What If You Pull Your CPU Out While The PC Is On?
Вставка
- Опубліковано 14 гру 2023
- Remove your personal information from the web at joindeleteme.com/techquickie and use code Techquickie for 20% off.
Thanks to Wendell of Level1Techs for his help with this episode: / level1techs
What happens if you remove your CPU, RAM, SSD, hard drive, or graphics card while your computer is still running?
Leave a reply with your requests for future episodes.
► GET MERCH: lttstore.com
► LTX 2023 TICKETS AVAILABLE NOW: lmg.gg/ltx23
► GET EXCLUSIVE CONTENT ON FLOATPLANE: lmg.gg/lttfloatplane
► SPONSORS, AFFILIATES, AND PARTNERS: lmg.gg/partners
FOLLOW US ELSEWHERE
---------------------------------------------------
Twitter: / linustech
Facebook: / linustech
Instagram: / linustech
TikTok: / linustech
Twitch: / linustech - Наука та технологія
Wait, a CPU socket is only rated for 10 insertions and an M.2 connector only for 60?! That's the actual thing I learned from this video. That's kinda shocking tbh.
Thats also new for me. Never even thought about that to be honest
No way that is true
Same, now I'm wondering if there's a reason they need to be so fragile, would be another interesting video
I need to see citations. Big wtf moment here.
PCIe and other power connectors are also limited.
USB: "Plug me in 10,000 times".
CPU: "About 10 times is enough. Thank you."
Dating vs Married
No one s speaking about hdmi? The shit broke in about 30/100 insertions.
Funny story about storage removal: in my early computer days I used a laptop with Ubuntu Linux, while not really knowing how to USE Linux. I tried formatting a USB stick, and somehow managed to FORMAT THE BOOT DRIVE. It actually completed the process of erasing itself and ran fine for 10 seconds after. I didn’t realize until I tried to open the usb stick and crashed.
That's like performing brain surgery on yourself and only dying when you see your brain on the table
@@Gigglesnix exactly still baffles me that Linux allows that
@@Gigglesnix 🤣
@@chielvandenberg8190 it's not that it allows that, it just happens to be possible because of how it's built
@@chiefdenis I know that it ran off ram for the last seconds but windows won’t allow you to erase the boot drive WHATEVER YOU DO
"When in doubt: Blue Screen" is actually really acurate to the mindset of a BSOD/Bugcheck.
It basically means that either some Kernel-Mode driver or the kernel itself has no damn clue what the crap is going on, and can't safely continue to function.
Hence why Linux sais "Kernel Panic: Not syncing". It's not syncing to the disk because something has gone catastrophically wrong (e.g. you are dereferencing a NULL Pointer in Kernel-Mode) and it would be unsafe to keep going.
Blue screen was how I learned my laptop ram died... Except the ram wasn't dead, the slot was.
I had a situation about a month ago where one of my RAM sticks was dying and the behaviour you described when "yanking" a RAM stick was very similar to what I was experiencing some times. At first I thought it was the MOBO going kaput but thankfully it was just a single stick. In all due honesty, nobody's going around yanking their components but faulty components can sometimes act like they're disconnected so knowing what it looks like when those things happen is quite useful.
That's how I figured out that my Chromebook's SSD had died.
Behaviour lol
Lol yanking
Relevant everytime I moved my optiplex I have to reseat all my ram and usually my GPU always the ram specifically slot 2 and 4
The most interesting useless tech video in UA-cam
In modern PCs, the CPU isn't who commands the system to turn on (at least not when fully off), it's actually a couple of chips on the motherboard that do: the "SuperIO" along with the "chipset".
In my experience, removing the OS drive in Windows doesn't blue screen immediately; instead appears to keep running but simply programs and basic OS UI elements begin to not respond; then the unresponsiveness gets worse over time that the computer becomes unusable. Mouse pointer never freezes, no BSoD, simply becomes so unresponsive that becomes unusable.
