skills to his Windows 7 and refused to relinguish his administrator user. Hence I would get a phone call in the middle of the night saying “My computer doesn’t work.” “What happened?” “I formatted the harddrive.” Thankfully, being an engineer he was happy to use a system other engineers would use as well, so I was allowed to install Linux on his computer.
interesting explanation thanks Dave. On a related note for years now I've recommended people configure their laptops to hibernate rather than sleep not just as a power saving method but also for stability reasons. its not uncommon for laptops to have issues waking up from sleep especially if its a new model this is doubly true for Linux users where there may be a driver fix for the issue on windows but not on Linux. Aside from the first few years when the feature was supported I have never seen a machine ever have an issue waking up from hibernation. Also if you are one of those now rare dual booters who haven't switched to WSL it can also be helpful if you want to quickly switch back and forth between operating systems. just remember on the Linux side the swap partition is used to save the state so you must have a swap file big enough for the content of both your system and GPU ram and you should probably reduce your "vm_swappiness" possibly to 0 so you don't burn out your SSD.
Disable 'fast startup' in settings. Might have to ungrey it by selecting "Choose what power buttons do" and "Change settings that are currently unavailable." Presto. Shutdown is shutdown. If paranoid, additionally disable 'hibernation.' In most BIOS, you can prohibit fast startups. Great video 👍 Kindest regards, neighbours and friends.
I've found that hybernation(only) has been far more reliable, i've noticed that sleep, and hybrid sleep often result in network and USB not always working properly after wake, its rare, but when you manage over 1000 different machines, you tend to notice a higher ticket volume for wifi and keyboards not working when you enabled sleep because you wanted to save power/heat across all campuses.
Exactly what we found. When I replaced everyone's machines a few years back we got some back that had uptime of months. And users do not know the difference between a shutdown and a restart. so even when you ask them to restart they don't and then you end up with a system with 90 days uptime and you can't convince them that they did the wrong thing. A quick disable of fast boot and hybrid sleep in group policy and we reduced our random error tickets by about 75%
@@adamgeorge8327 I find an alarming number of users will restart the computer by pressing the power button on the monitor, even in out computer science department
@@denvera1g1 With my first PC, the IBM PS/1 (model 2??? - the question marks being numbers I don't remember right now, but it was a 286), that was actually the proper way ;) It only had a power button in monitor and the monitor VGA and power cables to the PC itself. The PC had no power button :) It would have prevented anyone without some tech savvy from replacing the monitor with some generic VGA monitor and have it working... There would've been no way to boot up the system. Though the power connector on the PC was probably just a standard 3-prong connector, so it shouldn't have been that hard to get it powered up anyway. Just thought I'd mention.
@@denvera1g1 There are times where you have to use the power button (though you hold it rather than just pushing it). I always run an actual restart after that though.
@@karlrovey I still have the power button connected electrically, but group policy disables the windows power button/lid close action. However the BIOS will still normally cut power from the device after 4-15 seconds depending on the model. We have a group of laptops for a VR class that shutdown after 15 seconds of holding down the power button, i've never seen a machine take longer than 5 seconds before then, and if i didnt have more than one laptop i'd say the power button was defective and never would have held it for longer. The weird thing is that after 5 seconds of holding down the power button, Windows shows this "swipe up to shut down" option, which 1: i dont want to do because you're acting up and i just want to re-image you, and 2: i cant do because this isnt the touch screen version.
Turning off fast start-up is the first thing I do after setting up a new Windows machine. I'm more than happy to wait a few extra seconds every morning in order to ensure that patches etc. are all installed properly.
Great video! Really enjoyed the history of booting a system, really helps to visualize what is going on. I work in IT and ever since Fast Boot became a thing we have been disabling it for our end-users because when we ask them to do a restart they end up powering off thinking that it was even more helpful. Microsoft must have caught on to the issue Fast Boot was causing, as now when a your system needs a reboot due to a Windows update, the power-off option adds a new "update and power off" selection.
@@anon_y_mousse While that is true, it’s also true that at the same time they did this, almost no one noticed, because they made it that much more reliable (although still not perfect).
I had to turn off wake on lan because I couldnt send my computer to sleep: Every single time the computer would wake back up by itself shortly after. Sometimes literally 20 seconds after having gone to "sleep". Everyone had told me that it was my fault, that the mouse moved or whatever, but even unplugging any entry devices would bring the same situation. It only got solved upon disabling WoL and I havent had any issues since.
I've been straightened out like a pretzel. Thanks Microsoft! You should have included some tips on disabling prefetch and superfetch, and these pseudo shutdown options in the registry. I've disabled prefetch for my Windows systems since they run on SSDs and having copies of files for a quick boot is a waste of space. Great channel, thanks Dave!
You should've mentioned the (annoying) Modern Sleep that windows has on newer laptops. It's annoying because I don't want my surface to overheat in the bag and I can't enable regular sleep in any way. I just hibernate it way too often instead. For a long time I had fast startup switched off because I was told it was better, but when I turned it back on the boot speed went from 22 secs to 15 secs, which is worth it for me.
Great episode! I'm telling all my programmer friends about your channel. 3 things: 1.) So, for those us who got it but need reinforcement, like, need to write it down: Restart - clears everything and brings up a fully reset state. AND Hibernate (set for "When lid is closed") is more desireable than power off because it does power off but on wake quickly returns to where I left off with minamal power consumption. Right? 2.) To optimize the longevity of my laptop battery, I guess leave it plugged in and have "close the lid" hibernate the laptop? 3.) Do you happen to know or giess why MS doesnt have a battery meter that reminds the user to plug-in the charger when it drops below, say, 20%. And another notification or bell to tell the user the batts has reached 100% charge (or should it be 85%?) and to unplugg the charger? I like watching your channel. It brings back lots of good memories: The PDP-11/70 days were lots of fun, UNIX and C had there moments too.
Another great video, thanks. That shutdown/power-on no longer does a full system refresh by default is one of MS's worst decisions. Naive users dream that they are doing a system refresh and they are not. I ALWAYS disable Fast-Startup on my own system or others because I want to be able to choose Sleep/Hibernate/Shutdown myself. Naive users need shutdown to mean shutdown. There are ZERO cases where it makes sense to have Fast-Startup enabled, either users are naive and need it to be a shutdown or they are not and they can choose Sleep/Hibernate/Shutdown themselves. This feature caused no end of trouble with the hundreds of users that I recently did phone/remote support for.
Indeed, and the time it saves on boot (I can't tell tbh, can't imagine it being more then one or two seconds if that) on new pc's is thus negligible, like I'd guess my pc "slow" boots in five or six seconds or something, like what are we talking about, what does a second faster accomplish when it's already that fast to begin with? Not worth the trouble imo.
I always disable Fast Startup so it acts like old school shut down. SSD's have made cold booting so much faster, that it no longer seems necessary to have Fast Startup enabled with an SSD anyway.
The only REAL COLD boot is removing the power cord from the computer power supply. I've had networking issues surviving multiple cold reboots and restart cycles because the NIC card remains powered on via the standby 5 volt supply, that only goes down if the plug is pulled! I have lost count of how many times this has bitten me on-site! The NIC card holds the ARP lookup table, and DHCP gets confused and will NOT serve an IP reservation until you pull the plug!
Throwing the big red toggle switch in the back from 1 to 0 pretty much does the same thing. If you don't trust that, you could use a powerstrip with a switch on it too. Pulling the cord is not necessary, and has a slight chance of causing other issues, because of the loss of grounding, and possible sparking.
@@joesterling4299 It may surprise you but not every power supply has an off switch and and the one I have is black. Cheap chords have switches without dampening and bad design which can spark worse then pulling the cord.
Dave is the best technical teacher I've ever encountered. He knows his material perfectly, doesn't make the mistakes that many instructors make, and has an engaging and personable style. He's instantly likeable. Thank you, Dave, for doing this channel!
This is actually why I always (on my own laptops) always turn-off this. Do a shutdown/power-off/restart cycle every 1-2 weeks. Too many apps still have memory leaks or leave external hw in an undetermined state. My work computers always get shutdown at EOB and started from fresh SOB.
Disabling "Fast Boot" also will cut off power from USB ports, unless the motherboard have a feature to power some of USB ports (many of today MBs do offer that feature) in order to charge phone/tablet while PC is off. If the "Fast Boot" is enabled (default) the USB ports are powered even if PC is off.
the irony is microsoft brought us fast boot exactly at a time where its no longer beneficial. thanks to modern hyperspeed storage the differences are well... irrelevant in any practical way. so lets introduce more complexity into errors to have single digit percent faster boot shall we
Not everyone has SSDs for storage, especially laptops where if you want a less expensive one to have 1TB, it will be mechanical. If you want SSD but less expensive laptop, you get 120GB
@@danman32 I hate fast startup with a passion, with mech drives my experience from clients is that you needed 8gb ram even at windows 8 times, otherwise fast startup would bog down your PC with them missing reboot, it would also expose driver issues more often and after that it would require a real reboot. One of the first things I check when someone calls me for issues is the task manager wake timer, if it's more than 15 days is most likely that pc needs a real reboot plus disable the fast startup
@@MrCrashDavi It might not be because even though the kernel is saved to disk, it is still considered in a shutdown state. For WOL to work, it had to be in sleep or hibernate state.
It is not the same speed. Hibernation is indeed faster, as it only loads from disk to memory, without the need of processing. I've loved that fast boot feature, even though I have always hibernated since XP, because it was blindly fast back then compared to cold booting.
This reminds me of the days of Win95 and NT4. There was no sound driver for NT4 and the Win95 drivers didn't work. But, when you first booted up Win95, did a soft reboot and then started NT4 you did have sound like normal.
Nice. I consider myself one of the ancients as well sir. Started one of the earliest ISPs here in MN, owned one of the earliest clones available. Did all the memory mgmt of card installations in the old dos days. Saw the first browser come in with CERN, a DoD site, and one other (I think a college site) and that was it. I think people don't get the value of understanding how things work at a low level in relation to today's commonly accepted complexity. I learned a bit on this one and appreciate the vid :) I wish you the best.
Fast boot is something I've turned off instatly in BIOS on all new machines since its first introduction, but cool to konw more "behind the scenes" things as usual in these videos!
On Linux you can just set the WoL flag on your NIC at any time; I usually do it in a startup script. Interesting to hear how Windows handles this automatically depending on the sleep mode. Not sure if mentioning Linux on this channel is heresy though... 🤔
As a Puppy Slacko user I now recall my Windows days as like holding a very valuable thin crystal wine-glass that I could easily crush in my puny fist and therefore a constant worry that It would break. Using Slacko is like having a shipping container of restaurant-grade tumblers at my disposal.
Only maybe 10 days ago I set this up on my Linux desktop. It was a 2 step process, enabling in the BIOS and using the set option in Ethtool to enable, that apparently can get overwritten on restart so also make it an option in Network Management to respect the 'magic packet'
This channel is often informative, but apart from the clickbait title "You're Doing it Wrong", I won't be changing my shutdown/restart habits after watching this video.
