i also have had experiences where i have a buyer that would replace the parts that i sold him with exact same parts but are defective. so always make sure to have records of serial numbers of the stuff you have to make sure he didn't swap stuff out for defective ones.
@@paulf1071 early on in school for a+ cert, a few dead pentiums the old socket kind...then the gpu was a hd 5750 that was used and dead on arrival. And actually recently I delidded with a razor a i5 3470 and It starts but won't post/ no display so RIP...lol I replaced it with an e3-1235 though.
@@lagginswag I hope you're grounding yourself properly when doing your repairs! There's a lot to be said for wearing an ESD wrist-strap for this type of work.
@@paulf1071 yeah I always work in the same spot and touch part of the case/psu and also hold down power button to drain. I ofc have a wrist strap but never end up using it, static isn't the issue.
I bought a used 4770k on ebay for a friends budget gaming PC . It had very similar problems to the ones in this video the thing is that it was already delided and it wasnt overheating. Turns out the cpu just wasnt making contact with the motherboard pins and crashed etc etc. Real pain to get the thing working though good thing the YES man didnt have that problem with a actual faulty cpu lol also LOVE the content keep up the good work mate.
I once had a 4790k that had some weird issues with low power states (c6/c7), the system would freeze after ~5 mins in idle. Disabling these low power states fixed the problem.
I had that on a CPU (not Haswell), and it was the lower voltage causing the instability. It would run the lower speeds fine if the voltage was manually set. What you did makes perfect sense. It's been tested and the power difference is negligible between letting the clock speed and voltage drop and manually locking both. The current is drawn on load anyway, and the power difference between a small process completing at full speed vs taking longer at a lower power draw is small. It comes down to efficiency at that point. When I saw the numbers on a modern CPU with everything locked vs all of the power savings running, I stopped worrying about it.
Haswell was kind of a wonky generation. I had a 4790k that was an awful overclocker, couldn't barely get to 4.6 GHz and had it running most of the time on stock because it was so unstable, really.
@@pedrobastos8132 I'm actually still using a haswell CPU, an 4770k that I picked up rather cheap in a bundle. I overclocked it to 4.5GHz/1.32v and it runs perfect in combination with my 1060. Temps are also good using an Thermalright Macho Rev. B, the only issue is the core to core thermal difference, core0 to core3 are under load 7-10°C apart. I might delid it soon but since the temps in general are fine (70°C max at non-avx load) there isn't really a need for that.
I had a faulty CPU in my 2016 Touch bar Macbook (i7 6820HQ) and it was incredibly hard to get Apple to accept that there was a problem with the machine, their own internal stress tests couldn't replicate the problem, so even if say rendering in premiere would cause a crash report from the kernel saying the CPU was bad, or I'd have kernel panics they wouldn't accept a warranty claim because it'd pass their diagnostic.
Everyone keeps saying the same thing. A dead cpu is rare. But whats even rarer is a brand new 9900k dying on the first gaming session. How do i know its dead? Swapping out the Cpu to a 8700k, it boots. 9900k it just wont boot. So here i sit, listening/reading the countless amounts of “its very rare a cpu dies” being told to me. While slowly accepting the fact that im hella unlucky...
Hey dude, when I build machines for people I take pictures and keep a file on my NAS of the build and finished product. To many times people will take stuff apart and screw it up and then try to come back and say I made a mistake. Then I pull their file and show them the pics and that pretty much settles the issue then I charge them to fix their screw ups.
The only faulty CPU I ever saw was a Cyrix PR 200+ that ran way too hot and kept crashing no matter what standard cooler I used. Went back to the shop and swapped it over for an Intel 166 mmx that I overclocked to 200Mhz from the day I got it and ran perfectly ever since. But I guess no-one remembers these anymore... like Intel's first low budget overclocking beast, the Celeron 300a... I'm getting old...
i had a dual P2-333 system back in the day, and ran into it getting unstable and found that the fans on the backs of the cpu cards had clogged all to hell. after cleaning them out (positively nasty after seven years of multiple climates, and the fact that i smoked didn't help), it became stable again, with no actual damage as far as anything that i could use to test it.
I had a Cyrix 5x86-100. Cyrix chips were notoriously hot CPU's, but I found most of their problems stemmed from software bugs and incompatibilities. I started into overclocking with the Celeron 300a on an ABIT BH6 motherboard, it was a very simple overclock of the FSB from 66MHz to 100MHz and CPU clock speed increase of 50%. The Celeron 300a was actually a better CPU for gaming more often than not vs. the much more expensive Pentium 450, because of faster L2 cache even though it had 1/4 the amount. Overclocking back then was physical dip switches and jumpers on the motherboard, not fancy BIOS settings.
Memory, bent pin or dust in socket, blown PCI lanes, bad sata, fried chipset. It is rarely the CPU though i have had it happen but your right it's usually memory.
@Adam W I already have seen more faulty or even dead CPUs (i5 6600k, i5 6400, FX 8320) than any of these problems though. A buddy of mine bought an i5 6600k from Mindfactory (biggest German/European e-tailer for pc parts) and left it at stock (MSI Z170-A) and it died just months later, meaning it would crash @stock constantly, had the Windows boot screen watermarked in UEFI - it was weird all around - and ran fine lowest clock of @2.4 GHz in Prime95 for just three minutes until BSOD. Swapped memory, board and PSU and the problems persisted until I swapped the CPU with an RMA sample, no problems since then... I also built a PC for someone on a budget with an i5 6400 from eBay from a commercial seller with >5000 feedback (99.9% positive) as "known working, even optics checked" and it came bent. I know Skylake had some problems with bending, but never had that problem myself until then (have an i5 6400 in secondary PC @4.7 GHz with DDR4 @>3800 MHz xD), so we thought the bluescreens (somewhat counteracted by adjusting system agent voltage and similar) and program crashes would simply be caused by the (DDR3 bclk oc Z170) Gigabyte motherboard which was in for RMA already. Since we also tested and another chip (almost known good Skylake ES "QH8F" - guess who inspired me) also was unstable, this was pronounced. Turned out the motherboard was bad and was repaired just luckily a few days before warranty expiration from September 2016. After repair strangely everything worked fine only with memory in channel B, with any out of piles of memory in channel A and/or B the PC did not boot. Turns out the chip was damaged due to the bent. Since I life in Germany I thought: "No problem I have two years warranty by law with six months of "burden of proof reversal" (direct translation of the law), so I just need to tell the seller. The seller declined because I did not return the item within two weeks after purchase. Excuse me, someone does not know law here. After many fruitless mails I let the case rest until I met a lawyer that would take my case for free (only would take a cut of the fine), so I got 60€/71€ back from my purchase, after the seller received a legal letter. The chip however still works fine in single channel even @4.3 GHz (4.4 GHz was unstable @1.38V so we stopped there) with just two slots and it was almost free. :D Also know a guy whose FX 8350 was dying (random bluescreens, lacks and program crashes or complete boot failures). He against my advice did buy a new motherboard and had the same problems (PSU, HDD and RAM already known good and replaced). He eventually got fed up and sold the chip as non-working alongside the other used parts on eBay - so I borrowed him my aforementioned spare PC with overclocked i5 6400 and he (apart from his new motherboard, the CPU and RAM) still uses the same parts and only had one bluescreen since he hot-plugged a HDD and no further problems. And since he is no stupid fanboy, he is considering saving up for a Ryzen build this year because of the value and new components. He really did not want to switch to a random Intel platform just to keep his DDR3, which I second. Furthermore I have some CPUs with bent pins, but that is another story, since these CPUs are technically repairable unlike with internal damage which is irreversible. So yeah, CPUs get faulty or die more often than many may think, source: my experience: Built and mantained about ten PCs 1x memory (very cheap stuff) 2x bent pin (once did it myself, repaired all) 0x dust in socket or similar 1x blown PCI lanes 0x bad sata 0x fried chipset, maybe more but idk for sure, could be 0 to three times Meaning all known to me apart from mostly irrelevant or fixable cases of system failure mentioned in the list are less for me than dead or failing CPUs, so really not that uncommon sadly. Still way less likely than RAM, other motherboard component, HDD or PSU failures though. Only unfixable and stability affecting one was RAM in one case for me. Also never had any fan die on me ever apart from two very cheap ones which died within the first second of use (one the same fan header any of my other fans worked fine and they are similar CPU fans).