It can BSOD, had a boot drive totally die, and it took windows about 10 minutes to fully die. And I did get a BSOD, albeit a blank one
@@kuhljager2429 now that you mention it, only twice I did experienced a BSoD by bad OS drives, but suprisingly never by disconnected drives (or maybe after a while, not immediately).
@@lucario4483I pulled out the HDD on my latitude E6510 while running windows 7 pro and got a BSOD after a few minutes but I have the page file disabled on all my windows computers, I did it because I wanted to see what would happen and I didn't have anything important on the drive.
Since Windows 8, Windows actually attempts to survive a temporary main drive disconnect.
The kernel intentionally hangs practically all processes for a few seconds and sees if the drive comes back online. If the drive comes back, it keeps going as if nothing happened. Only after the drive reconnect window times out do you actually get the BSOD.
Presumably, this safeguard was implemented as part of Windows To Go. Windows To Go is a feature of Enterprise versions of Windows that assists users in installing Windows to external, removable storage. It requires fairly beefy, fast USB drives that can tank Windows' space requirements and constant disk activity. As a result of effectively providing official support for installing Windows to external storage, they had to make it reasonably resilient to the boot drive dropping off the bus.
Of course, this only works on the internal storage if the system BIOS/UEFI handles SATA/NVMe hotplug. Otherwise Windows is not informed that the drive actually disconnected and instead receives corrupt data, which either crashes or hangs, but the system basically can't recover then.
The OS kernels (both Windows and Linux) also support running with less than the total available RAM for debugging purposes. Then, removing a stick of RAM should be safe if it is completely unused by the software and firmware. And whether or not removing the GPU (or any PCIe devices) is safe relies mostly on the hardware and firmware not shitting their pants, and properly reporting to the OS. Assuming the motherboard supports PCIe hotplug, the drivers mostly handle it fine.
After all, eGPU drivers come from the same codebase as their PCIe siblings. Linux even supports securely ejecting PCIe hardware the same way it does for USB devices.
@@3lH4ck3rC0mf0r7 I guess that's why I've experienced some resilience in modern Windows: the drive wasn't actually disconnected, but having a faulty SATA cable/connection; or the drive was just really bad (that hangs loading due to physical bad sectors)
"When in doubt, BLUE SCREEN" made me laugh so hard xD
I hadn’t realised the insertion limit for motherboards but it makes sense. How do you get around it for benchmarking?
That's the neat part, you don't!
Carefully.
Ignore it.
Dongles could be used for power connectors. So the end of the dongle is what sees most of the wear. PCIe risers could help also. Then there's the old fashioned way, repair it, but that's just crazy.
It doesn't actually limit you, it's just that the manufacturer (AMD and Intel in this case) have only tested the socket to work for ~10 insertions. It may still work after 100, even 1000, it's just that they never actually cared about making the socket endure this many insertions. At worst, you might not get a warranty replacement if you did it that many times.
Not exactly the same, but related.
There’s a technique for reading RAM on a PC that is locked by quickly removing power, powering back up, and then booting into special recovery software. It relies on the RAM not completely losing data for something like a second or two (I forget the actual avg timing).
Because of this, there is a chance that you could pull the RAM and very quickly get it into a homebrew device to keep it powered until you get it somewhere safe
I remember the video of Linus doing PCIe hotswapping and the problems it entailed...
The only things I know of that really cannot be hotswapped are things like PS/2 keyboards/mice and M.2 drives. PS/2 hotswapping doesn't cause any problems other than the fact that you have to restart the computer.
In situations where PCIe, CPUs, RAM and SATA need to be swapped, daughter boards are used and are designed to prevent damage.
Even things like power supplies, in servers, can be hot swapped, but you require at least one PSU to be active at all times (which is why redundant PSUs exist).
My experience, mostly with computers that are 20+ years old at this point, is that PS/2 can actually handle being hot swapped and will work just fine after, though this is usually only at the BIOS. It's Windows that can't handle it correctly and requires a restart. On Linux, I've plugged a PS/2 keyboard while it was already booted and it just worked
@@redpheonix1000 I ended up with a system with a blown fuse once!