There's also usually a poorly-named option in the BIOS somewhere under the Advanced Power Management to say "when you're off, turn all the way off." Otherwise, the RGB lights keep going and dumb crap like that. The EU required motherboards to be able to actually turn all the way off.
Hey Dave, Great video as always. I was surprised you didn't cover the practice of holding the left shift key while click shutdown in Windows 10. I've found that to be quite useful for getting a fresh boot to run utilities or a live disk from a flash drive. Any thoughts or corrections?
My routine after reinstalling windows is to disable the hibernate function using Command Prompt in admin mode with "powercfg -h off" (I never use it as I had issues in the early days of it and disk space was more important than hibernating back in the day), then I disable "fast start up", disable screensavers and set it to turn monitors off after 15mins of no use. Then I'm very much like Dave, where my PC will stay on and will only get a restart every once and while.
A few years ago, I hibernated my windows 10 laptop and popped it into my rucksack to take home. On arriving home I could smell a very hot electrical smell. During my bike ride home windows had decided to power back up and attempt some update and due to the vents being blocked became dangerously hot. The end result was the battery pack being ruined and I no longer using Windows 10 on my laptop again.
I had installed windows on my laptop for about a year. And whole this year I was trying to fight with it's power ups. It does not matter if it was hibernated, or in the sleep mode, it always was turning on and checking for [censored] windows updates, mail/messengers notifications, and etc. I tried all the ways to disable this behavior I could find on the internet so far (turning off wake on lan, wake on usb device, etc.), but nothing could help me to resolve this start ups when the laptop is closed. It was so stupid that I close my laptop in the evening, and when I wake up it's completely uncharged, the battery is 0 percent. The only fix I could find is buying Mac. Now I can close my laptop and be sure if I open it next time it has the same battery level as before.
@@mekedron THose wakeups are really annoying. I stopped that by running Taskschd.msc as TrusterInstaller, disabling UpdateOrchestrator's wake task and denying write access for that task file for System. Now Windows can't reactivate that wake task after new updates are downloaded and ready for installation. :D
Dave, Wake on Lan does work on PC's that are "fully off" (not just hybrid sleep) if enabled in the BIOS. Sometimes this is called "Power-on on PCI-E" or similar. We have over 250 PC's of various models at work (mostly Gigabyte and ASUS motherboards) and I can assure you every single one supports Ethernet Wake on LAN from a fully off state, and did so both on Windows 7 (which only had a true shutdown) and now on Windows 10 with or without quick shutdown enabled. (We have it disabled with group policy) This is a feature of the BIOS and onboard ethernet hardware and nothing to do with Windows. (It works even if Windows is not installed or some other OS is) The Windows ethernet driver for some network adaptors offers the option to disable wake on lan - however this disabling of wake on lan in the windows driver only takes effect after windows boots up not when power is initially applied to the system, which is controlled by the BIOS setting. So wake on lan will still work once after power is first applied but will not work after a windows shutdown/sleep if disabled in the driver as it gets disabled when windows boots. So for it to always work it has to be enabled in both the BIOS and in the windows driver. There are also two types of wake on lan - what you describe with the MAC address repeated 16 times is "wake on magic packet", this is OS independent (supported in the BIOS from both fully off and standby) and works on all systems that support wake on lan. Then there is "Wake on pattern match" - where Windows or other OS can program additional custom pattern matches into the ethernet hardware at run time via the ethernet driver. This is how waking on arp resolution etc works as Windows configures a pattern match in the ethernet hardware on the fly which matches arp packets that the PC would normally respond to based on it's IP address, then goes to sleep. When a matching ARP packet is received the PC wakes. As far as I know this type of Wake on lan doesn't work when the PC is fully off, only when it is in standby/hibernate/sleep. Not all ethernet adaptors support this second wake method. If you don't see an option for it in the driver it probably doesn't. If it does have the option you can disable wake on pattern match while still leaving wake on magic packet enabled - which I've done on my PC at home because some other device on the network querying it using arp would keep waking it up, yet I still want to explicitly wake it using wake on magic packet.
I've also recently had to disable the second method on one of my machines as it was waking on spurious packets after a driver update. That said, wake on patten can work if off (well in ACPI S5) but it is highly hardware dependent the driver basically has to be able to save the patterns to wake on in the card itself.
“Pulling themselves up by the bootstraps” was always intended to be satirical - at the time in the Victorian era Malthusian philosophy was popular i.e. being poor was a choice and a moral failing. So, official policy on poverty was often shaped around this idea that a person could become a productive member of society on choice alone - if you just cleaned your room, took some responsibility and hustled then all your problems would go away, precisely as true then as it is now: misleading, naive and incomplete at best. To illustrate how ridiculous it was that policy aiming to solve a structural problem was entirely focused on individuals, the phrase was picked because of how obviously absurd it is. It’s only been “rehabilitated” relatively recently, and tbh I think I would prefer it with its older meaning
I enjoyed the entire video, but as a guy who has loaded in the bootstrap from front panel switches, I especially liked the explanation of the term "Boot".
What Dave didn't mention is the so called "Modern Standby" which Microsoft insists on for laptops since some time. This mode keeps your computer in S1 (fully on) state but is supposed to make your processes and devices aware the computer is supposed to be asleep and to behave themselves, only allowing for light activity like periodical e-mail checking, etc. It practice, the implementation is broken and rarely works as intended, often making your laptop cook itself to a near death in your bag or backpack, until the last resort thermal failsafes kick in (or not).
Yep. It's horribly implemented and ignores your power management settings. For example, if you close your lid but want the computer to 'do nothing', modern standby will still put your system to sleep after 15 minutes. Infuriating, but setting the BIOS/UEFI firmware sleep mode to 'Linux' restores classic, working, functionality.
I noticed this omission too, and agree that Modern Standby and Connected Standby as it was called in Windows 8 is a DISASTER. I have had the issue with the laptop overheating in the laptop bag several times, to the point that even the glue on the rubber strips on the bottom of my Dell XPS 15 melted and let loose. Fortunately, it can easily be disabled in the registry: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Power CSEnabled=0
I forgot what that was called and I never figured out how it worked before now. If it's keeping the computer fully on that's probably the stupidest thing ever. I'll need to make sure I have that fully disabled on my WIn10 laptop.
On my main system, which pretty much runs everything on the property including surveillance and media server, I have it set to do a restart at 5:30 AM every day. This might be excessive but it always gives me a fresh load to start the day with and that's worked out well for me. It's worth the 15 minutes it takes to restart and fully load everything from scratch with fast boot/fast start disabled and a full fresh load of PCI and USB in the BIOS. This also incidentally reveals any impending hardware failures that might manifest on a power recycle. Better to find them in the morning when I can do something about them than have them randomly fail at unexpected times. When I do shut the machine down all the way for any reason I have the surge suppressor that serves the machine mounted where it's easy to get to. I just flip the switch on the surge suppressor which removes power completely from the machine, wait at least 30 seconds for the caps to drain and then reapply power to the system.
I still have machines with Win7 that support WOL because WOL is a feature of the LAN card and lately the BIOS and not of the OS. A LAN card must have 5VDC always for it to work. And the IP is not needed in reality since the magic packet is broadcasted to the entire network, not just one or any IP. You can anyways use the IP to ping the OS when it's up and running :) Great content and channel, Dave!
On the first PC I had, a 486 DX-66 I actually used to enjoy watching the memory check as it literally counted the bytes of my 4MB ram. Imagine how exciting it was when I installed an additional 4MB and maxed out the 8 memory slots it had.
Once every 2 days was the statistic found by CERN (although different papers seem to disagree a lot). A single bit out of the billions in RAM every 2 days honestly doesn't seem like much to me, unless you never close any programs ever, and those programs aren't constantly modifying the state of their memory (fixing the issues). Not to mention that ECC RAM makes these errors even less likely to have any real-world impact since they can be corrected whenever they are found.
@@miguelguthridge Indeed, and I've only powered off any of my computers in the past decade because of power outages. Although I also don't use Windows, so that might make a difference.
@@miguelguthridge Well the overwhelming majority of PC's doesn't run ECC, so that's not really that relevant for most users. Now it'd be interested in learning how DDR5 handles bit flips as it uses a form of ECC internally on the memory modules. I've worked in the PC field far longer than I care to admit, and I've always tried to steer people towards ECC memory when they spec out workstations. If your personal gaming machine crashes because of a bit flip it's unlikely to cause any long lasting problems or major costs. But in a machine used for production or economics it can be disastrous. Now if we look at the likelihood that a bit flip will cause a crash it's actually quite small. Of all the memory used by the system the absolute majority will contain graphical data, sound data and things like that. The actual code that's executable is way smaller than all the "fluff" data that memory is filled with. And not even all the executable code will actually be executed. There is a lot of code in conditional branches and error handling that will only be run very infrequently. If the bit flip occurred in code like that you might never know. And if it's in graphics or sound data chances are you wouldn't notice. Even a crash isn't what you are really afraid of. A crash will self correct the problem as it makes you aware that something is wrong. When the bit flip occurs in critical data used for calculations or in data that is the result of calculations that hasn't been written to disk yet, that's when things get scary. Unless the data has built in parity or there is some kind of sanity check on the calculations or perhaps redundant computers performing the calculations then a bit flip can go by unchallenged and the corrupted data gets logged as being good. So bit flips happens, and no, that your computer doesn't constantly crash isn't prof that it's not happening to you. And as memory size increase as does data density there are more cells and they are smaller. That means the chance that a bit flips goes up both because there are more of them, but also because the smaller the cells get the easier they are to flip. The use of ECC in the DDR5 memory modules is one sign of things getting critical, but I wish they had chosen to standardize on full ECC capability for every thing memory related. Then we would be able to see when there is a correction done and have a chance to measure the frequency with which they occur. But then I guess they don't want end users to worry about these things. Out of sight, out of mind...
@@mikemx55 I'd argue that 90% or more of the memory is used for things other than executable code, and most of it is non vital data like graphics assets, sounds and so on. You would be hard pressed to even notice a bitflip in all this data, and so you can be sure that the absolute majority of them are never discovered.
It took 10 years for laptop bios to properly restore the computer after a suspension (not hibernation). The computer powered up, but the ethernet and WiFi wasn't connecting, external USB disks were lost, Bluetooth was disconnected (A2DP devices especially), screen brightness was messed up, and more. Perhaps you can shed a light on why it took so long for laptop manufacturers to mend an apparently straightforward operation. Thank you for the trip down the memory lane... P.S.: the best moments of my long IT experience are those of the launch of Windows XP (after being obliterated by the Trumpet Winsocks). My worst IT moments were at the time of the 486/66 and Local Bus computers. Thank you once more. Regards... Anthony - from the UK.
