A weird PC failure, freezing just seconds after booting
ฝัง
- เผยแพร่เมื่อ 5 ก.พ. 2025
- Check out Adamant IT 2 for the Pod Cast and More: / adamantit2
I stream on games Thursdays and Sundays, 7pm UK: / nethesem
Join the Discord right here: / discord
Support the channel directly at: / adamantit
Follow us on Twitter at: / adamantit
Check out our Instagram at: / adamant.it
Wow, that CPU had a massive dry spot!!!
Just to address this, because it's popping up in a lot of comments - I stuck my thumb in the thermal paste when taking the CPU out, so that clean spot was me!
@Adamant_IT ah OK. Thanks for letting us know. Great content buddy! Watching this for a long time...
I don't mind this format at all! It's nice to see desktop pc video's again, because those had kind of gone away. I get that the're a bit harder to record, but in my opinion this is a great way to do it.
Fix the camera focus for the monitor/PC distance. I saw it hunting every few seconds, looking like everything is zooming in/out just a tiny bit each time.
"It's never the CPU, except when it is." 😀 Great work Graham.
It's super rare! I think I've seen 3 bad CPUs in 20+ years of computer repairs.
It would be interesting to know how a CPU fails and if the failure is the same failure everytime.
I estimate it is likely an internal connection inside the CPU and likely heat related (thermal expansion) even dirty contacts can cause issues. I've experienced that with memory modules. With logic circuits it doesn't take much to cause lock-ups.
@@JustInspiredKent Another TH-cam channel (Greg Salazar) had a run of bad CPUs in his 'Fix It or Flop' series a while ago, from memory, usually AMD Ryzen 3600s. It would also be interesting to see the if the failure rate for Intel's high end 13th and 14th Gen CPUs reported by the tech sites is borne out in real life by the PC repair industry.
bro i was shouting that wooden sample needs more volts. from the beginning. i have a 3600 behaving like this at stock settings. degradation with years of usage in OC. needs overvolting to be stable
@@JustInspiredKent I had 2, an AMD with a burned out memory channel and a series 4 intel with a dead video chip.
"We got a problem here, we are running out of customer computer remaining."
That one really made me chuckle and laugh, especially because I have been in that situation with my own computer in the past.
“It’s never the CPU, except when it is.” Should be t-shirt.
That windows boot video is a must! Let's go!
Great video. Your Desktop repairs are my favorite. Thank you.
Thanks for doing your videos!
It's great to see a methodological approach, and the way you're explaining *why* you are doing things, explaining your thought process is *chef's kiss*. I've improved my soldering skills and made a good microscope choice because of you, so big thanks are in order!
Please never edit out the bloopers, like hitting the wrong power supply switch, or your recent "I need to thread the wires through the body of the camera first" oopsie. It's reassuring to see that even pros make these mistakes sometimes, and owning up to mistakes and showing how to fix them is super confidence-inspiring.
Love your videos Graham! We push out several 100 custom builds a year. Used the 5700G few years ago. Ran into more than one bad ones. Did the same thing stopped on the spinning circle. In the past was never the CPU Turned out all of them were in fact the CPU. Been doing this for 20+ years. Never is the CPU, except when it is!! Keep up the great content. Wish I could solder like you man!
Very nice testing and love that you explained your reasoning throughout!
God damn bro, me and my friend used to watch your videos when you had about 1000 subs and see if we could guess the issue before you did it. So glad you're doing so well
Use Ventoy and use one usb drive to rule them all. 😁🤣
My first thought that came to mind at the start of this video "It's never the CPU, except when it is".
High failure rates with Ryzen 5000 CPUs, seems the majority of them have dead memory channels.
But good to go over everything. Great video
Great to see a desktop PC under repair. Some of us still have a battery of capable PCs but have no requirement for a notebook. In my case it is because of multiple connections and break-out boxes for different device communication requirements, and high spec hardware and components to meet requirements for AI, graphics, photography, VFX, upscaling and video editing software, DAWs, racks of audio processors, samplers, synths, desktop midi controllers, synth keyboards, GFX tablet, scanners, and inputs for oscilloscope/test equipment for additional screen and data display. All feeding their vision output into decent spec 4K displays. The downside is if all equipment is running at once 96 AC outlets are required with total metering of current to avoid meltdown. And what seems like hundreds of metres of cabling. (After years of various "studio" setups I have learnt to have meaningful labels on both ends of a cable, AC sockets and plugs, and various input/output sockets.)