@desertrat810 That all makes sense, since I think the most failing CPUs are either old hot Intels like Pentium 3/4 or similar AMDs like Sempron, Duron or FX and newly those flashy new Intel chips. Hell I've never seen that many dead CPUs before Intel 's now infamous 14nm node and I don't think Intel's "rise" in popularity or dumb and careless customers are to blame - tech gets more fragile nowadays and who can repair CPUs that small? I think running them with bad components can kill them, but I admit I put an i5 2500k into a bent socket for testing and the socket was shorted (VCC to GND) so that it basically unsoldered a gold contact pad from the CPU which I had loose afterwards. The funny thing is I send the board to a person repairing it for 40 bucks and put the CPU back in and everything worked fine. So, I guess moral of the story: Don't do that even if it worked out for me. xD Really any of the CPUs I describe here are not very old just 7 months for the 6600k, at most since Skylake released (then 2.5 years ago) for the i5 6400 and I think 8 years for the 8320. Especially the latter ones die like flies, but who cares FX was long dead anyway I recon and any real gamer can attest that. So to anyone doubting: I have many CPUs and none of that that died/are dying were run overclocked at any point by myself, so that is ruled out as a factor. In contrary most of the other CPUs I have were tortured to the brink and many were in now dead motherboards and still work just fine.
Memory 9 out of 10 when you are installing it while power is running through the motherboard as well as handling the memory stick on the memory chips rather than the edges of the DIMM as shown in this video.
I recently picked up a Xeon E3 1270 V2 for $60 on eBay and it performs like a $100 i7 2600 if not slightly faster. I was worried that I would run into microcode compatibility issues cause HP does not list Xeons as compatible with their Q77 chipset boards, but the CPU worked without a single problem.👌
Another great video Bryan. It's a very rare occurrence, nearly always cpu cooler ( dust/ dried thermal paste etc ) related but worth bearing in mind after checking the more common faults.
Have had 6 years with my 4790k/ASUS Z95A MB box. Starting having intermittent shutdowns, got different BSOD errors every time. My work is tech so I tested/swapped RAM, PS, Graphics card etc. I changed CMOS battery, and reset BIOS with F5 in Asus to default. What fixed it was THIS VIDEO. The setting that made all problems go away was CPU CORE RATIO > Sync all Cores. This did not happen when using the F5 deafault. The bios had been running default since it was new. I suspect this CPU/MB combo won't last forever, but for now, this fixed it. Hopefully this helps someone. Thanks for an excellent video. Just the right level of detail to be of great help.
Nope... We shouldn´t have to. When paying up to a 1000 dollars for a product, it should be expected to already come in perfect condition and with good paste or liquid metal. It should actually be soldered to the cpu itself like amd has done with their cpu´s.
@@datreja9573 We should get a decent product even if we pay a fucking 100$, especially every K and X cpu, why are we paying the premium for them then? 1st time I bought from intel and was already disappointed with their garbage 7700k toothpaste. Had to spend additional 9$ for some liquid metal and 2$ for some knife to delid it. After that I could finally have 5ghz with 75c and not 100+.
I found this video completely by random and have been having an issue recently with my machine shutting down and up until watching this video i guessed it were my ram or motherboard. Turns out my i7 4970 needed delidding after all these years! Subscribing to you this video is a massive help!
Glad you mentioned it could be heat instead of something else TYC, I had an old FX 6100 and that thing would get dusty, so much so that after a couple months without cleaning that it would simply turn the whole system off without an error, just a black screen and everything powers off. It's important to maintain your system, and if your previously working CPU suddenly starts crashing, check those temps! General rule is:
The office space my mom works at has been using HP computers with i7 4770s for over 5 years I think. It was supposed to be amazing, too of the line performance, but I’ve nicknamed those computers the “Disappointments”. If there is a dilemma with 4770s, that would partially explain why they all suck. They also have HP’s horrible motherboards, no cable management, tiny tiny AIO cooler, and mismatched ram (an 8gig stick with a 2gig stick). So glad they are getting replacements and I’m more glad that I’m being commissioned to build them.
👍!👍!👍!👍! Awesome job, I appreciate you simultaneously showing & describing what you were doing when overclocking in thd last 1/4 (a quarter of) of the video.
Different story, but something was wrong with the PC of my mum. It kept crashing and never turned on when we hit the power button. We sended it five times to the merchant who sold us the pc. Every time he checked it, he said it was fine but never fixed anything. My mother was one of those people who threw everything away if something didnt work and would buy a new one, so I suspected the merchant never looked inside of the pc, which everyone should do, if a customer tells you something is wrong with the pc. We returned the pc again and he became salty because he had to check the pc again. Then I told him "Then you clearly never looked inside! Seriously! You cant do small tests and say everything works well." It was clear in his tests the pc was running fine, yet he never looked inside, simply because all what he did was to check if the pc is turning on. After telling him my mind, he did check the components. It turned out the PSU had a faulty, he replaced it and apologized for the mess. I asked him what kind of a technician he was and we didnt pay the last repair bills for all this mess. To that time I was 14 years old and didnt knew much about pcs.
Been in the computer industry for 30+ years and the only Intel CPU I had fail was a Pentium P5 90 MHz CPU in 1994. Had more problems with AMD CPUs, but not in along time. I have a 4770K in my current PC which has been running fine since I bought it new in 2015. Looking forward to the new 3rd gen Ryzen processors for my next upgrade.
THANKS! you saved me from throwing away my 7700.... I bought this second(?) hand cpu for an old mb and it worked fine for a couple of days, even benchmarks all below 85C until it started freezing dead before boot... it just froze either before post, during post, or after a few seconds at the bios screen when I managed to hit F2 in time... I swapped it for an 7100 i have laying around to modify the config to lower the ratio, fix the cpu core voltage at 1.125, and disable turbo-boost... then back to the 7700 and it FKIN WORKED :D... It's running hotter than usual but at least it's not an expensive paper weight now :D
I like this video, and conclusion you came to. I've had 2 3770k's overheat severely with even minor overclocking/over-volting. I ended up de-lidding the second one, applying liquid metal, and it now stays cool and stable at 4.7 ghz. I think de-lidding is almost mandatory at this point with these 3rd and 4th gen chips, if you want a decent, stable OC.