Most of the issues you are describing will mostly be OS (and to a lesser extend device-based) as AFAIK the main BIOS is only responsible for keeping RAM intact and re-entering the OS. Actually ACPI code (which is stored in the BIOS) is likely responsible for much of your problems as it can potentially tell hardware to cut power to certain components. In any case I don't remember having many problems like you describe so maybe you were just unlucky with your machine models. And of course some things like Bluetooth devices being disconnected are going to be because it's physically impossible for a device to stay connected to an unpowered host. For USB disks being lost, it could also be drives which aren't able to reconnect in a reasonable amount of time. I've encountered that with some devices but usually flash drives work just fine.
What would be better is if it built a 'blank hibernation' state where it boots the system, and BEFORE giving the user control but AFTER loading ALL services (including any post login services) it creates an image of the blank system. That would allow it to accelerate the boot AND fix issues that have built up over use. This would not replace standard hibernation, but be more for boot from power off. Save the boot bit by bit for recovery modes and such.
Very well made video. To be honest, I didn't learn anything new, but for users it is explained perfectly well and in understandable language. The most interesting part for me was your bio. I'm kind of an old dude and I've been using probably all or almost all of the OS you have worked on in the past -starting with MS-DOS. Interesting channel, so I've just subscribed.
Hibernation has been one of the best inventions ever in PC world. I've enjoyed it since XP times, when I was using a helluva slow Dell desktop, which took forever to boot up properly vs. a minute or two coming out of hibernation. so for me, fast startup was a welcome addition in later Windows versions. as a bonus, I always know what to tell folks who say "why is my Task Manager showing 5 days of uptime when I shut the PC down every night?"
Woahhhhh. For so long I've been meaning to look up why it takes noticeably longer for the computer to shut off when I picked Restart than when I picked Shut Down. Finally this great mystery is solved!
I always disable fast startup as I had a mystery for awhile where users would report bad performance and slow applications on their PC. They would shutdown their PCs nightly, so always appeared to be a program issue/hardware or the moon in the wrong phase. Wasn't until we started deploying RMM agents and started seeing PCs with an uptime of weeks when users said they were powering off every night did research discover that shutting down isn't actually shutting down. What was considered by software engineers as a good feature to me is just giving a user more rope to hang themselves with. I firmly believe in the phrase, "Make something idiot proof, the world will build a better idiot."
Love your stories Dave. I’ve been interested in computers & software since I was 14 and had access to the TRS-80 back in 1981. At first, I learned simple programming with an obsolete punch card system. It was a simple calendar generator we programmed with the only variables being the month & year and it would print out a one month calendar. Everyone did the current M/Y or their birth M/Y, and I did January 2000, thinking about how far away that seemed(little did I know about the havoc THAT date would create, eh?) Then I learned to program BASIC. I was introduced into manipulating the hardware with the common single to double side 5 1/4” diskette conversion via a hole punch. Next I learned to solder and there began my love affair with the computer & gaming world. You had my sub long ago and you get my like with every video I watch - keep up the good work!
obsolete punch card system? You implying they don't make 029 keypunch machines anymore? That means no chads for the ticker tape parade. (where did that memory come from? Chads?) More humor about the computer having a crank on the side. 1974 BSCS @darrenlesage2420
when in doubt you can always flip the button on your power supply (desktop assumed) easily some of the best content around. thanks for the education and laughs!
Would love to see your take on the hot mess that is Modern Standby (S0) and the removal of other standby options. I've not found modern standby to be all that good and have had laptops wake up while in a bag resulting in overheating. To make it worse on some (maybe all) laptops that have S0 enabled, all other sleep options are not available.
I have had this issue too and it freaks me out, making me think it's going to catch on fire while **I** am sleeping, catching my house on fire (it's a very high wattage computer).
Back in the 90's, me and my tech friends would refer to sleep mode as "coma mode" because upon awake, sometimes it ran slower or had trouble loading things into RAM. ... or maybe I just had crap hardware at the time.
That one watt of power matters a lot more when I place the laptop into a backpack immediately upon shutdown. I can't tell you the number of times I've opened my backpack to pull the thing out only to find a sauna inside the backpack and the fans running full blast. It's due to this reason that I use Hibernate so much.
Amazes me that after 40 some years of making laptops the industry still hasn't sorted that one out. Same thing's happened to me even with non-HD notebooks.
The problem with fast boot is that sometimes it doesn't work. It just won't allow you to shut down your computer. On the internet they say that it is only the problem with old motherboards but I actually had this problem on a 2020 motherboard with the latest UEFI.
Ha Ha Not much use on a laptop. And considering you’re dealing with windows, which needs to be able to juggle a gazillion different hardware configurations, not necessarily a smart thing to do. Back in the DOS days no problem, and sometimes even necessary.
The number of times in the 90s I had to physically attend to a workstation because the user would not accept that turning the monitor off and on was not actually power cycling the system
Interesting subject. Some time back at a Microsoft partner seminar (Win 8 time) one of the most senior Microsoft chaps answered questions saying the software had been designed to "clean up" when the MS Operating System when it closes down fully. Hence delays in start up from a complete full shut down (electric shut off) should be used regularly to ensure BIOS and device parts with BIOS type instructions can be refreshed. It was also why so many repairers and advice was to shut the system of, turning off power, to overcome possible corruption occurring when the system 'saved' in quick start methods and the device parts had actually been the cause source and the total power off restart had proved to best and so often got rid of the some issues which otherwise returned from a "hot boot". However, you do seem to support regular power off on the system as recommended by Microsoft. From experience in attempting to diagnose between hardware and operating system my vote is for power off if a hot boot does not resolve an issue or the type of issue recurs. Often the quickest resolution although I do know a few so called techs that do repairs who often just give up in seconds and claim that hard drive has given up or OS is corrupted. Then they proceed to make the quickest "repair" by charging for new hard drive installation or total wiping and re-installation of the OS which unnecessarily results in complete destruction of easily recoverable data. The old "Ghost in the hardware" however is a very interesting experience. Create a new system with all new parts and OS. Starts up and falls over just as quick. Software diagnosis reports all good. Test all hardware individually. All good. Hmm. As you are putting several systems together with exactly the same part requirements you construct all the others which work well. Take individual parts from the "Ghost machine" and put each in a different machine to each other. All systems work perfectly. ??? In the future you never hear from the users of those machines of them having any problem of failure. ??? So what caused the problem with that one machine which became just one of the number that finally went out the door? I just love computing. Some variables you can come across can be so exciting... NOT.
Dave, you might be upset to know that some companies restrict access to the task manager, requiring an admin password just to see what processes are locking up on your pc
That is not so much the fault of the companies as it is of Microsoft to allow admins to have the option of doing so. In my experience, most admins seem to take the attitude of "Who are YOU and why are you messing with MY computer system?" Out of laziness and arrogance they tend to lock down whatever they can just because they can. If you challenge it they will tell you that it is "for security" but cannot articulate how that is so.
@@ravenbarsrepairs5594 Did that a year ago on my computer and am in the process of putting together a Linux system for my wife as I will not allow Windows 11 and all of its data gathering into my house.
@@WatchesTrainsAndRockets part of restricting access is to protect the worker. "Did you install that program that stole all our data?" "No sir, the admin locked my pc, so I can't install anything"
@@LeifNelandDk I NEVER said that restricting access was a bad thing. What I said is that things such as task manager need not and ought not be capable of restriction. If it is built so that it will not allow a user to affect a system process nor other users' processes, there is no security issue with allowing a user to run task manager in order to control their own processes.
At my first job in 1967 at Purdue University's High Energy Physics research group, we had an IBM 1401 card system. There was no way to boot an application from a disk as it was a card-based system . One of the programmers (Roger Wagner) stored an image of the boot deck on a hard drive (IBM 1311 in this case) and created a single-card boot program which read that image from disk, thus only the one card was required to start up. That card was called a "RABIT" card, which stood for "Really A Bitch In Theory" NB: IIRC IBM didn't use the term IPL until System 360 in 1964
I am THAT guy. Common sense dictates that shutting the machine down is the mostest bestest. In our upside down inside out world, of course piddly restart works better. Massive tip. Thanks!
We use MS Win 10 Surface laptops in a production environment, rolling test stations with lots of USB devices attached. Occasionally, maybe once every few months in the summer, more frequently in the winter due to static electricity being worse, the USB hub internal to the laptop will get glitched and the only way to reset those internal hubs is to completely power down the laptop. You can reboot until the cows come home, run updates, reboot some more, jump up and down and cuss, complete power down is the only solution that we have found that resets the internal hubs, but once we do, things come right up and things work normally again. As far as reboot schedules, generally all our computers stay on continuously until a service update forces them to reboot, they are pretty stable outside of the static electricity issues hosing the usb hubs from time to time.
A mystery to me is that full shutdown is usually slower than hybrid. You'd expect the opposite since full shutdown isn't writing to the hiberfil. Also it should be noted that these features only work if you've done a UEFI install on GPT disk, many features aren't available in a BIOS/MBR install.
The main reason that a full shutdown is slower is because Windows sends a close program message to every running application and waits for its resolution. This could involve writing log files or saving other state data as each individual program closes.
Possibly but a hybrid shutdown also closes active programs so that overhead should be much the same. Maybe it's just slower to close the kernal than it is to write it out to disk.
Here is a registry key to speed it up - registry disclaimer etc. HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control WaitToKillServiceTimeout Default is 5000 (5 secs) Low this is force shutdown quicker or do shutdown -f -s -t 0 No question asked forced now shutdown
It's about time that a guy who ACTUALLY KNOWS WHAT THE HELL HE IS TALKING ABOUT puts up a video like this - way too many LTT-types running around on TH-cam these days.
Hey! Don't dis Linus like that! He and his team generally do know what they are talking about, and have the money rolling in to actually do practical tests and break stuff so you don't have to/so the geek side of you can see the inside workings, which itself is educational...
@@Dranok1 As someone who has been in the field for the better part of 3 decades, no, LTT and his crew do not generally know what they're talking about - they mostly fumble around and hope for the best. They are little better than the "consumer helper" you'd find at your local Microcenter.
I have a somewhat old machine and it seems to benefit from frequent restarting. I've set up Task Scheduler to run `shutdown /g` once a day (at 5 a.m.) and `shutdown /r` once a week. Did I mess up anywhere?
Well... you wouldn't want it to reboot at 5AM if you use your computer overnight (leave programs open with unsaved files, or it gets accessed over the network etc.). If you save and close everything in the evening and it's purely a client machine, then why not just shut it down afterwards, and turn it back on in the morning?
Best video to point to when people ask me.."why yo laptop restart so slow!" different expectations, you expect it to be fast..I always expect restart to actually boot cleanly
Working in IT I found a lot of even younger folk will still practice shutting down and starting up to restart. So I've had to include a small simple explanation of why they should restart instead for years. The tiny bit more people know means less hair-pulling support requests for me.
There's the possibility to disable fast boot In the "Choose what the power buttons do" menu of the Power Options. That way a shutdown becomes an actual full shutdown.