One of the things I always do when the mobo is doing weird stuff, is load defaults in the bios. There is no telling what the customer, or indeed other IT people, have been doing in there!
Had a much higher than normal failure rate on Intel CPUs in the last few years. Yes even 10/11th gen one.
You and I think alike. I just made the same comment before seeing yours. Award ourselves a star because AdamAnt_IT didn't think of that !
@@regwatson2017 what's the deal with this channel and armchair enthusiasts trying their hardest to outsmart Graham on every single video. Is it too hard to just respect his process knowing that he arrived at the correct solution in the end?
@@Bromon655 Mate this guy does this for a living every single day. I am an armchair enthusiast that builds a system every three or four years. I am going to pat myself on the back along with the other guy just like someone watching a game show if they beat the contestant in their lounge room. If you don't like that sorry tough luck.
Be careful with the assumption with memtest that "if the test passes then memory is not your problem."
I've had a couple head scratchers with memtest saying RAM was good that turned out not to be. One only showed up when I tested the memory one at a time. Another only showed up after a memory swap. That one I ran for a weekend and came in Monday morning and ended up swapping the RAM around for some reason and ran memtest again and it surprised me when it immediately showed up as bad.
On that system I was constantly having filesystem corruption issues; Until I caught that the RAM actually was bad and replaced it.
Just discovered you and really wanting to grow my knowledge with diagnostics. Love how you explain your through process on these!
Quality diagnostic. Breaking down a logical process. Ps channel 2 memory check next Saturday :) and a vid on how to make usb sticks covering memtest, a pxy boot and maybe a windows installer would be well received
I am now envisioning CPU chips spanking other CPU chips. Thanks Graham. 😊 Always love the videos. Keep them coming.
I had some OCZ memory back in the DDR3 days on a socket 1366 system that would run MemTest for 24-36 hours no problem. In Windows the system would randomly hard lock like your customer's, but it would take 5-6 hours to actually happen, sometimes more. I could stress test it all day sometimes, and it would run fine and then randomly hard lock shortly after. A friend of mine at the time, and a fellow tech, hated OCZ with a passion based on other systems he had worked on, and told me he was pretty sure it was just their crappy RAM even though it would run MemTest. Neither of us could find any definitive issue with the system after a lot of testing. I finally took apart my media PC at the time and installed the OCZ RAM into it. Sure enough, about 8 hours later the system hard locked. Completely different system. I stopped using OCZ products after that. Watching this video brought back that memory, so thought I'd share. Good job figuring out your customer's issue.
Once has a 16Mb stick of RAM (yes it was a while ago!), which would crash my PC. The shop wouldn't exchange it because it 'worked' with their testing. In the end, they DID exchange it, as I explained if there was nothing wrong with it, they could just sell it to someone else. I fitted the replacement in their shop and proved the computer now worked.
@@simmo1024 16MB? That would be well over 20 years ago now right? Mobo's used to be very very selective about RAM requirements back then.
@@DJ-Daz Erm, might have been 8mb, now I think about it, and in the mid-latter half of the 90s :) An DX4 100 board if I recall, or maybe an AMD '586'.
Nah, it was the memory, the replacement memory was exactly the same module, just a different one.
The problem with testing memory, is it is very difficult and very time consuming to do thoroughly. More so 25 years ago. Even today, just because the memory doesn't fail the test, doesn't mean it isn't faulty. Up until the point Graham swapped out the memory, I was 'It's the memory!'...
Always swap the memory to prove it isn't it!
@@simmo1024 I'm not doubting you, I was just explaining the difficulty in marrying the right motherboard and RAM back then.
I agree about testing RAM too, it's just easier (if you have spares) to swap out the parts.