CPU will burn your finger lol 😂
Or it basically crashes your PC before you even get a chance to get the cooler off
Dawg my GPU fan doesn't work
@@Fitnessdickinmymouth Which GPU is it?
Like bro😂
Or you could just pull it out with the cooler. PGA master race.
Only rated for 10 insertions? Must be pretty expensive to test CPUs huh...
Sounds like my ex.
@@garrisonfjordNot mine. They could handle hundreds.
@@TheRealSkeletorlet me guess, needed a rotation of different cpus to function?
MrYeester has has a couple of videos on this. One where he removes a CPU while the PC is running. He also has one where he removes different components while the PC is on
It's a fun channel to watch.
Wouldn't it be great if it were standard for PCs without a CPU to boot to the point where you could update the BIOS / UEFI? This way you could, for example, update it to make the board compatible with the CPU you want to install.
Edit: So that even the Brainy Smurfs are getting it.
Some motherboards actually do have functionality to flash the bios/uefi even without a cpu installed.
Well the cpu executes the BIOS to begin with, but perhaps a motherboard with BMC could do it, they're not consumer grade hardware though and I don't think they'd sell enough for the added cost, maybe on those $1000 elite ones
Full boot is asking a bit much with everything a modern board does... but it's absolutely idiotic that anything wouldn't have flashback in this day and age.
Some are but having this kind of features is expensive, since many components (specifically the GPU card) are connected directly to the CPU for best performance.
Adding a way to control them while there is no CPU will require redesigning the traces and adding extra switches and might not be a sensible use of both space, performance and money.
@@urgay1992 I know that some manufacturers offer this as a feature. But that doesn't make it standard.
There's an exception to this "Pulling out the OS drive will BSOD". Namly, if you system has the Windows To Go flag enabled (or if it's an actual Windows To Go workspace), then pulling the system disk will just freeze. If you reinsert the disk to the same port within 60 seconds, then the system will resume working. Otherwise, it will power off the machine.
It doesn't actually just "freeze". You get a nice warning message that asks you to plug in the USB drive back, and warns you "Your Windows To Go workspace might crash".
As someone who's accidentally restarted the wrong SAN node: Windows Server will keep running even without any IOPS response for extended periods of time and resume operations once the SAN is back up. That said, this is a case of 'system drive unresponsive' rather than 'system drive disappearing'.
The computer says "I'm sorry Dave, I can't let you do that" 😅
10 insertions ? So my socket is a legend ? 😂
In many server system a lot of component including CPU (in multi socket configuration), RAM, PCIe cards, fans, drives etc are hot swap able so can be safely changed while the system is still running.
At 0:45 you say a modern socket is rated for only ten insertions?
That stood out to me wildly because I work at a tech shop and have to remove CPUs A LOT.
I looked it up and that claim doesn't seem valid. A lot of sites like tomshardware where that "ten insertion limit" isn't even mentioned when talking about swapping CPUs... Could y'all back that up or expand on that? Did Intel/AMD tell y'all? Idk it stuck out like a sore thumb and now I'm super curios where you got that claim.
I've never seen or heard of this before and did some quick googling. I work at a certain cough cpu maker and never heard of it
I actually did some research because i thought there must be limits specified. Finding specific socket info is kinda hard but i found LGA 775 socket specs and there it says ''socket must withstand 20 insert cycles''. So i don' see it unreasonable that some sockets only are made to withstand 10.
It's a "rating" not a limit. It's like how you need to get your oil changed every howeversomany miles/kilometers. You can certainly drive it a lot further than that and continue to use it but the chances of things going wrong increases.
Look at @SmolPotatowo 's reply; with caution with removing CPUs from sockets, the socket can potentially last significantly longer than its rating. The rating only shows how many times it can be inserted before the likelihood of something wrong occuring increases.