Great video, lots of information. I've fixed lots of computer processors, but that was when they were still made from desecrate components. I had to troubleshoot and repair them to component level, replacing capacitors or resistors or IC's or transistors or diodes. Now that was before micro processors like today's, these were on one to six circuit board cards connected by via a BUS. A lot different than what you are doing, but at that time it was the in thing to do, lol...
I have an i7-920. Sometimes in summer it started to have weird crashes. Turned out it was overheating to 99°C. I completely de-dusted everything inside the computer and it has been running smoothly since that. It is easy to overlook heating problem possibilities until you face one first time.
A computer should be cleaned and thermal paste changed about every third year I would say. Depends of course on the environment. There is software like HWinfo64 that can help you keep track of your computer and thermals. Thermal Paste always gets hard with time. That and dust would probably be one of the two major reasons why laptops tend to die rather soon (3-6 years).
Recently I did a build with a 4770K. OC'ed to 4.4GHz on a Corsair H110, it maxed out at 79c during Aida64 (after some minutes). I was pretty sure it's been delidded by previous owner (Sometimes you get pleasant surprises when hustling 2nd hand hardware. :) Another way is to see how quickly the temp rises after the stress test startup. If it instant goes to max temp, is the connection between die and IHS is the limiting factor.
I had that once, where I replaced every single component of a computer. In the end I even replaced the case. It would run for a week, even in my office, then just hang. Of course the customer just assumed I was a lousy computer technician. It was very hard to convince them that this was an incredibly unusual situation.
Cool video. It is a two for one. You also show how to destroy memory by installing it while power is running through the motherboard as well as handling the memory stick on the memory chips rather than the edges of the DIMM as shown in this video.
My recent issue with 4th gen i7 4770 is it never gets heated up 😂😂😂28 to 30 c only. My system shuts down after 30 min Or an hour. Never starts after that. May start after a few hours. After watching your video, and changing CPU ratios it worked. On stress test the temperature was 70-c.b At idling 35-c. The system worked for 5 hours now, today. Thanks buddy
Don't know if anyone's pointed out but, next delid, use some nail varnish on the tiny resistors next to the die. The liquid metal can migrate and short them, which is not good. Other than that, great little 'how to' Brian. [thumbsup]
I've been building and repairing computers since 2008 and I've only come across 1 bad CPU, which was a dying Core i7-2700k. It needed extra voltage to boot into the system without crashing. One of the contacts (Lands) on the bottom of the 2700k was black, not gold. On some ebay listings, I've seen "Parts Only" processors that were missing capacitors on the bottom of the CPU. So it does pay to visually inspect the CPU. I didn't need to run the P95 method because I had a working 2600k. In fact, I've used Prime95 Blend to diagnose DRAM problems. Sometimes with the DRAM and CPU stability problems, a thread on the Prime95 test will stop running. With the memory I can swap and add sticks in order to find the bad stick. Another guy I know scrimped and saved to his funds to get a Core i7-6700k, which had a problem with one of the memory channels not working. This resulted in 2 of the memory slots not working. At least he was upgrading from a Core i5-6500 which recognized all 4 sticks, so his problem was easy to diagnose. He tried raising the VCCIO, VCCSA, and DRAM voltages but this didn't help him. He even tried reseating his cooler and everything was good.........no bent pins. There was another time I put a mono-block on my system, and the uneven mounting pressure caused one of the DRAM channels to quit working properly. All I did here was I reseated the mono-block, and everything was fine after that. So it pays to make sure the cooler was mounted properly. These days when I delid my processors, I use liquid electrical tape on the contacts that are next to the die. This will keep the liquid metal compound from shorting those contacts and killing the CPU. Some people like using clear fingernail polish here, but I like using the liquid electrical tape better.
My 6600k randomly freezes the entire computer (it had nothing to do with OC or C-States, if it didn't freeze PC, it would run any test). I have been looking for a problem for a long time, blaming RAM, motherboard, power supply and disks. Until now, I had no idea that the CPU has more states than two "works" or "doesn't work". Well, man learns his whole life;)
There is something perversely enjoyable about watching someone track down what is causing a computer issues, jay did a few of these videos a few months ago but sadly stopped.
I'm not sure why, but despite the exact same architecture and only 100mhz speed difference I've never come across a Haswell xeon that's had the same issues with temperature, even spite of the fact many where used in much higher duty cycle loads with smaller cases/coolers than there consumer breathern. Could it have something to do with crystallization of the TIM with heat cycles rather than actual drying, or perhaps that these CPU's never saw attempts at being overclocked? Or may there have been a difference in the binning/selection criteria? Either way, they are a pretty cheap/widely available part still, so if your buying used and your not getting a K skew, I'd compare prices.
You are onto something there, I noticed with some of the mining experiments I did back over a year ago, that some GPUs would be fine* to just keep on running, but take them down for a week then try booting them back up and you would have to change the thermal paste. Something to do with cycling from hot to cold?
umm I'm gonna say voltage is the answer? my E3-1246 v3 needs only 0.89v to maintain 3.5ghz and 1.01 for 3.9ghz "turbo" speed, the 4770 I got rid of needed 1.19v at 3.5ghz and 1.22 at 3.9ghz... both kinda won the silicone lottery, and tested on the same boards. the 4770 needed more volts, hence ran hotter than the Xeon at the same speed...
@@C-S-J True, however the Xeon E3-12xxv3 of the LGA1150 socket is in fact thermal paste, not solder, and are the directly comparable parts to the i7 4770 of the video referenced in the video... and yet doesn't seem to face the same age related overheating or instability as the consumer i5 and i7 chips. This is why I'm speculating about either binning differences of the core itself or effects the different use cases have on the longevity of the paste. And the conventional logic that more use and hotter running would correlate to drier paste and higher temperatures later in life doesn't seem to hold up with my experiences on the platform.
Xeon's are better binned parts for Enterprise systems that need power efficiency and stability. So of course it makes sense when a Xeon counterpart from the same architecture runs cooler and with less voltage than their desktop counterparts. But yes, Haswell was a very hot architecture in general (even hotter than Ivy Bridge) compared to Sandy Bridge, and most overclockers were better off with the 2600K due to it's best performance without the hot temperatures since it was soldered. That may be why a lot of people never upgraded from their 2600K until Coffee Lake came around. This was fixed when Skylake came out with the 6700 being quite cool when you're not bumping up the voltage, just to be undone once the 7700K came out due to the high clock speeds and obviously the poor choice from Intel to use TIM on enthusiast parts. Ivy Bridge and Haswell overclocking ARE NOT the same procedure when compared to Sandy Bridge. The 22nm Tri-Gate transistors are much more sensitive to voltage than the 32nm transistors we saw with Sandy Bridge. This can prove to be a pain when overclocking since you may need to bump the voltage to reach higher clockspeeds, but that may cause the heat to ramp up with it. And we also know that voltage does cause a linear increase in heat either. As for thermal cycling, that's a lot worse on a CPU and GPU than if you ran it constantly at the same temperature. Large temperature spikes can prove detrimental to hardware. It's why mining GPUs are usually actually in better condition than a GPU used for gaming because a GPU used for gaming will be thermal cycled A LOT between hot and cold temps. And there's miners that actually ramp the fans up and undervolt their GPUs also to keep low temps. Thermal cycling can cause PCBs to warp over time and solder to crack. In fact, the Xbox 360 had this issue where it would get very hot under load, cool off when not in use, and over time this caused the solder that held the GPU in the socket to crack and cause the GPU to stop working, and thus the red ring of death would occur.
i usually just use IPDT from intel and when it says it's good, i just trust them. but if something else goes wrong, it usually is very accurate what's a bad processor or not. takes five minutes.