Go to Control Panel\All Control Panel Items\Power Options\System Settings and untick "Turn on fast startup.." That means no more fake shutdown, slightly slower boot time.
I find very little difference, so might as well do the full boot
Actually, there's a CMD console command that used to do all that. For instance, to turn off hibernate completely, you would use the command "powercfg -h off", and you would use "on" to re-enable it... Of course, I'm not 100% percent sure if that's something that's still available in windows or not in windows 10 and 11, aside from the fact that it APPEARS to work on my windows 10 system
Powercfg /h off
Pro-tip. if you dual boot, or plan to image the hard drive, **always** do this. Otherwise the filesystem will likely not be in a clean state.
Dave I have a suggestion: A complete, step by step cold boot of a modern day PC - from PSU power on to login screen. Mobo, intelME, memory check, TPM verification, BIOS, bootloaders, kernel... To really show us what really goes on while we wait for PC startup. I saw an article about Linux booting and it was fascinating to learn how top-level user accessible OS actually is and what is REALLY happening underneath it - and before it can happen.
Seconded on that. PCs have gotten wildly complex now. One of the things that has always frustrated me is the way that graphics on UNIX is such a s***show so it becomes actually quite difficult to do something like a perfectly clean boot with a simple logo all the way from BIOS to finished boot without the graphics flicking in an out. For instance on NetBSD version 7 they brought in DRM graphics from Linux and the way they did it doesn't really play that nice with the existing framework, it just seems to take over so the system would start in text mode then half way through suddenly jump into hi res DRM mode. And you often see a lot of Linux distros jumping around between a boot logo and then crap being spat out from the kernel to the console etc, screen flickering as it switches modes etc. Seamless boot is quite tricky to pull off it seems.
Doesn't help that the freedesktop people aren't exactly the most transparent bunch of people and that PC graphics is still such a dark art. I once tried to fix a screen brightness support issue on NetBSD and ended up going through the kernel code for like ACPIVGA, DRM, wscons framework etc and it seemed such a god damn mess, it was like some Kafka nightmare. In the end I found an extremely dirty hack which just involved writing a PCI register that let me control brightness on my particular system but it turned out to be a VERY dangerous method. It worked on one laptop just fine but when I tried it on it on another laptop I couldn't get screen brightness working on it appeared to damage the LED PWM driver and the backlight would flicker badly for a few minutes every time you started it up.. god knows what I did but it did some actual damage!
didn't they teach you that in school?
@@kingjames4886 I don't think what I learned about PCs booting up 20 years ago in school has much relevance today.
@@treborrrrr it's not reeeally all that different
To the person pushing a button to turn on a magic box, no. To a person writing a startup driver to make your magic box work, yes.
It's not really a good reason to stop learning either way.
As a desktop support tech I hated Fast Startup. I'd have to explain to everyone why restart and shutdown were now different, and if I told someone to reboot I now have to specify restart and not shutdown.
Me too. Even brought it up to my fellow IT support engineers
A couple of enterprises I work with have all-SSD machines and they globally disable fast startup. The difference in boot time isn't worth the "What do you mean I don't reboot, I shut down every night" support cases.
I went out of my way to straight up disable it on every device in our environment.
Just be glad you never supported Windows 95.
It didn't take me long to notice that Windows 10 Shutdown was actually hibernating: 1: I had 64 GB RAM. 2: I used DriveSnapshot to backup my system, and it would not backup a hibernated system. Disabling Fast Startup in Power Options is easy enough, but I decided that I didn't want to keep blasting my SSDs with massive writes, so I simply disabled hibernation, and would just use Sleep. If I had a power loss, I wouldn't lose much, as I would make sure any current work was saved before Sleeping. If the power loss corrupted my system, I could just restore my system image if Windows was unable to repair the damage. Why Microsoft didn't simply add a Fast Shutdown option and keep Shutdown to mean what EVERYONE had got used to was just madness!
Fast Start in Windows is something I always disable, in today's age where SSD's are pretty much the standard, the bluescreens it often causes is more a bug than a feature.
That also lets you enable the hibernation option, which is occasionally useful without automatically using it every time.
Same here, I've always found it boots FASTER on an SSD with Fast Start disabled, it was designed to speed up loading from HDD and is detrimental on SSDs.
Raid arrays boot terribly slow. Fast boot doesn't matter to me because saving seconds on a 30 plus second boot isn't too impactful but I still leave fast boot on. I don't experience bluescreens or errors so there's no problems for me. So people reading this should probably opt to try it before disabling it. I use a lot of systems the newest being a 12900k/samsung 980pro Windows 11 based system and the oldest being a dual core from 2007 running 8.1 that runs media center, and machines in between 2007-2022
@@alexatkin And Fast Start Boot may cause problems when booting into another OS (if you use Windows and Linux on the same machine). As Windows with Fast Boot enabled stores the state of the NTFS partitions in the hybernation file and restores it on the next start, it may interfere with changes to that partitions from Linux (or any other OS). The worst case would be data loss.
I use ultra fast boot and i have never had a blue screen and im on the insider build. So i think thats more a fallacy.
I always find your videos informative about subtle things "under the hood" with Windows that I never knew before. Oh, you're of course entertaining as well and have an excellent delivery on the info. Great channel.
Glad you enjoyed it!
Love your channel. I was a pc consultant from 83 to 2004 then retired. Trying to get back into the field now with studying for my A+. the background you provide is priceless. Thank you.
In the control panel under System and security->Power Options->Change what the power buttons do; there is a checkbox for turn on fast startup that is checked by default. Unchecking it makes shutdown a proper shutdown again.
Rofstyx4 - in my PC, W11Pro, the checkbox you refer to is greyed out and connot be unchecked.
@@davybloggs1564 Near the top of that window there should be a UAC icon next to the sentence "Change settings that are currently unavailable" clicking on that should unlock the checkbox, but you may need administrator access for it to work. I am using win10.
@@RofStyx4 Thanks
Can't seem to find it on my Win10.
Changes what the power buttons do; there is a checkbox for turn ON
There is a difference in meaning between ON OFF and on off
that people do not realize and are not taught by the corrupt education system
Neither do the people actually notice - even when it is in plain sight - as the people
'have been turned into ZOMBIES
ON OFF ( in the ALL CAPS ) relates to switching - -
such as Turn the Light ON = or the Computer refuses to turn OFF
{ IT is also marked that way on switches - - so pay attention to detail )
on off is used in a different context:
Such as - I start work on Monday and I am off on the week-end
So learn it and reach it to others who are ignorant of this difference
And remember - pay attention to detail - Male sure the brain sees exactly
what the eyes are looking at
Such as HOLLYWOOD and not Hollywood
It is written on a Huge sign yet ZOMBIES cannot see it.
ZOMBIES have eyes but cannot see - ears and cannot hear
Ironically by the time they started making these changes, SSDs were almost already ubiquitous, making the changes not only meaningless, but decreased the user's understanding of their system at the same time. Good job MS.
Nonono...they know this, but they still wanted faster boots! The goals were (as I recall) initially to get boot times down under 30 seconds, and then down under 10 seconds. So, full BIOS POST was ditched via "fast-boot" to reduce that initial 10-15 seconds of hardware verification by BIOS. Some of the shenanigans that were being pulled in this race to the bottom were ridiculous. And the canned response any time something Went Wrong was, just do a full boot. It was enough to make those of us in OS Test just want to pull out what little hair we had left.
@@Ughmahedhurtz Good thing the memory training times of DDR5 completely undid most of that effort spent on reducing boot times, haha. Well, on AM5 systems, anyway. The situation with Intel systems may be different, idk.
@@aaronjenkins9573 Memory (re)training is only one part of this, though. Granted, if it's taking stupid amounts of time on some systems, that's a problem. Funny thing is, my current systems where I run dual NVME SSDs in RAID-0 take much, MUCH less time to load the OS from scratch than the BIOS does to boot/detect/handoff. I haven't seen any of this retraining time BS on DDR4. Hopefully by the time I upgrade my 3900X to a 8950X, they'll have all that figured out.
Reading data from the SSD isn't the only thing that happens on boot. Your computer still has to process a lot of things before it can get into a usable state. And depending on how many things you want to have started up, this can still create a significant boot time. By packing away things that don't really need to be restarted from scratch and only need their state saved, you can still shave off a lot of time that way. You just load it back into memory and hit go.
@@Ughmahedhurtz which is very ironic when you have a 1.5GB/s M2 SSD capable of booting the system from cold boot in 5sec, which is faster than all that complex code running and doing a lot of shite .
"Fast-Boot", more like complex-boot.
I basically always disable it, with hibernation too, and I disable all the power saving extra states that only serve to throttle the throughput of the CPU when you need it.
You know when you have a car and hit the pedal to the bottom, and it gasps, that's what I feel when I use stupid power saving states and speed-step or whatever Intel call that misfeature. I fucking don't want to save power, I want to have faster computer, if it uses 800W I literally don't care, I paid for it, and I want to use 100% of it all the time forever, until the planet becomes a death ball of global warming effects or until the heat death of the universe.
To be fair, the system only uses 70W when doing nothing, which is very acceptable, its like leaving a incandescent lamp bulb on. It uses less power than my A/C unit which is always on and uses much more power than that.
Another great video. Thank you Dave.
I got to 162 days on my MacBook several years ago. I bragged on some forum where a Windows guy (a friend) was complaining about having to reboot every week on average. Later that day, karma hit me. Xcode suddenly refused to compile my project. I tried all sorts. Reverted changes. Shut down Xcode etc etc. The errors made no sense whatsoever. Until I found a post online about Xcode having a memory leak. The humanity of it! I had to reboot.
Almost on topic - I once took a call from an elderly member of the public who started with "I believe I have information which will lead to the arrest of theee international hacker" (if you're trying to imagine the accent, think of Tucker out of something about Mary). Temptation was to hangup on the premise it must be a colleague winding me up. After a short chat, it became obvious his computer was "waking up" every time his phone line rang, which his modem was connected to.
I have seen FreeBSD boxes going without reboot for 7+ years...
@@MrIlovesubaru A good system administrator can use Linux or BSD for years without reboot.
@@anon_y_mousse
My most recent reboots have been due to hardware failure or moving the computers without UPS.
@@cigmorfil4101 I've got UPS's on all of my equipment too. Definitely a good idea.
If Xcode has a memory leak closing it and reopening should fix it unless there's something I'm missing. Worst case logging off and logging back in would probably help if there are background components you can't normally close. If rebooting was the only fix for a memory leak it _must_ be an OS fault.
In Windows, _Shift + Shutdown_ (or _Shift + Ok_ in the _Alt + F4_ dialogue, with “Shutdown” selected in the select-box) overrides Fast-boot. The OS will shutdown with Fast-boot _temporarily_ disabled. _Shift + Restart_ restarts into a boot-options (recovery) menu.