Matter of fact, the channel replied to another comment saying the exact same thing we are. Notice how the three of us also never specified a number - different mobos will have different socket ratings, but even with the scarce amount of research I did (thank you comment/replies :D), the consensus seems that 10 and 20 are the most common values for a given CPU socket rating. It all depends on the mobo. If anything, just another detail to think about when working with PCs :)
@@tmanx2724 Yeah i have never heard socket breaking other than bend pins. That's probably why no one ever talks about those numbers.
Techquickie is officially out of sane topics to cover 😂
Lol
What if you cut your computer in half while it's on?
@@50-50_Grind speedrun strats
next : what happens if you pee on your mobo.
I'd like an LTT video demonstrating this in action
In Linus Tech Tips channel they had a video where Linus yanked a CPU out of working PC. The GPU just proceeded to display the last frame in its buffer
"Jim, his brain is gone."
the damage is usually caused by a spark that is created for a short amount of time a high potential difference will be generated to damage FET transistors (which are very sensitive), diodes and possibly capacitors therefore a protective circuit is used in external ports to avoid this spark that can damage those components
It'd be cool to see a video about the potential of FPGAs for hardware acceleration
The windows + ctrl + shift + b seems like an excellent key combo to troll streamers with.
thank you for saving me some experimentation time this weekend.
I really like your videos as I’m not too good with tech things but recently got into learning about them and you explain it very clearly.
Bro brought out Filthy Frank
I know from experience, removing power to the hard drive on my old Windows 7 computer caused the computer to immediately reboot and boot to a "Boot drive not found" screen. The computer was built in 2013 and is a relic nowadays but it didn't hurt anything once I restored power to the hard drive. And no, I didn't just pull the hard drive out, it was a bad connection that dropped out all of a sudden.
I've pulled SATA-DATA from System's (Win10) drive by mistake yesterday, due to poor connection. System wasn't responsive, but was able to start browser and even task scheduler. This data was probably cached in RAM anyway, and a few minutes later completely froze, even tho the cursor was moving. No BSOD appeared.
The system was a haswell-based desktop I believe.
I'm surprised you've mentioned U.2, but not SAS or IDE.
In the 70s, I programmed on a small main frame, the power went off from lighting, the computer used core memory. When the power came back on, the computer continued processing where it left off with no loss of data.
This video was very well done! 👍
Thanks for the video!
Is Riley starting a new Computer Fetishes series ?
Maybe what happened if you cut down one of the CPU Pin ?
Com'on they have like more than 1K pins
The CPU will work perfectly fine because CPUs usually have duplicate pins
Well, I happened to know that the LGA 1155 socket used for Ivy and Sandy Bridge was rated for 20 insertions, and that there was a lot of talk when LGA 775 came out about processors no longer being user-serviceable because they don't use PGA anymore... LGA was touted as a reason why only technicians should be installing or removing CPUs now, and people were up in arms about it. Did future processors become that much more fragile? I wouldn't expect LGA 1150, 1151, or even 1200 to be a lot more fragile... but maybe with 1700 pins, now it is so fragile that it can't be replaced more than 10 times.
This is the sort of question I expect my non-techie friends to ask me out of sheer boredom.
I'd probably tell them what a dumb question it is, but secretly be like "but what happen though? 🤔"
LOL "When in doubt Blue Screen" hahahha
I had a hard drive fail while the system was working and the system just froze. I think it depends on the age and era of the hardware and windows
I must of changed my CPU about ten times before the pins started to bend (Intel S1151), had to try and straighten them again and it can be done, but it's not easy. Also blew up a GPU (years ago, maybe a AGP slot) by not plugging it into correctly, the end of the slot, and one of the pins was burnt.
these were similar symptoms after my boss knocked his cpu off of the desk while it was powered on lol
I had no idea parts are only rated for a certain number of insertions
This video crashed my smart tv. I thought it was a gag till it went on for too long
Should there still be a link to LTX 2023 in the description?