Well, prime95 is nice, but usually takes a lot of time before you can encouter an error. What is somewhat better to test CPU/RAM is Intel burn test which uses linpack. Usually it is enough to run 10 loops with maximum ram to find if the computer is stable or not :)
Filip Doležal just throw in call of duty 4 blackout run zombies it’s best stress test yet trust me got my i7 9700k at 5.2 ghz at 1.325 volts its amazing !!! Stays in 50-65 while gaming on just 120mm aio in push pull and fan over vrms!!!!
you can only truly test stability with a mix of stuff. start with the IPDT, then stuff like aida64, then intel burn test, then prime with FMA3, then real-world crunchers like seti@home. i have found that everything would be fine, until seti@home (lunatics avx client) would show it to be unstable. if you are looking for an unconditionally stable oc, you need to have a large set of synthetics and some real world stuff.
AVX generates more heat, and you may need to set up an AVX Offset in the bios, to downclock the CPU by -1 or -2 on the multiplier when AVX instructions get hammered like that.
@@mrlithium69 I think AVX offset was introduced later. I have the Gigabyte Z97 flagship board for my 4790K, the GA-Z97X-Gaming G1-WIFI-BK, and it doesn't offer that. I'm pretty sure that ASSRock didn't either. Not sure about Asus, as my first modern Asus board is a Strix Z370-I (prior to that, my previous Asus was a 286 board).
For testing CPU only use Small FFTs in Prime95 NOT the blend test, and use HWMonitor for temperature monitoring when Prime95 runs at the background, Prime95 will give errors if the CPU is faulty or unstable overclocking. For memory testing use memtest
A faulty CPU rare? Nahhhh, at the store I work for, I even had clients, who had faulty power lines that caused problems and resulted in random shutdowns under load. Or a faulty case that also caused random shutdowns. Yes a case, not sure what it was, but it was gone after we changed the case (, after we had changed everything else. The client thankfully was a very understanding one). Anyway, great video.
@【】_【】 Not sure what you mean by that? If you want to imply that I lied to the client, it was him who suggested the cause. It even was fixed after he installed a UPS.
I had the same problems with an Engineering sample cpu, it is running mostly fine now without changing much of the settings. I did change out the motherboard battery and that increased the booting stability a bit.
The bad thing about these CPUs overheating is that over the long term will last shorter and will have more probabilities to die sooner. So that brings me to the next question: and how about the alarming high temperatures found in gaming laptops? These usually get to a heat level that is borderline the maxium temperature that these dies can handle.
My 4790 is on the warm side too, but since it's not giving problems and as you say it's still somewhat expensive, I'm not going to risk delidding. But maybe some day.
@@webbie7503 I have the same cpu. But it gets really hot. (50 - 55 C idle) (70 - 85 when gaming) I dont know if it is the cooler or the cpu whos causing the issue. The cooler isn't great i think its a spire sigor iv, but anyways i just wondered what cpu cooler you have i need tips :D
I had a customer bring me a Dell optiplex 9020 they bought from the off lease section. The computer would not boot and the warranty had been 90 days and they were at 100 days. So they just told me to get them a used 9020 board and I5 to replace their I7 4790. At that time , nobody had listed an issue with the power switch. I still have not read the issue and now I know this issue. I still have the board and chip I pulled but I did discover that the power supply was blown. Since I did that as a favor to them, I no longer had any extra parts around for checking stuff. My guess was the power supply blew hard and took the board and chip. There was an issue with the 8 pin power supply they used having a very high failure rate. Anyone else disagree with the assessment ? The power supply did not have any power and when their used same Dell 8 pin power supply came in, I plugged in the I7-4790 board and did not get any fans or lights or beeps. I was unable to check the CPU without old parts, The customer was using the computer daily so, I did not pull the I5 and take a chance on damaging their board in any way. So chance the I7-4790 is alive ????? I vote it is dead because of the definitely dead PSU and Board. PSU had the reputation of popping as it died with a high failure rate.
Over 2 years now I've gone through 3 sets of RAM, a 1070 and 2070 GPU, an SSD and NVME.2 storage for the OS, and reinstalled windows several times. Yet my system still crashes semi-regularly. Sometimes daily. Sometimes weekly. Raaaaaarely I can get 3 weeks without a crash. I'm confident I have either a bad CPU. Why? Because while installing my AIO cooler I was trying to squeeze it on suuuuuper tight with a LOT of pressure, before I realized I could loosen the screws to get the AIO mount on, with ease. And you know how AMD's PGA sockets are.
Im definitely a budget shopper so going with a ZALMAN cooler i don't mind. Plus i always used them. I found the biggest one that supports up to 300watts and stuck that on my 8700K. I got a Gigabyte Aorus Gaming 7 cause of it's high end VRM components and after watching Der8auer's video on it, I had to get it. Plus it looks good with the ZALMAN 9900MAX and a Sapphire Nitro+ RX 480 8GB OC (RX 580) with it's back plate almost perfectly matching the design on the board. All in i built a z370 monster for only $670. I see PC's prebuilt going for over $1500 with similar specs. I can say im happy with the choice to go z370 and not the i9 9900K route. I might upgrade to that in the future being that the MB can handle it. (8700K $280, ZALMAN 9900max $20, Z370 Aorus Gaming 7 $150, 16GB Gskill 3200MHZ ram $100, 650watt Corsair PS with case $120) had the rest of the parts SSD's and HDD's. $670 is what i would've had in just the 9900K with a proper IO and MB.
I've been building systems since the mid 90s and the only new CPU I ever had bad out of box was a Ryxen 5 1500x, it had the dreaded segfault bug. It took me months of diagnosis and fighting with AMD to get an RMA because the serial number on my CPU was not supposed to be an affected model. Beware of Ryzen 1000 series parts floating around on the second hand market, you may well buy a defective CPU and be plagued by random inexplicable crashes.
Theres 10% wiggle room on ATX voltage specs. And only 12v is relevant here, which further gets converted both in the motherboard VRMs and in this case on the CPU FIVR.
I've only ever seen one CPU failure, was a real pain to diagnose. it was an AMD A-series APU. The computer would just shut off randomly, ruled out heat related issues and swapped the power supply and nothing changed, it would just shut down a couple times a day. I pulled the GPU and used the onboard graphics and had artifacts even in the BIOS screen. Apparently the GPU portion of it had failed. Replaced it with the best CPU I could get for that socket which was I think a shitty Athlon X4, re-installed the discrete graphics card and everything was fine after that.
Good to see you use Prime95 too Brian since it puts a more demanding load on the CPU. I would not worry too much about Haswell CPUs. My Haswell is fine. My guess it was abused before you got it because my 4790k has been overclocked to 4.6ghz for years now and no thermal problems
@@Brandon-uy1uv Lol i know, but down under we just call all types of spray wd40, its a weird thing. The aussie export, multipurpose stuff is the best thing going.