Wiggle the ram in the slot while running memtest. I've found several intermittant RAM faults this way, the PC would run for hours or days no problem then suddenly lock up. Wiggling the bad RAM in the slot would cause the lockup almost immediately.
As a PC builder / Repair since 1999 I have had a few Ryzen AMD CPU fails, Never had an Intel fail.
That is very true, from what I've seen intel chips rarely fail but AMD chips have do at times.
True many AMD CPUs that fail
Intel is much more stable than AMD
Well, as computer engineer that have worked with all kind of hardware, for 30 years, i can tell you that i had more Intel failures than AMD, we never used AMD at a profissional level😉. But with my computers and people that i have helped with hardware problems, AMD and Intel, almost the same. Usually, the problems are user induced! Rarely was an hardware failure, by itself.
Love the desktop content, thats why I subbed in the first place and i enjoy the wide angle setup. Hope to see more in the future!
A great nostalgic run!
Thanks Sir!
Good job, awesome content! 👍👍👍👍
Woo a desktop repair! There's that phrase, I won't spoil which one!! The format worked just fine!
Just something to note. The same colored DIMM slots are on different memory channels to give dual/triple/quad channel memory. Moving the slots over 1 doesn't change the memory controller channel, just the slot.
Sort of. I wouldn't rely on colours for this, most of the time this is correct, but not always. You need to read the mobo to make sure you're moving to a different channel. But the gist of what @HueMongus101 is saying is correct. Vast majority of modern mobos are going to be A1B1C2D2 with alternating colours to indicate the channel pairs.
@14:20 very good statement about it being able to run an hour without freezing, yet not under Windows... I haven't finished your video and I will in a minute, but I'm going to put it in writing that it sounds like a specific hardware component or driver issue - since Memtest won't access special drivers or hardware.
Update: well, that was a surprise! CPUs now have so many integrated components that I could easily see how part of it could be broken without being obvious to the user. Not only the memory controllers but also USB controllers and of course PCIe controllers. So any number of failing ports or memory issues could actually be inside the CPU. You're right that it's extremely rare but I suspect my Ryzen 2700 to have a really crappy USB controller since it has always had issues with the USB ports disconnecting if they draw even slightly too much current. A powered USB hub that I tried did not solve the problem and that was a huge surprise. It got so bad that I decided to buy a USB add-on card and this solved my issues.
Small tip about that faulty cpu. Try disabling cores, and/or increasing voltage and/or lowering/locking clocks. Sometimes it's just one out of eight cores that's faulty. I live in China and lots of cpus with damage are still used just at a lower capacity. Who knows, it might still have 95% or original performance. Also you might want to check for bent pins or cracks around the edges.
Many yes'es to WinPE tutorial video!!!!
I had a 3600 CPU with "random freezing". It was unstable when the voltage dropped for low power modes. So I turned off "generate ACPI objects" in BIOS which prevented the OS from putting the chip in low power 'C states'. It's not a "fix" as such, but the PC works fine and if you got the CPU for free or less than half price it can let you keep using the PC for gaming etc.
I've seen symptoms like this twice before.
The first was a Pentium 75 with an external cache module which would randomly freeze in Windows. Disabling or removing the external cache module made the problem go away, so in that case it was the Cache not the CPU. Nowadays the cache is built into the CPU Die.
The other was an AMD AM2 CPU. The motherboard was top end and had a tweaker BIOS, which allowed you to turn cores on and off and override the CPU Microcode to enable features of the higher models. The CPU was officially a two core, but the BIOS would let you enable the other 2 cores. It turned out that one of the disabled cores was bad, but the other was fine. So using the BIOS I used to run it as a 3 core Phenom with the extra cache enabled.
When the bad core was enabled on the system behaved exactly as the one in the video did, so I think you have a dying core on the CPU Die.
fantastic educational videos mate, plz plz keep uploading, tnx
Can I ask how do you test watercooling could there be air in the system?
which may have caused the cpu to fail?
Im no expert am just asking out of curiosity.
Love your channel 👍
Very good video. I have a few computers (most AM4), and if one of them ever does "weird things", I will remind myself of this video to try and find the cause.
I appreciate the artwork in the frames. Both amazing games.