MOrnetc - In my W11 Pro PC, pressing alt + F4 just closes my browser! Please be more specific in your explanations re key presses.
@@davybloggs1564 - I have not upgraded to W11. I rely on desktop Linux as my primary OS. I keep a W10 partition only as a test platform.
btw- _MorningStar_ is what I intended with my YT_ID.
@@davybloggs1564 - Actually, I am quite sleep deprived. Alt+F4 will close any _application_ (i.e. active window) in Windows 10, also.
I do not have W10 loaded at the moment, but I believe that selecting the _desktop_ (i.e. left-click on desktop wallpaper) will place focus on the operating system for Alt+F4; else, closing all windows and then pressing Alt+F4.
Not certain --- I always close all apps _before_ I attempt to shutdown/restart.
I prefer Win+X to get the power context menu, then getting the shutdown/restart options from there, a bit cleaner than getting focus on the desktop then Alt+F4ing. I suppose you could Win+D to get the desktop first but the Win+X menu seems more reliable to me.
@@ChristopherWoods - I prefer to close all open apps, one-by-one, first. Alt+F4 works well for that.
Dave has reached 250k subscribers. Woot-woot! 🎉
When he gets to 640k then he wins. 640k should be enough for anyone!
@@amcadam26 Done.
Came into this video thinking I already knew all the things end users should know about shutdown/restart/sleep/hibernation... But this video still taught me the new knowledge that hibernations, which I use daily on my laptop, still consume a tiny bit of power to facilitate WOLs. Since I have no utility for WOL, it's now disabled on all my network adapters. Thanks Dave!
No mention of modern standby (in newer laptops) which breaks all the normal power management settings. What a pain it was trying to figure out why my system would go to sleep even if I had power management set to never sleep.
Dave. What can I say? I hate watching TH-cam videos that are more than 5 minutes long. Your videos are an exception. I listen to them more than once. A veritable wellspring of sagacity and practical knowledge. Well done. I gotta get back to my second listening. I like to vary the playback speeds as well. Both faster and slower than the default.
Yes. More than once to get all the content transfer to my memory.
Thanks, Dave. As a true geek who hales all the way back to the preDOS days, I truly enjoy your videos… I invariably learn something from each one and love the ‘inside baseball’ commentary.
watch out everyone, a "true geek" has graced us with their presence lol
Always a fan of your videos... The bit of history always brings me back to some period of tinkering around... This time adding hibernate to my win98 desktop PC back in the day and feeling cool. 😎
I think you mean Win2k. I might be wrong.
I KNEW IT!!!!!!!!!!! Thank you!!!
I thought there was something odd about Rebooting. When we do it to help an end user, some programs are sometimes still running like Excel or Word.
Being good at troubleshooting and BS I tell people I think it may only be going to hibernate before starting again.
I just now shared this video with my coworkers to confirm what I have been telling them the past year.
I learned from my experienced computer tutors in the eighties, the term ‚bootstrap‘ was related to a short piece of punched paper tape, about the length of a bootstrap, that was inserted into the reader to start up a computer and enable it to start the programs fed in from the upcoming spools of punched tape / cards or similar storage devices. So ‚boot‘ the machine was likely an abbreviated way of saying ‚let the run through the machine‘.
That story may be a “back etymology,” since not all computers in the “old days” used paper tape. The Digital PDP-5 used papertape but the “boot” was entered using the panel switches (IIRC, 7 12-bit words), which then allowed a full papertape program to be loaded. Most IBM computers before the 360 had a separate LOAD button on the card reader, and/or each magnetic tape drive. The 360 allowed booting from any device, usually an installed operating system, by dialing the device address into a set of switches and pressing the LOAD button next to them (but IBM always called it Initial Program Load, or IPL, from the 360 onward).
So I would go with the “pull yourself up” theory, especially since an operating system had to open files before the full file open/close software was loaded, read JCL to set up the JCL reader and initiator tasks, etc, requiring partial function duplication in different programs.
In a comp sci class in the 70s I was told it was from "pulling yourself up by your bootstraps". There were also some old cartoons that played on this trope of being able to pull yourself off the ground by grabbing your shoe laces. It's an old joke.
"Bootstrapping" is the term for a self-starting process in multiple fields, it's the self-starting part that is the common root, not anything computer-related. That comes from the pulling yourself up by your own bootstraps saying. (Which itself is derived from other earlier sayings.)
Back in the day, I had spliced together a dual-boot Windows 98 and Loop-Linux machine. The two sides could comfortably work on the FAT32 partition. I had to compile the source for the Linux network driver or the Linksys card I had in this unit. It worked great-- except when it didn't. At first, it seemed very random. Sometimes when I booted into Linux, the network card just wouldn't work. After a day or so, I realized-- when I cold-booted to the Linux side, network card always worked. But when I rebooted to Linux, after running Windows, network card wouldn't work. Back then machines turned all the way off, or you had to turn them off physically with a switch or button after "shutting down". Then I realized what was happening: the NIC wasn't completely refreshing after being initialized by Windows. It was remembering whatever Windows had told it between reboots-- i.e. It's internal memory wasn't being cleared. This prevented Linux from communicating with it properly. Oddly, the reverse was not true-- I could reboot from Linux to Windows and it would work fine. So, either Linux was being nice and clearing the NIC's memory when going down on a warm boot, or Windows was being a jerk and not clearing the NIC's memory on warm boot, or the opposite: Windows may have been smart enough to clear the NIC's memory when starting up on warm boot, and Linux was not. Hard to say. Anyway-- it is true, memory may not be cleared between shutdown and startup on a warm boot. Today, thankfully, things are generally simpler. However, there is that: As long as power is supplied to RAM of any sort, it only clears it's contents if those contents are explicitly over-ridden. The only other way to clear content's of RAM is to completely remove power.
I usually turn off fast startup for two reasons:
1. This does not work correctly on some of my (slightly outdated) hardware.
2. As far as I understand, this does not record information that the NTFS partition was correctly unmounted. Which causes problems in other OSes. They assume that the filesystem has not been correctly unmounted, it may contain some errors, and, as a precaution, do not allow write-mounts.
Regarding point 2, I believe it _isn't_ properly unmounted even though it's fully consistent. The FS driver in Windows probably has some state stored in "RAM" and if the actual state is changed under it bad things could happen. Having said that it _is_ very annoying.
I always use GP to disable fast boot across client networks, especially helpful when you ask a user to restart and they shutdown and turn on again.
Reminds me something about the 1982 Sinclair Spectrum design: its CPU didn't utilize the Z80's NMI vector so asserting the NMI pin on CPU just caused a hard reset. So the trick was to wire a press button to the NMI and patch the ROM with code to do something really neat like run an executive user routine with options to save or load a snapshot 'hibernate' image to/from FDD or print the screen etc.
The hibernation file is a great resource for anyone who wants to know what you were doing at the time the PC last went into hibernation. If you were accessing encrypted data, it will all be there in its unencrypted state in the hibernation file, as will all the encryption keys, passwords etc that were active at the time the PC went into hibernation, and any documents you had open at the time.
false - unless you using windows wrong, things are encrypted whenever written to disk, no matter what. I guess it depends on your settings and windows version, though.
Even while I knew most of this, I still learned quite a good bit! Thanks Dave for the knowledge. May I ask your take on the CPU Management Engine features - which sort of steps into some of the key points discussed on this video,?
That's what I liked about it. I'm very technical, so I know what the different options are. But this video casually gets into where these options make the most sense, and what to expect (boot time, power consumption, hard drive grinding, etc) with each of the optons. Pretty thorough presentation.
My father was an experienced MS-DOS user and solved most issues with “format c:”, not simply a reboot. When he got old it became a problem because he started to apply his DOS skills to his Windows 7 and refused to relinguish his administrator user. Hence I would get a phone call in the middle of the night saying “My computer doesn’t work.” “What happened?” “I formatted the harddrive.” Thankfully, being an engineer he was happy to use a system other engineers would use as well, so I was allowed to install Linux on his computer.
interesting explanation thanks Dave. On a related note for years now I've recommended people configure their laptops to hibernate rather than sleep not just as a power saving method but also for stability reasons. its not uncommon for laptops to have issues waking up from sleep especially if its a new model this is doubly true for Linux users where there may be a driver fix for the issue on windows but not on Linux. Aside from the first few years when the feature was supported I have never seen a machine ever have an issue waking up from hibernation. Also if you are one of those now rare dual booters who haven't switched to WSL it can also be helpful if you want to quickly switch back and forth between operating systems. just remember on the Linux side the swap partition is used to save the state so you must have a swap file big enough for the content of both your system and GPU ram and you should probably reduce your "vm_swappiness" possibly to 0 so you don't burn out your SSD.
Disable 'fast startup' in settings. Might have to ungrey it by selecting "Choose what power buttons do" and "Change settings that are currently unavailable."
Presto. Shutdown is shutdown.
If paranoid, additionally disable 'hibernation.'
In most BIOS, you can prohibit fast startups.
Great video 👍
Kindest regards, neighbours and friends.
I've found that hybernation(only) has been far more reliable, i've noticed that sleep, and hybrid sleep often result in network and USB not always working properly after wake, its rare, but when you manage over 1000 different machines, you tend to notice a higher ticket volume for wifi and keyboards not working when you enabled sleep because you wanted to save power/heat across all campuses.
Exactly what we found. When I replaced everyone's machines a few years back we got some back that had uptime of months. And users do not know the difference between a shutdown and a restart. so even when you ask them to restart they don't and then you end up with a system with 90 days uptime and you can't convince them that they did the wrong thing. A quick disable of fast boot and hybrid sleep in group policy and we reduced our random error tickets by about 75%
@@adamgeorge8327 I find an alarming number of users will restart the computer by pressing the power button on the monitor, even in out computer science department
@@denvera1g1 With my first PC, the IBM PS/1 (model 2??? - the question marks being numbers I don't remember right now, but it was a 286), that was actually the proper way ;) It only had a power button in monitor and the monitor VGA and power cables to the PC itself. The PC had no power button :) It would have prevented anyone without some tech savvy from replacing the monitor with some generic VGA monitor and have it working... There would've been no way to boot up the system. Though the power connector on the PC was probably just a standard 3-prong connector, so it shouldn't have been that hard to get it powered up anyway.
Just thought I'd mention.
@@denvera1g1 There are times where you have to use the power button (though you hold it rather than just pushing it). I always run an actual restart after that though.
@@karlrovey I still have the power button connected electrically, but group policy disables the windows power button/lid close action. However the BIOS will still normally cut power from the device after 4-15 seconds depending on the model. We have a group of laptops for a VR class that shutdown after 15 seconds of holding down the power button, i've never seen a machine take longer than 5 seconds before then, and if i didnt have more than one laptop i'd say the power button was defective and never would have held it for longer.
The weird thing is that after 5 seconds of holding down the power button, Windows shows this "swipe up to shut down" option, which 1: i dont want to do because you're acting up and i just want to re-image you, and 2: i cant do because this isnt the touch screen version.