I had a problem with a CPU i7 6700 for months! It always had bluescreen with the WHEA uncorrectable error. We noticed it would bluescreen when idle or doing light tasks but never on load. Long story short we fixed it by putting voltage to 1.35V.
hey m8 i've a question for the 'Yes man' ^^ can you explain why you use prime95 over aida64 ? Keep your awesome work, i really enjoy your happiness and that you always keep smiling !
in about 20 years, I have only ever known 2 CPUs die, one was an AMD K62-350, WAAAAY back.... I don't remember much about the cause, and another was more recent, it was during a re-build of an i5-2400, built into new case and it just stopped working, swapped everything out, even put everything on a new board and was still nothing - swapped out the CPU and it all came back to life... I do however believe the CPU took one of the RAM channels out on the board, as only 3 of the 4 slots will work now... But also recently (2017) I built a 7700k system, and I ended up returning it TWICE to Intel as temps were not under control, were hitting low-mid 90's in less than 2 minutes when gaming...
Great video on a lightly covered topic. Ive had a couple cpus appear faulty due to thermal paste getting on the contact pads, even a tiny amount that is barely visible can cause issues. Same thing happened to a video card as well
I had a i7 4770 that I delidded, I didn't seal the IHS to the CPU little board. I used Arctic MX4 both between CPU die and IHS and between IHS and cooler. It worked fine for about one year and a half.. until last week when it simply refused to boot showing nothing on the screen. I took it out, I re-placed it in socket, and now it works again. Before delidding I had 70 Celsius in idle.
i recently brought the i7 4770k and i can honestly say ive not had any issues at all.. mind you i upgraded mine originally from a i5 4460 cpu that id had for years.
i'm still running the i7 3770k, oc'ed to 4,7ghz and still running strong after 7 years... i was getting high 90° at 4,2Ghz and i delidded it immediatly after getting it, and now it's sitting in the low 80 at 4,7Ghz, glad i chosen the i7 instead of the fx8350 since the fx nowadays is really weak compared to the i7
Tough break, kinda sux you aren't certain if some of the hardware is what you actually supplied... Q: Do you have any sweet spots for 3770 non-k cpu's? I delidded mine a few days ago with great temp success running at 4.644ghz with 1.235 volts via 43x108 peaking at 66c under stress. Anything I can do to improve, is that my lottery sweet spot? Great vid as always, I know a dude that bought one of your builds and am always surprised by the bang for buck with whatever he demands of it, you sure know how to tune!
i mean it's a testament to Intel that we look everywhere else first before blaming the CPU. back in the 90s, Asus was like that for their MBs and Kingston ram had a lifetime warranty for a reason.
I have a i7 4770k and have no idea why the pc keeps just shutting off. Sometimes the pc is just idleing and it shuts off. Just watch low stress content and boom off. Thank you for letting me know about it's thermal grease. It's come down to the motherboard and CPU. Retesting the ram one more time before I delid the cpu. Other option would be to buy a motherboard first and see if that works fine. That would be sad though had an amazing formula VI motherboard. Going to just buy new set of ram. Worst case scenario I end up with extra ram for my system. Do let me know what would you recommend I do?. I have changed the power supply, tested both the rams, added a new cpu cooler with new grease, tested it on unraid still turns off, remove all hdds and tested it still turns off, removed most usb plugs, changed grease on gfx card which made it 12c cooler understress. Ram test just turned off was at 92%. Temps are generally really good on the cpu. Below 37 max 50 to 60. Will try Prime95 tomorrow.
I had some boot-looping problems with my 4770k (and or mobo?) a couple years ago, and I still occasionally run into issues especially when overclocking. Without a doubt the most frustrating thing to fix. I could get it to boot with only 2 sticks of ram after clearing cmos but it would only boot once (it would start looping after shutting down or restarting), so I had to clear cmos after each shutdown (or oddly enough change which ram sticks I had inserted). I finally fixed it after messing around in the bios by setting all voltages from the default "auto" to "normal" with the exception of vcore and ram voltage which I set manually. After that it had no issues even with 4 sticks (it wasn't vcore or ram voltages that were the issue since those were the first I tried experimenting with). If I were to guess, there must be some sort of auto-voltage-calibration thing that's not working properly but I have no idea whether its a problem with the motherboard or cpu, but it must be a hardware issue since updating/switching bios didn't help. I've been wanting to sell my 4770k and upgrade to ryzen but I'm afraid someone will buy it and cant get it to work and I'll have a situation similar to yours. :/
Not many people know this but if you place the faulty CPU at the large end of a celestial grade telescope you can spot the microscopic bug that is causing the CPU to fail. At this point a pair of surgical tweezers can be used to carefully extract it, at which point the CPU should work fine again.
i had a faulty i3-7100 on a cheap computer I built for someone. It would BSOD shortly after getting into windows. I replaced everything, motherboard, power supply, memory etc.I finally tried a 7700k cpu and it worked absolutely fine. The CPU was almost brand new so RMA through intel, they asked a bunch of questions and had me mail in to verify faulty. Sure enough it was and they sent me a brand new in the box cpu, even got another cpu fan.
My first pc was a Q9550, I picked the parts but had the store build it. Once arrived it worked but not great and had higher temps and seemed to thermal throttle. In transit the cpu cooler must of been shaking because the thermal paste was cured with airgaps on all but the very center of the cpu. Cleaned it up and new paste and it was running normal. This system also had the OCV-Reaper 1066ddr2 that would cause issues on stock volatages or if placed in certain slot combinations with bootloop.
On a side note to whoever reads this comment: nowadays every cooler out there will come with some sort of pre-applied thermal compound which most of the time is $0.10/g crap (some coolers do come with decent stuff so do your research if you can). I recommend everyone to scrap that white stuff off, get yourself some decent thermal compound like Kryonaut or MX-5 (if you want more longevity but decreased performance over Kryonaut) and replace it, your temps will probably be about 5C lower. Plus it will probably last longer unless if you're constantly running 80C+. Kryonaut's break-down point is 80C, MX-5 *should* last longer in a high temp scenario but will not offer the same level of performance. There are more expensive solutions that might endure longer than Kryonaut in a high temp scenario but I personally think that by the time it starts losing effectiveness, your system will also need some cleaning, so it's not hard to also repaste at that point. On GPUs I highly recommend to test it on day 1 and then straight up repasting with Kryonaut once you're sure it's working fine. You will get better performance after the repaste, specially on Nvidia GPUs with the way GPU Boost works. In my case, I use a slim gaming laptop as my daily, my CPU (i7-8750H) runs 80C+ under load and these high temps lead to the thermal paste to break down/dry out faster than normal (I use Kryonaut). So I just replace the thermal paste whenever it's time to open my laptop to clean dust build up, which I probably do every 3 months or so. Every slim laptop I've seen out there running an 8750H will inevitably CPU thermal throttle at some point due to design limitations, so every degree counts if you want max performance and longevity.
My Ryzen 2700X was a faulty CPU. The impossible happened. Wasn't hard to identify it/ It was just unstable. System was freezing whatever the memory settings, whatever the voltage. I managed to keep it stable at 4.2 GHz and 1.34v (yes the chip didn't crash at around 1.3v in prime95: "golden chip") with 16GB of cl14 3200 memory. XFR2 and PBO2 didn't work at all. System hang up while booting. After a few weeks nothing worked. Even 3.7 GHz with 1.4v and cl18 2133. I tried different hardware. Wasn't the powersupply either. And my Crosshair VII went through RMA too, they sent it back, so by conclusion it wasn't the motherboard either. Well RMA just gave me a new Ryzen 2700X. Been working fine in the last 6 months. And probably will just do fine in the next 5 to 10 years.