Great video! It's something I'm sure many people wonder about but, for good reason, don't even get to try it themselves!
What did Linus do that made everyone hate him in the last month? I haven't been watching them in a while
"Keyboard not found, press F1 to continue..." 🤣
I noticed that am5 cpu socket is a thing now. Could you do a "different cpu socket generations" video?
Ahh yes the age old question
Could you please post your sources regarding how many insertions each interface is rated for?
Today I learned about the Win + Ctrl + Shift + B shortcut. Neat!
new for me was the limited duty cycles of any slots of the mainboards
The craziest thing i ever hotswapped was a bios chip way back in time when it still was a "removable" eeprom on the mainboard.. i use " here because it was not that easy without the pc resetting.. but after a couple of tries i succeeded and could recover my friends corrupted bios flash on his identical mainboard....
Back in the day, I used to plug 1060 to a laptop using a riser cable to mini pcie. Hotplug worked if plugged in while in bios or in bootmanager os select list (I.e before os boot). The weird thing is that hot unplug sometimes worked with errors but sometimes just went to bsod or just black screen (on laptop screen). Never found out why it sometimes worked and other time didn't.
pci-e supports hot-plug by spec. unfortunattely, mobo manufacturers and os'es rarely bother to implement it in consumer products.
What about the memory extra option in Windows where if the ram is full it would use some of the system storage a essentially back up ram
what is Mobius from Loki doing in Tech quickie? 😂😂😂
It's very fun when your game harddrive fails on you while you're gaming.
Your game keeps running for a while, but gets progressively worse with each second.
Like i wasn't able to use the menu in the game at all hahaha
I’ve done this before. I pulled my ram stick out while pc was on. Computer froze but remained completely on with out any problems. Shut pc down removed all power and put ram back in. Started up fine. This is why I love micron ram. I was 14 when I did this by the way. I am 36 now :)
This like removing a game from a Gameboy while playing a from it at the same time, but on a whole other level 😂
I remember playing Dragon Age 2, off an external drive where the cable had a tendency to... fall off sometimes.
When it did so, the game would run perfectly well... UNTIL it needed to load something new (like a new area) and then i got an infinite loading screen...
IIRC there where some music glitches too, but cant be sure.
Something similar happened to me when a hard drive started failing while I was playing Skyrim. Right in the middle of a tavern, the game suddenly got super laggy as it had trouble reading data in. Might've been as simple as the tavern bard changing songs, or two NPCs starting a conversation.
It was a known failing drive I'd had issues with before, on an old PC I'd already replaced as my main PC -- so not a huge surprise. The only reason I was playing on it at all was that I had Linux on the new PC, and had never gotten Mod Organizer (with its virtual file system for easily loading and unloading my 200+ mods from the game) working properly on Linux.
"Heck, even just wiggling the heat sink can do this if you're too rough with it, and, *that's cool to know* "
🤪
0:43 Only 10 insertions? So does that mean you can only install a CPU 10 times?
yes it does
Yes and no, it means warranty won't protect past 10 insertions but odds are it will be able to handle more
@@FlashDrive356 how they gonna know how many times I inserted 😂
I’m glad I don’t have to listen to my inner child and try this myself anymore 😂 👍 For the next video though, please use some background “noise” that doesn’t sound like a phone is ringing all the time (and No, ofc I did not start looking for the ringing phone in my room while watching this video😅)
We need a full LTT video with Dennis or someone pulling a bunch of components while they’re on.
Awesome vid
Would you one over SLC Cache Mode where they simulate SLC memory in MLC, TLC, or QLC memory?
the number of people who tried the win + ctrl + shift + B shortcut after pausing the video on the exact frame has to be staggering
I used to unplug hard drives and plug them back in immediately to jump start a dead hd, sometimes it worked but more often than never, freezing the hard drive and then hot plugging it in did the trick.
Good jokes in this video, great writing
Had sticky thermal paste in my dads prebuilt, never had paste get sticky. I neglected the fact that my pga cpu would be yanked out if I pulled on the cooler. Seemed like nothing happened but I damn near had a heart attack.