Turning off fast start-up is the first thing I do after setting up a new Windows machine. I'm more than happy to wait a few extra seconds every morning in order to ensure that patches etc. are all installed properly.
Great video! Really enjoyed the history of booting a system, really helps to visualize what is going on. I work in IT and ever since Fast Boot became a thing we have been disabling it for our end-users because when we ask them to do a restart they end up powering off thinking that it was even more helpful.
Microsoft must have caught on to the issue Fast Boot was causing, as now when a your system needs a reboot due to a Windows update, the power-off option adds a new "update and power off" selection.
Kind of funny, and quite predictable, that they made what was traditionally the safer option the one that performs worse.
@@anon_y_mousse While that is true, it’s also true that at the same time they did this, almost no one noticed, because they made it that much more reliable (although still not perfect).
@@babybirdhome Still not as reliable as a well maintained Linux system.
I had to turn off wake on lan because I couldnt send my computer to sleep: Every single time the computer would wake back up by itself shortly after. Sometimes literally 20 seconds after having gone to "sleep".
Everyone had told me that it was my fault, that the mouse moved or whatever, but even unplugging any entry devices would bring the same situation. It only got solved upon disabling WoL and I havent had any issues since.
I've been straightened out like a pretzel. Thanks Microsoft! You should have included some tips on disabling prefetch and superfetch, and these pseudo shutdown options in the registry. I've disabled prefetch for my Windows systems since they run on SSDs and having copies of files for a quick boot is a waste of space. Great channel, thanks Dave!
Hey Dave, Just wanted to say YOU ROCK!
For years, every time I need to shut down a PC, I'll use the "shutdown -s -t 0" command.
I was told that it bypasses Hybrid Sleep and stuff like that.
shutdown -r works well, if you need to reboot from an RDP session...
@@tekvax01 Throwing down a “shutdown -s” fresh out of muscle memory instead of “shutdown -r” was always an interesting situation on remote sessions!
@@WilliamHaisch You subconsciously just wanted to go for a drive.
You should've mentioned the (annoying) Modern Sleep that windows has on newer laptops. It's annoying because I don't want my surface to overheat in the bag and I can't enable regular sleep in any way. I just hibernate it way too often instead.
For a long time I had fast startup switched off because I was told it was better, but when I turned it back on the boot speed went from 22 secs to 15 secs, which is worth it for me.
Great episode! I'm telling all my programmer friends about your channel.
3 things:
1.) So, for those us who got it but need reinforcement, like, need to write it down:
Restart - clears everything and brings up a fully reset state. AND
Hibernate (set for "When lid is closed") is more desireable than power off because it does power off but on wake quickly returns to where I left off with minamal power consumption. Right?
2.) To optimize the longevity of my laptop battery, I guess leave it plugged in and have "close the lid" hibernate the laptop?
3.) Do you happen to know or giess why MS doesnt have a battery meter that reminds the user to plug-in the charger when it drops below, say, 20%. And another notification or bell to tell the user the batts has reached 100% charge (or should it be 85%?) and to unplugg the charger?
I like watching your channel. It brings back lots of good memories: The PDP-11/70 days were lots of fun, UNIX and C had there moments too.
Another great video, thanks. That shutdown/power-on no longer does a full system refresh by default is one of MS's worst decisions. Naive users dream that they are doing a system refresh and they are not. I ALWAYS disable Fast-Startup on my own system or others because I want to be able to choose Sleep/Hibernate/Shutdown myself. Naive users need shutdown to mean shutdown.
There are ZERO cases where it makes sense to have Fast-Startup enabled, either users are naive and need it to be a shutdown or they are not and they can choose Sleep/Hibernate/Shutdown themselves.
This feature caused no end of trouble with the hundreds of users that I recently did phone/remote support for.
Indeed, and the time it saves on boot (I can't tell tbh, can't imagine it being more then one or two seconds if that) on new pc's is thus negligible, like I'd guess my pc "slow" boots in five or six seconds or something, like what are we talking about, what does a second faster accomplish when it's already that fast to begin with? Not worth the trouble imo.
I always disable Fast Startup so it acts like old school shut down. SSD's have made cold booting so much faster, that it no longer seems necessary to have Fast Startup enabled with an SSD anyway.
The only REAL COLD boot is removing the power cord from the computer power supply.
I've had networking issues surviving multiple cold reboots and restart cycles because the NIC card remains powered on via the standby 5 volt supply, that only goes down if the plug is pulled!
I have lost count of how many times this has bitten me on-site! The NIC card holds the ARP lookup table, and DHCP gets confused and will NOT serve an IP reservation until you pull the plug!
Throwing the big red toggle switch in the back from 1 to 0 pretty much does the same thing. If you don't trust that, you could use a powerstrip with a switch on it too. Pulling the cord is not necessary, and has a slight chance of causing other issues, because of the loss of grounding, and possible sparking.
@@joesterling4299 It may surprise you but not every power supply has an off switch and and the one I have is black. Cheap chords have switches without dampening and bad design which can spark worse then pulling the cord.
Dave is the best technical teacher I've ever encountered. He knows his material perfectly, doesn't make the mistakes that many instructors make, and has an engaging and personable style. He's instantly likeable. Thank you, Dave, for doing this channel!
Thanks for the kind words!
This is actually why I always (on my own laptops) always turn-off this. Do a shutdown/power-off/restart cycle every 1-2 weeks. Too many apps still have memory leaks or leave external hw in an undetermined state.
My work computers always get shutdown at EOB and started from fresh SOB.
THIS x1024
Disabling "Fast Boot" also will cut off power from USB ports, unless the motherboard have a feature to power some of USB ports (many of today MBs do offer that feature) in order to charge phone/tablet while PC is off. If the "Fast Boot" is enabled (default) the USB ports are powered even if PC is off.
the irony is microsoft brought us fast boot exactly at a time where its no longer beneficial.
thanks to modern hyperspeed storage the differences are well... irrelevant in any practical way.
so lets introduce more complexity into errors to have single digit percent faster boot shall we
Not everyone has SSDs for storage, especially laptops where if you want a less expensive one to have 1TB, it will be mechanical. If you want SSD but less expensive laptop, you get 120GB
@@danman32 I hate fast startup with a passion, with mech drives my experience from clients is that you needed 8gb ram even at windows 8 times, otherwise fast startup would bog down your PC with them missing reboot, it would also expose driver issues more often and after that it would require a real reboot. One of the first things I check when someone calls me for issues is the task manager wake timer, if it's more than 15 days is most likely that pc needs a real reboot plus disable the fast startup
It's beneficial because of WoL.
@@MrCrashDavi It might not be because even though the kernel is saved to disk, it is still considered in a shutdown state. For WOL to work, it had to be in sleep or hibernate state.
It is not the same speed. Hibernation is indeed faster, as it only loads from disk to memory, without the need of processing. I've loved that fast boot feature, even though I have always hibernated since XP, because it was blindly fast back then compared to cold booting.
Always wondered why it takes my computer longer to reboot than if I power it down and turn it back on. Very informative this podcast.
This reminds me of the days of Win95 and NT4. There was no sound driver for NT4 and the Win95 drivers didn't work. But, when you first booted up Win95, did a soft reboot and then started NT4 you did have sound like normal.
Nice. I consider myself one of the ancients as well sir. Started one of the earliest ISPs here in MN, owned one of the earliest clones available. Did all the memory mgmt of card installations in the old dos days. Saw the first browser come in with CERN, a DoD site, and one other (I think a college site) and that was it. I think people don't get the value of understanding how things work at a low level in relation to today's commonly accepted complexity. I learned a bit on this one and appreciate the vid :) I wish you the best.
Thanks Dave! Superb update for those of us who “used to know” the details of the boot cycle. 😉
Fast boot is something I've turned off instatly in BIOS on all new machines since its first introduction, but cool to konw more "behind the scenes" things as usual in these videos!
On Linux you can just set the WoL flag on your NIC at any time; I usually do it in a startup script. Interesting to hear how Windows handles this automatically depending on the sleep mode.
Not sure if mentioning Linux on this channel is heresy though... 🤔
Linux is not forbidden here. Amiga rules. Windows is just for gaming ;)
As a Puppy Slacko user I now recall my Windows days as like holding a very valuable thin crystal wine-glass that I could easily crush in my puny fist and therefore a constant worry that It would break. Using Slacko is like having a shipping container of restaurant-grade tumblers at my disposal.
Only maybe 10 days ago I set this up on my Linux desktop. It was a 2 step process, enabling in the BIOS and using the set option in Ethtool to enable, that apparently can get overwritten on restart so also make it an option in Network Management to respect the 'magic packet'
@@airjuri 'Tell Max , Amiga rules'
it is still in the adapter options in device manager. And BIOS also often needs to be setup for it.
This channel is often informative, but apart from the clickbait title "You're Doing it Wrong", I won't be changing my shutdown/restart habits after watching this video.
Thanks Dave, your videos always teach me something while being emotionally calming. All the best, keep the content coming. :)
There's also usually a poorly-named option in the BIOS somewhere under the Advanced Power Management to say "when you're off, turn all the way off." Otherwise, the RGB lights keep going and dumb crap like that. The EU required motherboards to be able to actually turn all the way off.
Hey Dave,
Great video as always. I was surprised you didn't cover the practice of holding the left shift key while click shutdown in Windows 10. I've found that to be quite useful for getting a fresh boot to run utilities or a live disk from a flash drive. Any thoughts or corrections?
I'm pretty confident it flushes the disk caches, but not much more!
My routine after reinstalling windows is to disable the hibernate function using Command Prompt in admin mode with "powercfg -h off" (I never use it as I had issues in the early days of it and disk space was more important than hibernating back in the day), then I disable "fast start up", disable screensavers and set it to turn monitors off after 15mins of no use.
Then I'm very much like Dave, where my PC will stay on and will only get a restart every once and while.
A few years ago, I hibernated my windows 10 laptop and popped it into my rucksack to take home. On arriving home I could smell a very hot electrical smell. During my bike ride home windows had decided to power back up and attempt some update and due to the vents being blocked became dangerously hot. The end result was the battery pack being ruined and I no longer using Windows 10 on my laptop again.
Blaming Windows for your own fault xD
What was his fault?
@@thedeegee1601OK we'll blame you if you'd locked your car and then at 3 in the morning your car decides to start itself up and burn all your fuel
I had installed windows on my laptop for about a year. And whole this year I was trying to fight with it's power ups. It does not matter if it was hibernated, or in the sleep mode, it always was turning on and checking for [censored] windows updates, mail/messengers notifications, and etc. I tried all the ways to disable this behavior I could find on the internet so far (turning off wake on lan, wake on usb device, etc.), but nothing could help me to resolve this start ups when the laptop is closed.
It was so stupid that I close my laptop in the evening, and when I wake up it's completely uncharged, the battery is 0 percent.