@13:57, we can clearly see a "gap" in the glue on the PCB. May that be contributing to the thermal paste actually drying off ? i.e. the humidity just escaping ? If so, what would be a good test we could come up with, without delidding, to test the glue between heat spreader and PCB ?
I once had an AMD system shutting down due to 'overheating'. I replaced the cooler a couple of times and the thermal paste a couple of times. Eventually, I became convinced that the motherboard 'thought' that the system was overheating and shut down. A new motherboard solved the problem.
I see a blast from my past with that mobo, seen the word RANGER and i got goosebumps. I had the A88X ROG Crossblade Ranger paired up with a A10 7870k Black Edition and x4 4gb Corsair DDR3 vengeance Pro Editions at 2400Mhz AND X2 ASUS STRIX GTX 970S BOTH WITH nzxt g12 bracket AND corsair H110I V2 280MM RADS AIO ON THEM, SADLY THE PC WAS LOST TO A FIRE IN MY FLAT, WAS MY VERY FIRST REAL BIG PC BUILD, CASE WAS THE HUGE PHANTEK ENTHOO PRIMO SE EDITION, NOW I JUST HAVE LOTS OF PICS TO LOOK AT, WOULD SERIOUSLY LOVE TO BE ABLE TO BUILD IT AGAIN BUT ,OMEY VERY TIGHT, SORRY PPL, FOR GO ON ONE, MEANT A LOT TO ME
Here’s what I’d tell that customer. Good luck finding a tech as knowledgeable and experienced as this guy who has half the business sense to get you an actual good deal on a nice system.
Generally it's a pretty good idea to run a Temperature Monitoring Program while doing this to make sure your not pushing the CPU Temperature outside of it's recommended operating limits because a overheating CPU can cause this exact problem........
Bought my 4770 when it launched, about year ago temps on it got horid, delided it and put on new thermal paste, didin't bother to seal heat spreader, temps are now great and cpu still going strong.
Is the i5 4690k in the same boat for the thermal paste?? Or did they fix it going from haswell to devils canyon?? Awesome video, had no idea something like this could be fixed relatively easy :)
The only 'faulty cpu' encounter I've had was a i7 860, which had a faulty memory controller so half the motherboard ram slots wouldn't work (I tried a different cpu and the all of the slots did work, so I was sure it was the i7 860)
Wow that's bad. I think I have a faulty X399 board that acts the same. The only defect/weak CPU memory controllers I've seen are Ryzen (My Ryzen 1700 can only do 2666 reliably, though that's technically not defective since anything over 2133 is "overclocking", its still REALLY weak!). My Threadripper 1950x does great with quad channel 3600 (plus some overclocking on top of that!), BUT.... If its cold, around freezing, (which sometimes it is in my room), it can't run full speeds. At roughly 0c, the Threadripper starts dropping memory connection, and 32GB turns into 24, and then 16... And then you need to drop it down to 2133 just to boot, then run some loads on it to heat it up so the rest of the memory detects. I'm 99% sure the motherboard is faulty based on how it just drops sticks of RAM completely when its cold. The lowering of stable memory speeds however is normal, and a known fact for Ryzen based processors acting up when cold (so don't run dry ice or phase cooling on them, but liquid nitrogen gets cold enough it starts working again for whatever reason). This memory issue was a huge pain in the butt when I was testing my LN2 cooler with regular ice (so the coldest it could possibly get was above 0c since there was melted water at the bottom of the cooler, still had these issues, so it can happen in the morning on any cooler). I'm very certain that its a bad board though, the CPU seems to bench fine at 4.15 Ghz on a 120mm AIO and 3600+ Mhz quad channel RAM most of the time, it only seems to crash when its doing nothing in the desktop (or 4x Handbrake encodes at once, but that could be a software conflict, though I normally run 2x at once...)
I have a 4770 and I'm now wondering if my CPU has gone bad. Temps are fine, usually 30's when idle, 50-60c when gaming, but 2 times in the last 3 months my PC has done a slow freeze and lock up. When this happened, I was playing a game on one screen with Chrome open on the other with task manager and Afterburner background. One time the game froze 1st, then Chrome. Other time it was reversed. After those 2 froze, the task bar would slowly become unresponsive. Afterburner freezes. Any other programs open would also slowly freeze. Task manager is always the last to freeze, but clicking End Task on frozen programs does nothing. Then it's frozen too. After about 5min or so, the mouse and keyboard stop responding. Only 3 things I haven't swapped out is CPU, main SSD and MB.
![[안방1열 풀캠4K] 베이비몬스터 'DRIP' (BABYMONSTER FullCam)│@SBS Inkigayo 241110](http://i.ytimg.com/vi/AQ4RIghEvUs/mqdefault.jpg)
i also have had experiences where i have a buyer that would replace the parts that i sold him with exact same parts but are defective. so always make sure to have records of serial numbers of the stuff you have to make sure he didn't swap stuff out for defective ones.
This is such a scummy thing to do
( I know this is obvious but I needed to say these words out loud)
@@aboyaser5608 yes, well, it also obviously happens.
If it's a large issue, or something without obvious/ easy way to track, you can also consider using a uv pen to mark parts that won't easily be seen.
Great advice for when i'm selling parts.
thanks for the tip
In over 15 years of tech only had like 2 dead cpus and a dead gpu...in terms of actual silicon. Rare but it definitely happens. Good video!
What type of CPUs failed for you? I'm having some strange workstation issues lately.
For me, CPU is the last component that could fail in a PC.
@@paulf1071 early on in school for a+ cert, a few dead pentiums the old socket kind...then the gpu was a hd 5750 that was used and dead on arrival. And actually recently I delidded with a razor a i5 3470 and It starts but won't post/ no display so RIP...lol I replaced it with an e3-1235 though.
@@lagginswag I hope you're grounding yourself properly when doing your repairs! There's a lot to be said for wearing an ESD wrist-strap for this type of work.
@@paulf1071 yeah I always work in the same spot and touch part of the case/psu and also hold down power button to drain. I ofc have a wrist strap but never end up using it, static isn't the issue.
I bought a used 4770k on ebay for a friends budget gaming PC . It had very similar problems to the ones in this video the thing is that it was already delided and it wasnt overheating. Turns out the cpu just wasnt making contact with the motherboard pins and crashed etc etc. Real pain to get the thing working though good thing the YES man didnt have that problem with a actual faulty cpu lol
also LOVE the content keep up the good work mate.
I once had a 4790k that had some weird issues with low power states (c6/c7), the system would freeze after ~5 mins in idle. Disabling these low power states fixed the problem.
Some of the c-states can act a little funky on Haswell. I shut off most or some times need to shut off all the c-states with Haswell for overclocking.
Get the same craptasting thing,the C6 setting,on an AMD cpu'ed board.
Disabled it,running fine even now 8)
I had that on a CPU (not Haswell), and it was the lower voltage causing the instability. It would run the lower speeds fine if the voltage was manually set. What you did makes perfect sense. It's been tested and the power difference is negligible between letting the clock speed and voltage drop and manually locking both. The current is drawn on load anyway, and the power difference between a small process completing at full speed vs taking longer at a lower power draw is small. It comes down to efficiency at that point. When I saw the numbers on a modern CPU with everything locked vs all of the power savings running, I stopped worrying about it.