Wow .. only 10 insertions over its whole lifetime? That's rough.
Once I was wondering why Cities: Skylines was all of a sudden very slow and really bad at drawing light rays like it had to draw them but didn't have any shaders to customise the rending of them. Turned out the HDMI cable was plugged into the CPU rather than the GPU because the monitor was being hot swapped for laptops back them.
one time i was testing a couple different CPUs in a pc, swapping them out-benchmarking then swapping again, I accidently pulled the CPU out before the PC shut off and there was a Loud zzzzZZZZZAP-POP with a nice bright blue lightning bolt between the CPU pins and the socket, right in my hands
Didnt check if that CPU worked and sold it for parts, mobo worked fine surprisingly.
frank has been gone for so long, yet he is still here
Make a video demonstrating what happens...
I've done the ram removal stuff... Got a screen with lot of diagonal lines
I was not expecting Papa Franku
I've removed old HDDs, my GPU(by mistake) and a DVD drive while my PC was running, even spilled water in one. But the only thing that's ever killed a component was a single splash of orange juice that hit my GPU on the back and fried it.
What I've learned is that as long as it's not orange juice, it won't kill anything. Probably.
Can you do arbitrary code execution by unplugging components from a computer while it is running?
For the people gasping at the low rating numbers.
These sockets are fragile. They are RATED at these low numbers because the manufacturer can't guarantee the component beyond this many insertions. It will work just fine if you use the slot more times than this on most occasions, but the metal pins may warp and the connections may become less reliable. Especially on LGA sockets.
Linus letting intrusive thoughts win.
What if a graphics card was pulled out without cutting off power you could solder wires to the ground and VCC pins of the pci-e connector, depending on the program or game it was running it would probably keep going.
So much to learn.... So much destruction, such little time.
Linux is "very popular" for servers in the same way that water is very popular for humans and plants 😂
I am a human being and I like water
I thought I was going to see blood and sparks, a more empiric video but I got Keys just talking about it.
it was like asking what if..organs were removed while the person is alive and not on anesthetics.
Filthy Frank!?! My software just became hardware.
I remember pulling out the HDD on my laptop once when I was a teenager while it was running. You're correct; it's a BSOD. :)
How about the opposite, plugging them in while the PC is on? I tried to plug in an HDD to a running PC and it stopped working entirely. I still think it could be recovered if anyone's got any ideas for me to try.
I whip mine out wherever I go. Work, library. When I'm at the coffee shop on my laptop, wait. They can't come out there it's soldiered down but you get the overall idea.
Your voice cadence reminds me of William forte's Chester V in cloudy with a chance of meatballs 2.
what happens when you plug in a new driver or something the like whilst the computer is running?
I never realized there was an insertion limit for the CPU ZIF socket. I would think the limit would be different for a PGA type CPU (e.g. AM4) vs LGA (AM5 or Intel) too.
Evidence suggests the size of the CPU can impact the total number of insertions it can experience.
I do remove power cable from gpu while running because amd gpu have bug which freeze gpu when running under kvm so removing power is only option
bare in mind that to use the on board graphics of your mobo, your cpu has to have support for graphics. Amd processors that do have a G at the end of the model number.
wow 10 insertions? so when we tested those trays of old cpus i guess we voided the waranty :P
I'v pulled my gpu out while the system was technically was on to clean it then put it back in and did not even reboot, the system was in sleep mode so the gpu did not have power, the os had no idea the gpu was ever removed
Your computer actively protests if you start tearing chunks of its brain out? Shocking.
"Daisy, Daisy, give me your answer do. I'm half crazy all for the love of you. It won't be a stylish marriage, I can't afford a carriage. But you'll look sweet upon the seat of a bicycle built for two."
I have a couple computers that allow hotswap sata drives and I love it. What a great feature
the mere idea of taking anything out while pc is running is just terrifying