That was bizarre indeed. Hour long memory test but 30 seconds then crash in windows. And it wasn't windows !
Great problem solving tutorial, keep up the good work!
Great job, Mr. Lord! Keep up the good work!
Awesome channel and you always do awesome work
I do like this desktop problem solving videos, reminds me when I did these kind of things in the early 90's ;-)
Great video; have a fantastic Christmas.
I dread having an issue like that to troubleshoot. You took a very logical approach, but I hope I never have to follow it! Good job. 😊
issues like this are bound to happen to us pc users
why, it is as simple to diagnose as this, discard parts, you will find the source, it just takes time
@ Sure if you have a spare CPU you can try. Most of us don’t. If you don’t have onboard graphics you’d need a spare graphics card too.
@@somewaresim when you work with this kind of stuff or you have another desktop pc, that is where you get your parts from
Had something similar happen to me, where it would crash after 5 minutes. Turned out it was a defective batch of AsRock B350 motherboards. When I pointed a blower fan at the chipset heatsink, it would run for hours but still crash. Replaced it with a Biostar X370 which ran for 6 years before a cap exploded like a roadside IED.
I have a heap of Voltage issues on my R9-7950x when I got it. Had to spend about a year tracking down all the weird shiz. Turned out to be bios+volts. Finally the new new new BIOS from a month or so ago is actually stable (just set ram voltage to 1.3 now).
I had a very similar experience like this. I had an issue with windows trying to install drivers for NZXT ARGB Controller and it would instantly cause my PC to crash/BSOD. I then installed CAM and drivers manually and it fixed my issue.
My best guess as to why it could pass memtest but not run windows is memtest was probably running at a fixed clock (the same as whatever speed the bios runs itself at, likely baseclock), while windows would engage the power management and allow it to boost.. which would mean the stock vf curve baked into the chip at the factory wasn't stable any more. This is similar to what people can run into on these chips while using Curve Optimiser to lower the requested voltage too far, they become unstable on transitions in very light workloads when boosting really high and not getting enough voltage.
If you still had the chip around, you could likely make it functional by using Curve Optimiser in the other direction with a positive offset of say +5 to +8, or just disable turbo.. just to save on ewaste.
Years ago when I didnt have much money I tried diagnosis. Read online a bit. The memtest threw up errors. I tried taking one ram module out then the other. Then swapping slots. It was the slot.
I got about another year out of it before the Mobo caps blew under the Cap fluid controversy of the 2000's. Now I am old and cant focus my eyes on close up small stuff, so I get a shop to build and fix now. Also I suffer patience issues, so there is also that.
One thing hasn't changed is the gfx card upgrade treadmill. First card Powercolor Voodoo 1 1997.
Haven't watched the video yet, but I had this happen once and it turned out that a single core on the CPU had failed. Disabling that core in the BIOS was enough to get the machine running again. Took me a bloody age to figure that our of course.
Could there have been any value in looking at crash dumps (if any) from the couple of occasions where the machine auto rebooted instead of freezing?
Definitely an odd one. Great job, Graham!
great vid per usual graham. loads to take from this one. (Like many others) thank you
I once had a CPU fail when my wife turned on the shower. There was an immediate pop and the magic smoke escaped. When I got the machine apart there was a small crater in the middle of the CPU itself.
Note to self - don't install a computer in the shower.
for best results change the orientation of the AIO radiator. Radiator hoses at bottom 👎🏽.
@8:29 "caps lock is unresponsive" thats interesting my goto is smacking numlock to check for keyboard responsiveness.
Yup it is numlock for me too - except, now I think about it, if you have a TKL keyboard, there is no numlock!
Sorry if its a stupid question, Is there any point in checking windows event logger when trying to diagnose faulty hardware? Or is it only useful for software related issues??
Good video/information. I'm currently working on my PC. Was getting blue screens up to 6 times a day. Always after a Windows OS update. I din't think that it could be my Power supply. It served well for 14 years. Almost continuously on and running FAH. Hopefully the PSU is the only problem.
Strange problems I do the reverse... I pull everything out that I can. If the problem is gone I start adding in.