Haswell was kind of a wonky generation. I had a 4790k that was an awful overclocker, couldn't barely get to 4.6 GHz and had it running most of the time on stock because it was so unstable, really.
@@pedrobastos8132
I'm actually still using a haswell CPU, an 4770k that I picked up rather cheap in a bundle. I overclocked it to 4.5GHz/1.32v and it runs perfect in combination with my 1060. Temps are also good using an Thermalright Macho Rev. B, the only issue is the core to core thermal difference, core0 to core3 are under load 7-10°C apart. I might delid it soon but since the temps in general are fine (70°C max at non-avx load) there isn't really a need for that.
I had a faulty CPU in my 2016 Touch bar Macbook (i7 6820HQ) and it was incredibly hard to get Apple to accept that there was a problem with the machine, their own internal stress tests couldn't replicate the problem, so even if say rendering in premiere would cause a crash report from the kernel saying the CPU was bad, or I'd have kernel panics they wouldn't accept a warranty claim because it'd pass their diagnostic.
Yeah because apple is stoopid
Stay away from Muchbar Touchbars
If I was stupid enough to purchase a 2016 macbook and it went wrong I wouldn't be advertising that fact on the Internet
W19 ELY What is the point of being a dick? Also, why are you guys so concerned with how someone else spends their money?
Thanks for being judgemental pricks, I use a Mac because I primarily edit with Final Cut. I have a desktop gaming PC for my other needs.
Everyone keeps saying the same thing. A dead cpu is rare.
But whats even rarer is a brand new 9900k dying on the first gaming session.
How do i know its dead? Swapping out the Cpu to a 8700k, it boots. 9900k it just wont boot.
So here i sit, listening/reading the countless amounts of “its very rare a cpu dies” being told to me. While slowly accepting the fact that im hella unlucky...
rip monies
Had a i5 6600 with a faulty memory control sensor, he was constantly restarting my computer. Intel refund the money and i bought a 7700k
SKYDRUM bought a i7 7700k with refund from i5 6600?
Rui Heng Chua I need to give just more 50$ and they sent me a i7 7700K
how did you know it was a he? Maybe it was a she. just saying
I have a problem I turn it on cpu light flashes then goes to dram then does it all over again forever, Same as you?
@@ahsheigjeej873 bruh U HAVE TO BE KIDDING YO LMAO
Good video Bryan, keep the content coming mate.
A tip: you can watch series at flixzone. Been using them for watching a lot of movies lately.
@Jett Landon Definitely, been watching on flixzone} for months myself =)
@Jett Landon Yea, been watching on flixzone} for since november myself =)
Hey dude, when I build machines for people I take pictures and keep a file on my NAS of the build and finished product. To many times people will take stuff apart and screw it up and then try to come back and say I made a mistake. Then I pull their file and show them the pics and that pretty much settles the issue then I charge them to fix their screw ups.
The only faulty CPU I ever saw was a Cyrix PR 200+ that ran way too hot and kept crashing no matter what standard cooler I used. Went back to the shop and swapped it over for an Intel 166 mmx that I overclocked to 200Mhz from the day I got it and ran perfectly ever since. But I guess no-one remembers these anymore... like Intel's first low budget overclocking beast, the Celeron 300a... I'm getting old...
Dolf M I had a intel 8088 Ibm with a 20Mb double 5.25 drive dos 5.0
@@thefinalroman Nice, a bit before my time though! Did you overclock it? ;)
I was 7 had no idea how to overclock till the Pentium 233mmx came out
i had a dual P2-333 system back in the day, and ran into it getting unstable and found that the fans on the backs of the cpu cards had clogged all to hell. after cleaning them out (positively nasty after seven years of multiple climates, and the fact that i smoked didn't help), it became stable again, with no actual damage as far as anything that i could use to test it.
I had a Cyrix 5x86-100. Cyrix chips were notoriously hot CPU's, but I found most of their problems stemmed from software bugs and incompatibilities.
I started into overclocking with the Celeron 300a on an ABIT BH6 motherboard, it was a very simple overclock of the FSB from 66MHz to 100MHz and CPU clock speed increase of 50%. The Celeron 300a was actually a better CPU for gaming more often than not vs. the much more expensive Pentium 450, because of faster L2 cache even though it had 1/4 the amount. Overclocking back then was physical dip switches and jumpers on the motherboard, not fancy BIOS settings.
Usually it's the Memory 9 out of 10 times
Memory, bent pin or dust in socket, blown PCI lanes, bad sata, fried chipset. It is rarely the CPU though i have had it happen but your right it's usually memory.
@Adam W I already have seen more faulty or even dead CPUs (i5 6600k, i5 6400, FX 8320) than any of these problems though.
A buddy of mine bought an i5 6600k from Mindfactory (biggest German/European e-tailer for pc parts) and left it at stock (MSI Z170-A) and it died just months later, meaning it would crash @stock constantly, had the Windows boot screen watermarked in UEFI - it was weird all around - and ran fine lowest clock of @2.4 GHz in Prime95 for just three minutes until BSOD. Swapped memory, board and PSU and the problems persisted until I swapped the CPU with an RMA sample, no problems since then...
I also built a PC for someone on a budget with an i5 6400 from eBay from a commercial seller with >5000 feedback (99.9% positive) as "known working, even optics checked" and it came bent. I know Skylake had some problems with bending, but never had that problem myself until then (have an i5 6400 in secondary PC @4.7 GHz with DDR4 @>3800 MHz xD), so we thought the bluescreens (somewhat counteracted by adjusting system agent voltage and similar) and program crashes would simply be caused by the (DDR3 bclk oc Z170) Gigabyte motherboard which was in for RMA already. Since we also tested and another chip (almost known good Skylake ES "QH8F" - guess who inspired me) also was unstable, this was pronounced. Turned out the motherboard was bad and was repaired just luckily a few days before warranty expiration from September 2016.
After repair strangely everything worked fine only with memory in channel B, with any out of piles of memory in channel A and/or B the PC did not boot. Turns out the chip was damaged due to the bent. Since I life in Germany I thought: "No problem I have two years warranty by law with six months of "burden of proof reversal" (direct translation of the law), so I just need to tell the seller. The seller declined because I did not return the item within two weeks after purchase. Excuse me, someone does not know law here. After many fruitless mails I let the case rest until I met a lawyer that would take my case for free (only would take a cut of the fine), so I got 60€/71€ back from my purchase, after the seller received a legal letter. The chip however still works fine in single channel even @4.3 GHz (4.4 GHz was unstable @1.38V so we stopped there) with just two slots and it was almost free. :D
Also know a guy whose FX 8350 was dying (random bluescreens, lacks and program crashes or complete boot failures). He against my advice did buy a new motherboard and had the same problems (PSU, HDD and RAM already known good and replaced). He eventually got fed up and sold the chip as non-working alongside the other used parts on eBay - so I borrowed him my aforementioned spare PC with overclocked i5 6400 and he (apart from his new motherboard, the CPU and RAM) still uses the same parts and only had one bluescreen since he hot-plugged a HDD and no further problems. And since he is no stupid fanboy, he is considering saving up for a Ryzen build this year because of the value and new components. He really did not want to switch to a random Intel platform just to keep his DDR3, which I second.