The only fix I could find is buying Mac. Now I can close my laptop and be sure if I open it next time it has the same battery level as before.
@@mekedron THose wakeups are really annoying. I stopped that by running Taskschd.msc as TrusterInstaller, disabling UpdateOrchestrator's wake task and denying write access for that task file for System. Now Windows can't reactivate that wake task after new updates are downloaded and ready for installation. :D
Dave,
Wake on Lan does work on PC's that are "fully off" (not just hybrid sleep) if enabled in the BIOS. Sometimes this is called "Power-on on PCI-E" or similar. We have over 250 PC's of various models at work (mostly Gigabyte and ASUS motherboards) and I can assure you every single one supports Ethernet Wake on LAN from a fully off state, and did so both on Windows 7 (which only had a true shutdown) and now on Windows 10 with or without quick shutdown enabled. (We have it disabled with group policy)
This is a feature of the BIOS and onboard ethernet hardware and nothing to do with Windows. (It works even if Windows is not installed or some other OS is) The Windows ethernet driver for some network adaptors offers the option to disable wake on lan - however this disabling of wake on lan in the windows driver only takes effect after windows boots up not when power is initially applied to the system, which is controlled by the BIOS setting. So wake on lan will still work once after power is first applied but will not work after a windows shutdown/sleep if disabled in the driver as it gets disabled when windows boots. So for it to always work it has to be enabled in both the BIOS and in the windows driver.
There are also two types of wake on lan - what you describe with the MAC address repeated 16 times is "wake on magic packet", this is OS independent (supported in the BIOS from both fully off and standby) and works on all systems that support wake on lan.
Then there is "Wake on pattern match" - where Windows or other OS can program additional custom pattern matches into the ethernet hardware at run time via the ethernet driver.
This is how waking on arp resolution etc works as Windows configures a pattern match in the ethernet hardware on the fly which matches arp packets that the PC would normally respond to based on it's IP address, then goes to sleep. When a matching ARP packet is received the PC wakes. As far as I know this type of Wake on lan doesn't work when the PC is fully off, only when it is in standby/hibernate/sleep.
Not all ethernet adaptors support this second wake method. If you don't see an option for it in the driver it probably doesn't. If it does have the option you can disable wake on pattern match while still leaving wake on magic packet enabled - which I've done on my PC at home because some other device on the network querying it using arp would keep waking it up, yet I still want to explicitly wake it using wake on magic packet.
I've also recently had to disable the second method on one of my machines as it was waking on spurious packets after a driver update. That said, wake on patten can work if off (well in ACPI S5) but it is highly hardware dependent the driver basically has to be able to save the patterns to wake on in the card itself.
I do mention that, and explain that it has to be turned on in the BIOS and then enabled for the device... I think!
“Pulling themselves up by the bootstraps” was always intended to be satirical - at the time in the Victorian era Malthusian philosophy was popular i.e. being poor was a choice and a moral failing. So, official policy on poverty was often shaped around this idea that a person could become a productive member of society on choice alone - if you just cleaned your room, took some responsibility and hustled then all your problems would go away, precisely as true then as it is now: misleading, naive and incomplete at best.
To illustrate how ridiculous it was that policy aiming to solve a structural problem was entirely focused on individuals, the phrase was picked because of how obviously absurd it is. It’s only been “rehabilitated” relatively recently, and tbh I think I would prefer it with its older meaning
On the flip side, not cleaning your room, not taking responsibility and not hustling will pretty much doom you to a life of poverty.
Ever heard of Baron Munchausen ?
To use mathematical terminology, hard work (or luck) is a necessary but not a sufficient condition for prosperity.
@@kcgunesq Shhh, don't tell the truth to a Ieftist, at best it goes over their head.
I enjoyed the entire video, but as a guy who has loaded in the bootstrap from front panel switches, I especially liked the explanation of the term "Boot".
What Dave didn't mention is the so called "Modern Standby" which Microsoft insists on for laptops since some time. This mode keeps your computer in S1 (fully on) state but is supposed to make your processes and devices aware the computer is supposed to be asleep and to behave themselves, only allowing for light activity like periodical e-mail checking, etc. It practice, the implementation is broken and rarely works as intended, often making your laptop cook itself to a near death in your bag or backpack, until the last resort thermal failsafes kick in (or not).
Yep. It's horribly implemented and ignores your power management settings. For example, if you close your lid but want the computer to 'do nothing', modern standby will still put your system to sleep after 15 minutes. Infuriating, but setting the BIOS/UEFI firmware sleep mode to 'Linux' restores classic, working, functionality.
I noticed this omission too, and agree that Modern Standby and Connected Standby as it was called in Windows 8 is a DISASTER. I have had the issue with the laptop overheating in the laptop bag several times, to the point that even the glue on the rubber strips on the bottom of my Dell XPS 15 melted and let loose. Fortunately, it can easily be disabled in the registry:
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Power
CSEnabled=0
I forgot what that was called and I never figured out how it worked before now. If it's keeping the computer fully on that's probably the stupidest thing ever. I'll need to make sure I have that fully disabled on my WIn10 laptop.
On my main system, which pretty much runs everything on the property including surveillance and media server, I have it set to do a restart at 5:30 AM every day. This might be excessive but it always gives me a fresh load to start the day with and that's worked out well for me. It's worth the 15 minutes it takes to restart and fully load everything from scratch with fast boot/fast start disabled and a full fresh load of PCI and USB in the BIOS.
This also incidentally reveals any impending hardware failures that might manifest on a power recycle. Better to find them in the morning when I can do something about them than have them randomly fail at unexpected times.
When I do shut the machine down all the way for any reason I have the surge suppressor that serves the machine mounted where it's easy to get to. I just flip the switch on the surge suppressor which removes power completely from the machine, wait at least 30 seconds for the caps to drain and then reapply power to the system.
I still have machines with Win7 that support WOL because WOL is a feature of the LAN card and lately the BIOS and not of the OS.
A LAN card must have 5VDC always for it to work. And the IP is not needed in reality since the magic packet is broadcasted to the entire network, not just one or any IP. You can anyways use the IP to ping the OS when it's up and running :)
Great content and channel, Dave!
Just remember to park your hard drive before you shutdown
Or wait for the "It's now safe to turn off your computer" screen if you have more recent OS.
On the first PC I had, a 486 DX-66 I actually used to enjoy watching the memory check as it literally counted the bytes of my 4MB ram. Imagine how exciting it was when I installed an additional 4MB and maxed out the 8 memory slots it had.
Bit flips(solar radiation flipping ones and zeros) happens more frequently than people realize. It's a good practice to reboot once a week imo.
Once every 2 days was the statistic found by CERN (although different papers seem to disagree a lot). A single bit out of the billions in RAM every 2 days honestly doesn't seem like much to me, unless you never close any programs ever, and those programs aren't constantly modifying the state of their memory (fixing the issues). Not to mention that ECC RAM makes these errors even less likely to have any real-world impact since they can be corrected whenever they are found.
@@miguelguthridge Indeed, and I've only powered off any of my computers in the past decade because of power outages. Although I also don't use Windows, so that might make a difference.
@@miguelguthridge Well the overwhelming majority of PC's doesn't run ECC, so that's not really that relevant for most users. Now it'd be interested in learning how DDR5 handles bit flips as it uses a form of ECC internally on the memory modules.
I've worked in the PC field far longer than I care to admit, and I've always tried to steer people towards ECC memory when they spec out workstations. If your personal gaming machine crashes because of a bit flip it's unlikely to cause any long lasting problems or major costs. But in a machine used for production or economics it can be disastrous.
Now if we look at the likelihood that a bit flip will cause a crash it's actually quite small. Of all the memory used by the system the absolute majority will contain graphical data, sound data and things like that. The actual code that's executable is way smaller than all the "fluff" data that memory is filled with. And not even all the executable code will actually be executed. There is a lot of code in conditional branches and error handling that will only be run very infrequently. If the bit flip occurred in code like that you might never know. And if it's in graphics or sound data chances are you wouldn't notice.
Even a crash isn't what you are really afraid of. A crash will self correct the problem as it makes you aware that something is wrong. When the bit flip occurs in critical data used for calculations or in data that is the result of calculations that hasn't been written to disk yet, that's when things get scary. Unless the data has built in parity or there is some kind of sanity check on the calculations or perhaps redundant computers performing the calculations then a bit flip can go by unchallenged and the corrupted data gets logged as being good.
So bit flips happens, and no, that your computer doesn't constantly crash isn't prof that it's not happening to you. And as memory size increase as does data density there are more cells and they are smaller. That means the chance that a bit flips goes up both because there are more of them, but also because the smaller the cells get the easier they are to flip. The use of ECC in the DDR5 memory modules is one sign of things getting critical, but I wish they had chosen to standardize on full ECC capability for every thing memory related. Then we would be able to see when there is a correction done and have a chance to measure the frequency with which they occur. But then I guess they don't want end users to worry about these things. Out of sight, out of mind...
Most people nowadays don't reboot for months (especially in tablets), and all is well and fine. Bit flips are not that common
@@mikemx55 I'd argue that 90% or more of the memory is used for things other than executable code, and most of it is non vital data like graphics assets, sounds and so on. You would be hard pressed to even notice a bitflip in all this data, and so you can be sure that the absolute majority of them are never discovered.
Finally, Dave talks about the film industry.
Thanks, Dave. Very interesting. I've always wondered about this stuff.
It took 10 years for laptop bios to properly restore the computer after a suspension (not hibernation). The computer powered up, but the ethernet and WiFi wasn't connecting, external USB disks were lost, Bluetooth was disconnected (A2DP devices especially), screen brightness was messed up, and more.
Perhaps you can shed a light on why it took so long for laptop manufacturers to mend an apparently straightforward operation.
Thank you for the trip down the memory lane...
P.S.: the best moments of my long IT experience are those of the launch of Windows XP (after being obliterated by the Trumpet Winsocks).
My worst IT moments were at the time of the 486/66 and Local Bus computers.
Thank you once more.
Regards...
Anthony - from the UK.
Most of the issues you are describing will mostly be OS (and to a lesser extend device-based) as AFAIK the main BIOS is only responsible for keeping RAM intact and re-entering the OS. Actually ACPI code (which is stored in the BIOS) is likely responsible for much of your problems as it can potentially tell hardware to cut power to certain components. In any case I don't remember having many problems like you describe so maybe you were just unlucky with your machine models.
And of course some things like Bluetooth devices being disconnected are going to be because it's physically impossible for a device to stay connected to an unpowered host. For USB disks being lost, it could also be drives which aren't able to reconnect in a reasonable amount of time. I've encountered that with some devices but usually flash drives work just fine.
I always wondered about this. Thank you Dave!
What would be better is if it built a 'blank hibernation' state where it boots the system, and BEFORE giving the user control but AFTER loading ALL services (including any post login services) it creates an image of the blank system. That would allow it to accelerate the boot AND fix issues that have built up over use. This would not replace standard hibernation, but be more for boot from power off. Save the boot bit by bit for recovery modes and such.