Were there any pins missing on the CPU? I've had a missing pin before and the system ran but was slower visibly, other pins do nothing, it all depends which ones.
Few times I solved this problem with cpu by cleaning contacts with something like crc contacts or wd40 contacts. Some times people use too much cpu paste that ends on the cpu contacts. Then they cleaning it and it's not obvious if a contact is dirty. Probably to a good inspection you can see it. It's true that it's almost never the cpu but dirty contacts can lead to this kind of problems. Anyway your procedure to isolate the problem is flawless and reasonable of course. Thank to share it.
Its possible but at the end of the day your own PC you might spend hours diagnosing an issue but as a business labour costs (by time), are likely mounting to the replacement cost of the replacement cpu.
@@Arachnoid_of_the_underverse I understand but what I suggest take about 3 mins. Just spary the contacts and then place the cpu on the MB 4 or 5 times so that contacts can react. I still think it's worth a try. After all this problem manifests in few minutes every time.
love seeing desktop pc for repair and keep them coming
Very round about way to go thru it but you did find the problem.
Had this exact failure about 4 months ago. Turned out to be the CPU. Blew my mind. I was getting pretty frantic and running out of ideas.
First thing I noticed when I watched this video was that the pipes from the cooling are going into the top of the radiator. That's kind of sub optimal cause the pump might suck only air when there is air in the system. So, the CPU might have run to hot in the past and got damaged?
Strangely had a ryzen 5950 fail on me. Same symptoms as you demonstrated.
Soak tested in BIOS for 24 hours with no problem, as soon as it went into the OS, it froze.
First ever cpu failure
I had this exact same issue too. 5950x.
Not really a surprise, issues started with the ryzen 3000 series. As the 5000G series are actually not the same architecture as the non G, they are plagued with the same 3000 series issue had
I know there's a lot of talk about Intel CPU reliability at the moment, but I feel like Ryzen 3000 onward have been a bit dodgy in terms of dying for seemingly no reason. Greg Salazar's fixing PC series seemed to find many dead R5 3600s for a while. They are only 5 years old.
There's a little bit of that, but also the fact that AMD sales have sky rocketed over the past 5 years, meaning there's _a lot_ more of them in the wild to fail. Actual percentage failure rates are really hard to find. I'd say Intel have a good track record of bomb-proof CPUs, but well... recent 13/14th gen failures kinda make that irrelevant as well.
The fact the computer stayed on for an hour through the memory test with a dead cpu has me baffled. Flabbergasted. Confused and hurt inside. What a weird ending.
Lol, not really confusing. On AMD, the RAM is sort of separated from the cores with the Infinity Fabric. So with virtually no load on the CPU, it passed the RAM test. However on Intel, the IMC is heavily integrated with the cores, which explains how an AMD user can pass y-cruncher with barely stable RAM for instance, but an Intel user would just instantly crash their system, because RAM problems cause even an known-good Intel chip to crumble under almost any load. Different design, different consequences. Also, the 5700G is a horrendously bad chip. Low cache, and mine which I got for fun a couple years ago came with bad PCIe 4 support so it would hang randomly if not set to PCIe 3 which destroyed performance. Sent back, refunded. That chip also ran hotter with a good cooler than any other Intel or Ryzen chip I'd ever seen. Using a half-cached CPU with a dGPU that one never needs is pure moronicism. No one should ever do such an idiotic thing. Get a 10 dollar Radeon card from ebay or keep an old card for emergency situations without a dGPU. This user likely suffered the very, very, very bad advice circulating on reddit that you should get a half-cached APU for the 1 in a million chance your GPU will go bad. The user will be astonished how much the extra cache benefits performance on the 5600X.
@@lePurpleDragonAMD has the issue that it doesnt instantly crash on memory corruption and it just ignores it to some point
@@AlpineTheHusky AMD has ERROR CORRECTION (basically data retransmission), which in mildly bad cases keeps the whole system going with mild blips, like audio cutouts and slight slowdowns, usually very brief (fractions of a second) and as RAM timings become too ridiculous (or unmanageable) or the FCLK is set too high, this turns into heavier issues, hanging, and finally crashing.