I knew this, but then I foget it.
Now I know it again.
Very well made video. To be honest, I didn't learn anything new, but for users it is explained perfectly well and in understandable language.
The most interesting part for me was your bio. I'm kind of an old dude and I've been using probably all or almost all of the OS you have worked on in the past -starting with MS-DOS.
Interesting channel, so I've just subscribed.
Hibernation has been one of the best inventions ever in PC world. I've enjoyed it since XP times, when I was using a helluva slow Dell desktop, which took forever to boot up properly vs. a minute or two coming out of hibernation. so for me, fast startup was a welcome addition in later Windows versions. as a bonus, I always know what to tell folks who say "why is my Task Manager showing 5 days of uptime when I shut the PC down every night?"
Woahhhhh. For so long I've been meaning to look up why it takes noticeably longer for the computer to shut off when I picked Restart than when I picked Shut Down. Finally this great mystery is solved!
I always disable fast startup as I had a mystery for awhile where users would report bad performance and slow applications on their PC. They would shutdown their PCs nightly, so always appeared to be a program issue/hardware or the moon in the wrong phase. Wasn't until we started deploying RMM agents and started seeing PCs with an uptime of weeks when users said they were powering off every night did research discover that shutting down isn't actually shutting down.
What was considered by software engineers as a good feature to me is just giving a user more rope to hang themselves with.
I firmly believe in the phrase, "Make something idiot proof, the world will build a better idiot."
Love your stories Dave. I’ve been interested in computers & software since I was 14 and had access to the TRS-80 back in 1981. At first, I learned simple programming with an obsolete punch card system. It was a simple calendar generator we programmed with the only variables being the month & year and it would print out a one month calendar. Everyone did the current M/Y or their birth M/Y, and I did January 2000, thinking about how far away that seemed(little did I know about the havoc THAT date would create, eh?) Then I learned to program BASIC. I was introduced into manipulating the hardware with the common single to double side 5 1/4” diskette conversion via a hole punch. Next I learned to solder and there began my love affair with the computer & gaming world. You had my sub long ago and you get my like with every video I watch - keep up the good work!
obsolete punch card system? You implying they don't make 029 keypunch machines anymore? That means no chads for the ticker tape parade. (where did that memory come from? Chads?) More humor about the computer having a crank on the side. 1974 BSCS
@darrenlesage2420
when in doubt you can always flip the button on your power supply (desktop assumed)
easily some of the best content around. thanks for the education and laughs!
I always turn off "Fast Startup". With SSD's that feature wasn't even needed anymore. As a gamer, stuff can get wonky, restart is in order.
Would love to see your take on the hot mess that is Modern Standby (S0) and the removal of other standby options.
I've not found modern standby to be all that good and have had laptops wake up while in a bag resulting in overheating. To make it worse on some (maybe all) laptops that have S0 enabled, all other sleep options are not available.
I have had this issue too and it freaks me out, making me think it's going to catch on fire while **I** am sleeping, catching my house on fire (it's a very high wattage computer).
Fascinating.
Now if only I could figure out why my desktop randomly wakes from sleep in the middle of the night sometimes.
Back in the 90's, me and my tech friends would refer to sleep mode as "coma mode" because upon awake, sometimes it ran slower or had trouble loading things into RAM. ... or maybe I just had crap hardware at the time.
That one watt of power matters a lot more when I place the laptop into a backpack immediately upon shutdown. I can't tell you the number of times I've opened my backpack to pull the thing out only to find a sauna inside the backpack and the fans running full blast.
It's due to this reason that I use Hibernate so much.
Amazes me that after 40 some years of making laptops the industry still hasn't sorted that one out. Same thing's happened to me even with non-HD notebooks.
The problem with fast boot is that sometimes it doesn't work. It just won't allow you to shut down your computer. On the internet they say that it is only the problem with old motherboards but I actually had this problem on a 2020 motherboard with the latest UEFI.
Just pull the cord out the psu like a man
Ha Ha
Not much use on a laptop.
And considering you’re dealing with windows, which needs to be able to juggle a gazillion different hardware configurations, not necessarily a smart thing to do.
Back in the DOS days no problem, and sometimes even necessary.
The number of times in the 90s I had to physically attend to a workstation because the user would not accept that turning the monitor off and on was not actually power cycling the system
Interesting subject. Some time back at a Microsoft partner seminar (Win 8 time) one of the most senior Microsoft chaps answered questions saying the software had been designed to "clean up" when the MS Operating System when it closes down fully. Hence delays in start up from a complete full shut down (electric shut off) should be used regularly to ensure BIOS and device parts with BIOS type instructions can be refreshed. It was also why so many repairers and advice was to shut the system of, turning off power, to overcome possible corruption occurring when the system 'saved' in quick start methods and the device parts had actually been the cause source and the total power off restart had proved to best and so often got rid of the some issues which otherwise returned from a "hot boot". However, you do seem to support regular power off on the system as recommended by Microsoft. From experience in attempting to diagnose between hardware and operating system my vote is for power off if a hot boot does not resolve an issue or the type of issue recurs. Often the quickest resolution although I do know a few so called techs that do repairs who often just give up in seconds and claim that hard drive has given up or OS is corrupted. Then they proceed to make the quickest "repair" by charging for new hard drive installation or total wiping and re-installation of the OS which unnecessarily results in complete destruction of easily recoverable data. The old "Ghost in the hardware" however is a very interesting experience. Create a new system with all new parts and OS. Starts up and falls over just as quick. Software diagnosis reports all good. Test all hardware individually. All good. Hmm. As you are putting several systems together with exactly the same part requirements you construct all the others which work well. Take individual parts from the "Ghost machine" and put each in a different machine to each other. All systems work perfectly. ??? In the future you never hear from the users of those machines of them having any problem of failure. ??? So what caused the problem with that one machine which became just one of the number that finally went out the door? I just love computing. Some variables you can come across can be so exciting... NOT.
Dave, you might be upset to know that some companies restrict access to the task manager, requiring an admin password just to see what processes are locking up on your pc
That is not so much the fault of the companies as it is of Microsoft to allow admins to have the option of doing so. In my experience, most admins seem to take the attitude of "Who are YOU and why are you messing with MY computer system?" Out of laziness and arrogance they tend to lock down whatever they can just because they can. If you challenge it they will tell you that it is "for security" but cannot articulate how that is so.
@@WatchesTrainsAndRockets idk dude. Locking things down keeps stuff secure. Depends on how much risk you accept.
@@ravenbarsrepairs5594 Did that a year ago on my computer and am in the process of putting together a Linux system for my wife as I will not allow Windows 11 and all of its data gathering into my house.
@@WatchesTrainsAndRockets part of restricting access is to protect the worker. "Did you install that program that stole all our data?" "No sir, the admin locked my pc, so I can't install anything"
@@LeifNelandDk I NEVER said that restricting access was a bad thing. What I said is that things such as task manager need not and ought not be capable of restriction. If it is built so that it will not allow a user to affect a system process nor other users' processes, there is no security issue with allowing a user to run task manager in order to control their own processes.
so can we blame you for ME?
Hahahahahahahahahahahahaha, partly
At my first job in 1967 at Purdue University's High Energy Physics research group, we had an IBM 1401 card system. There was no way to boot an application from a disk as it was a card-based system . One of the programmers (Roger Wagner) stored an image of the boot deck on a hard drive (IBM 1311 in this case) and created a single-card boot program which read that image from disk, thus only the one card was required to start up. That card was called a "RABIT" card, which stood for "Really A Bitch In Theory"
NB: IIRC IBM didn't use the term IPL until System 360 in 1964
I am THAT guy. Common sense dictates that shutting the machine down is the mostest bestest. In our upside down inside out world, of course piddly restart works better. Massive tip. Thanks!
We use MS Win 10 Surface laptops in a production environment, rolling test stations with lots of USB devices attached. Occasionally, maybe once every few months in the summer, more frequently in the winter due to static electricity being worse, the USB hub internal to the laptop will get glitched and the only way to reset those internal hubs is to completely power down the laptop. You can reboot until the cows come home, run updates, reboot some more, jump up and down and cuss, complete power down is the only solution that we have found that resets the internal hubs, but once we do, things come right up and things work normally again. As far as reboot schedules, generally all our computers stay on continuously until a service update forces them to reboot, they are pretty stable outside of the static electricity issues hosing the usb hubs from time to time.
A mystery to me is that full shutdown is usually slower than hybrid. You'd expect the opposite since full shutdown isn't writing to the hiberfil. Also it should be noted that these features only work if you've done a UEFI install on GPT disk, many features aren't available in a BIOS/MBR install.
The main reason that a full shutdown is slower is because Windows sends a close program message to every running application and waits for its resolution. This could involve writing log files or saving other state data as each individual program closes.
Possibly but a hybrid shutdown also closes active programs so that overhead should be much the same. Maybe it's just slower to close the kernal than it is to write it out to disk.
Here is a registry key to speed it up - registry disclaimer etc.
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control
WaitToKillServiceTimeout
Default is 5000 (5 secs)
Low this is force shutdown quicker
or do
shutdown -f -s -t 0
No question asked forced now shutdown
If you turn off the “fast start up” and disable the hibernation file you can click shutdown and power it back up and it starts everything from scratch
It's about time that a guy who ACTUALLY KNOWS WHAT THE HELL HE IS TALKING ABOUT puts up a video like this - way too many LTT-types running around on TH-cam these days.
Hey! Don't dis Linus like that! He and his team generally do know what they are talking about, and have the money rolling in to actually do practical tests and break stuff so you don't have to/so the geek side of you can see the inside workings, which itself is educational...
@@Dranok1 As someone who has been in the field for the better part of 3 decades, no, LTT and his crew do not generally know what they're talking about - they mostly fumble around and hope for the best.
They are little better than the "consumer helper" you'd find at your local Microcenter.
2:02 that was good 🤣
I have a somewhat old machine and it seems to benefit from frequent restarting. I've set up Task Scheduler to run `shutdown /g` once a day (at 5 a.m.) and `shutdown /r` once a week. Did I mess up anywhere?
Well... you wouldn't want it to reboot at 5AM if you use your computer overnight (leave programs open with unsaved files, or it gets accessed over the network etc.). If you save and close everything in the evening and it's purely a client machine, then why not just shut it down afterwards, and turn it back on in the morning?
Best video to point to when people ask me.."why yo laptop restart so slow!" different expectations, you expect it to be fast..I always expect restart to actually boot cleanly
Working in IT I found a lot of even younger folk will still practice shutting down and starting up to restart. So I've had to include a small simple explanation of why they should restart instead for years. The tiny bit more people know means less hair-pulling support requests for me.
Dave: In Hibernating, the machine is completely off.
Also Dave: Wake on lan works when Hibernated.
I thought that I was paying attention, but I missed where Dave tells us how to perform a full reboot !!