Error Correction in and of itself is good, because even casually tight timings may have rare, 'fixable' errors on a nanosecond level that do not need to take the whole system down. And in bad cases, why crash the system and risk data loss/corruption if it can be avoided?
It's good design, not AMD idiosyncrasy. AMD basically just used what Radeon and NVIDIA already know and use from GPUs, which REQUIRE good error correction, and implemented it on the CPU side. Same as why they use clock stretching.
@lePurpleDragon ECC is meant to CORRECT errors. Ryzen ECC only really reports errors and expects the OS to correct them. And since the architecture EXPECTS for ram errors to get corrected it ignores feedback errors. Thats why many NAS systems either use Intel or Epyc or very basic ram speeds. Intel stops execution once ram errors cause corruption. Ryzen tends to keep going. EPYC got true ECC
@@AlpineTheHusky Lol, ok. I didn't call it ECC for a reason (which is on the MEMORY not the CPU/IMC)... scurry off boyo...
Windows (and other complex software) uses many other cpu opcoee that (probably) memtest does not. Using in some trigered/timed functions inside OS or services those opcodes with silicon defect can cause cpu nmi/halt and not even kernel had the time nor interrupt code for that exception, hence freezing.
Fun video, nice change of pace
like your show. any format is fine
Did you tried cleaning the cpu pins with Isoprophyl alcohol and toothbrush?
I've seen a similar failure with a Lifebook A359 (i5-8250U) as well, except that even Memtest86 would faceplant with protection faults and other weird issues (including a black screen), only the BIOS worked OK. I suspect the CPU might need a reflow, but we're not doing that around here.
Had an interesting one recently, a prebuilt gaming system (3700X/32G/3060Ti/512G SATA SSD + 2 TB SATA) that was dog slow. Task Manager revealed that the Netac 512G SSD was maxed out at about 7 MB/s writing, and CDI showed 111 TB host writes and 722 TB NAND writes. In short, the controller seems to have a bit of a problem with write amplification and the NAND is pretty much worn out now as the SSD has seen its fair share of thrashing. Might be the first time I've seen that. Being 3/4 full (after cleanup) wasn't helping matters. I recommended a new M.2 SSD of at least 1 TB.
Had up to 4 AM4 with warriors same issues most is down to dead memory channel / Greg Salazar has had a few of those to on FIX/Flopp series 🙂
No bent pins?
bent pins prevent ypu from inserting cpu, unoess ypu are a stubborn mofo and break them
it is most likely a voltqge problem, i wouldnt trust on the motherboard eithet, possibly caused this
I’ve had this happen just from an abundance of dust. I actually set up my air conditioner to blow directly into the casing while I worked on it.
Awhile back I bought a 16Gb set of Mushkin RAM which booted and ran fine in Linux. When I tried installing Windows it got stuck in a boot loop cycle. I went through many of the experiments you did in this video with much the same result. It was a massive headache. In this instance where you're trying to fix something of this nature I would guess it's chip creep or a faulty processor. When you took the AIO off I noticed a portion of the IHS that wasn't covered by heatsink grease which may be the cause of the problem. That is unless you wiped that part off in the process of removing the water block.
The hardest part to diagnose this type of fault using this method (which is the best method, IMO) is to have the spare parts to test...
I've always wondered if it would be possible to check for a failing CPU using the thermal camera. Maybe one specific corner is getting much hotter than the rest of the die. Maybe worth a curiosity video in the future!
@16:19 its windows light, can't believe it's not windows.
It happened to me to find memory errors on a second pass, they were new memory modules and the shop I bought them accepted them as faulty and they refunded me. I get your point though.
I had a 3800x exhibit similar fault posted fine, install windows comes under load conks out! You diagnosed the fault all the comments are neither here nor there - fault found and fixed! Although this was a hardware fault, when I'm fishing for potential OS/software faults I like to look at Windows error logs to see what Windows was trying to do just before the crash for clues.
The process of elimination, will always rule in our universe, Brother!
"We are running out of customer computer" 😄 Still, it would have been more fun if the fault was actually the keyboard
Had a similar problem with a Ryzen 5 3600 in my own PC a few years ago. Drove me nuts trying to diagnose it -- but eventually the only possible conclusion was faulty CPU and I got it RMA'd.
Nice to see a desktop computer in for repair for a change, obviously not so much for the customer!
What about CPU warrantying or just off warrant.
Out of curiousity I would disable some cores of the CPU. Check if 1 or more cores are the culprit. It then still can work with the cores left.
Why the cpu is dead. Just wasting time no point.
first step is clear the bios-
next check the memory- use a known working stick and in different slots to check for a dead memory channel
does it have a debug light- it will show where it is stuck in the boot
test the power supply
try a different GPU
last but probably the problem is swap out the CPU, then last but not lest is the mother board
Those AM4 chips have a history of dead memory channels. I have seen that issue many times. Wonder what the failure rate could be?
my son had a simular problem, freezing crashing, memory error's graphics card problems, I checked the various components, it was looking like the mother board, I changed the power supply and everything all working ok. checked the power supply under load and the 12 volt rail dropped to 10 volts at 5 amps "power supply 750W"
Pls read your comment again🤣🤣🤣🤣
SO much can go wrong with computers.. 🤔 I had similar problem with one PC build back few years. It turned out that the new PSU was faulty in it. That caused it to hang on at start like that. 👀 And also that it did workes erratically. It freezed or booted and worked like 3-5min them restarted it self..
What an interesting mystery to solve.
It has to be overheating, it runs longer when cold then locks after warming up faster and faster. AIO is at the top which means possibly more air than liquid is getting in or even a random clog that gets pushed out and builds up. When grabbing the AIO hoses only one side will feel stronger than the other which means that hose is pushing the liquid and the other is the return. If your not mounting the cooler at least put a small dot of thermal paste to bond the copper and IHS for testing only. I am editing while watching the video, I do not understand why you did not try an air cooler?
I have had a similar problem but with various symptoms including the freeze on boot but also sometimes running slow and sometimes not recognizing the GPU in Adrenalin software. The difference with mine is after a few restarts it will run all day.
This is something that 'comes good' after warming up. These cold-start faults are horrifying to track down as you have to wait hours between each test for the system to go cold again...
wow has been a long time seeing one of your vids in my feed, how are you? It could have been bad luck on the silicon side of things, while intense gaming sessions don't really hurt a CPU it can lead to it, I hope the customer was happy seeing it back agian
It's never the cpu... yeah, maybe it's time to change it that it's most likely cpu 😅 you know Intel plague and amd problems with cpu 😂 never the less, it was a great video ❤ one question, cpu light during some crashes was off, and I was ohh was it cooling pump off to? But no, it was a cpu 😅
funny that you call that ram ceappy kit 😂 i run two times crappier 😢
probably it's time to start to think for upgrade all system
Worth it if you're running a modern CPU - anything ryzen 3000 onwards really does appreciate RAM speed that's 3000m/t or more. Might only count for a few FPS in some situations, but in others it can be 10-20% performance from 2133 to 3400
So if the memory channels were ok on the dead CPU, could it be some or all of the data channels, had become damage/corrupted?
i had with my system where my memory itself was fine but something was faulty on one of the modules causing it too intermittenly fail too post and cause crashes (I did memtest it and whenever i did it came back clean). rmad the memory everything has been fine since. im assuming there was a broken solder joint causing the issue.
I've noticed quite a lot of issues with these 5000-series Ryzen CPUs for some reason, at least compared to previous generations.
I've had to RMA two 5600Xs and one 5500 for customers because they've started to cause boot/freezing issues. One of them started working if you heated the CPU to around 100c but only for a short while...
Interestingly I remember recently seeing return rate data from an OEM... It was some high performance system builder in the USA, can't remember the name. This series of Ryzen had their worst return rates by far, it was data they shared during the Intel 13900/14900K scandal.
Of course no further detail was shared on why the machines were returned and apparently they sold a lot more Intel machines than Ryzen based ones but it was a statistically significant failure rate.
Never the CPU, calling it before intro
First guess, PSU now I will watch and see if I was right or wrong.