Furthermore I have some CPUs with bent pins, but that is another story, since these CPUs are technically repairable unlike with internal damage which is irreversible.
So yeah, CPUs get faulty or die more often than many may think, source: my experience:
Built and mantained about ten PCs
1x memory (very cheap stuff)
2x bent pin (once did it myself, repaired all)
0x dust in socket or similar
1x blown PCI lanes
0x bad sata
0x fried chipset, maybe more but idk for sure, could be 0 to three times
Meaning all known to me apart from mostly irrelevant or fixable cases of system failure mentioned in the list are less for me than dead or failing CPUs, so really not that uncommon sadly. Still way less likely than RAM, other motherboard component, HDD or PSU failures though. Only unfixable and stability affecting one was RAM in one case for me.
Also never had any fan die on me ever apart from two very cheap ones which died within the first second of use (one the same fan header any of my other fans worked fine and they are similar CPU fans).
@desertrat810 That all makes sense, since I think the most failing CPUs are either old hot Intels like Pentium 3/4 or similar AMDs like Sempron, Duron or FX and newly those flashy new Intel chips. Hell I've never seen that many dead CPUs before Intel 's now infamous 14nm node and I don't think Intel's "rise" in popularity or dumb and careless customers are to blame - tech gets more fragile nowadays and who can repair CPUs that small? I think running them with bad components can kill them, but I admit I put an i5 2500k into a bent socket for testing and the socket was shorted (VCC to GND) so that it basically unsoldered a gold contact pad from the CPU which I had loose afterwards. The funny thing is I send the board to a person repairing it for 40 bucks and put the CPU back in and everything worked fine. So, I guess moral of the story: Don't do that even if it worked out for me. xD
Really any of the CPUs I describe here are not very old just 7 months for the 6600k, at most since Skylake released (then 2.5 years ago) for the i5 6400 and I think 8 years for the 8320. Especially the latter ones die like flies, but who cares FX was long dead anyway I recon and any real gamer can attest that.
So to anyone doubting: I have many CPUs and none of that that died/are dying were run overclocked at any point by myself, so that is ruled out as a factor. In contrary most of the other CPUs I have were tortured to the brink and many were in now dead motherboards and still work just fine.
Memory 9 out of 10 when you are installing it while power is running through the motherboard as well as handling the memory stick on the memory chips rather than the edges of the DIMM as shown in this video.
My i5 4460 died on new year at 24:57
I recently picked up a Xeon E3 1270 V2 for $60 on eBay and it performs like a $100 i7 2600 if not slightly faster. I was worried that I would run into microcode compatibility issues cause HP does not list Xeons as compatible with their Q77 chipset boards, but the CPU worked without a single problem.👌
More like a 3770 even, since it's Ivy. Very nice find!
Another great video Bryan. It's a very rare occurrence, nearly always cpu cooler ( dust/ dried thermal paste etc ) related but worth bearing in mind after checking the more common faults.
Have had 6 years with my 4790k/ASUS Z95A MB box. Starting having intermittent shutdowns, got different BSOD errors every time. My work is tech so I tested/swapped RAM, PS, Graphics card etc. I changed CMOS battery, and reset BIOS with F5 in Asus to default. What fixed it was THIS VIDEO. The setting that made all problems go away was CPU CORE RATIO > Sync all Cores. This did not happen when using the F5 deafault. The bios had been running default since it was new. I suspect this CPU/MB combo won't last forever, but for now, this fixed it. Hopefully this helps someone. Thanks for an excellent video. Just the right level of detail to be of great help.
DELID ALL THE CPUS
Nope... We shouldn´t have to. When paying up to a 1000 dollars for a product, it should be expected to already come in perfect condition and with good paste or liquid metal. It should actually be soldered to the cpu itself like amd has done with their cpu´s.
@@datreja9573 We should get a decent product even if we pay a fucking 100$, especially every K and X cpu, why are we paying the premium for them then? 1st time I bought from intel and was already disappointed with their garbage 7700k toothpaste. Had to spend additional 9$ for some liquid metal and 2$ for some knife to delid it. After that I could finally have 5ghz with 75c and not 100+.
And how do u plan delid a Ryzen chip? The IHS is soldered on the CPU die.
You got scammed if it was changed and not how you gave him it. He lied and swapped stuff or switched cpu with his bad one for a good one.
Always mark the side of the ihs with a very small dot or something to tell that it's the original CPU.
yeah
@@jlrockafella or just record the serial number. It's right on the CPU.
Nah dont think he scammed or anything.I believe he just tried to unplug everything and plug them again. After watching Verge.
That video got pulled way long time ago. I think he sold person after as he said few months are in "1-4 months possibility".
I found this video completely by random and have been having an issue recently with my machine shutting down and up until watching this video i guessed it were my ram or motherboard. Turns out my i7 4970 needed delidding after all these years! Subscribing to you this video is a massive help!
Glad you mentioned it could be heat instead of something else TYC, I had an old FX 6100 and that thing would get dusty, so much so that after a couple months without cleaning that it would simply turn the whole system off without an error, just a black screen and everything powers off. It's important to maintain your system, and if your previously working CPU suddenly starts crashing, check those temps! General rule is:
The quantity of the tech yes lovin is going down and down.
Step it up Bryan we need the lovinz
I got something coming soon don't worry lol
@@techyescity oooooo
Goodie. :)
So the tech yes loving will intensify?
The office space my mom works at has been using HP computers with i7 4770s for over 5 years I think. It was supposed to be amazing, too of the line performance, but I’ve nicknamed those computers the “Disappointments”. If there is a dilemma with 4770s, that would partially explain why they all suck. They also have HP’s horrible motherboards, no cable management, tiny tiny AIO cooler, and mismatched ram (an 8gig stick with a 2gig stick).
So glad they are getting replacements and I’m more glad that I’m being commissioned to build them.
👍!👍!👍!👍! Awesome job, I appreciate you simultaneously showing & describing what you were doing when overclocking in thd last 1/4 (a quarter of) of the video.
Different story, but something was wrong with the PC of my mum. It kept crashing and never turned on when we hit the power button. We sended it five times to the merchant who sold us the pc. Every time he checked it, he said it was fine but never fixed anything. My mother was one of those people who threw everything away if something didnt work and would buy a new one, so I suspected the merchant never looked inside of the pc, which everyone should do, if a customer tells you something is wrong with the pc. We returned the pc again and he became salty because he had to check the pc again. Then I told him "Then you clearly never looked inside! Seriously! You cant do small tests and say everything works well." It was clear in his tests the pc was running fine, yet he never looked inside, simply because all what he did was to check if the pc is turning on. After telling him my mind, he did check the components. It turned out the PSU had a faulty, he replaced it and apologized for the mess. I asked him what kind of a technician he was and we didnt pay the last repair bills for all this mess. To that time I was 14 years old and didnt knew much about pcs.
Been in the computer industry for 30+ years and the only Intel CPU I had fail was a Pentium P5 90 MHz CPU in 1994. Had more problems with AMD CPUs, but not in along time. I have a 4770K in my current PC which has been running fine since I bought it new in 2015. Looking forward to the new 3rd gen Ryzen processors for my next upgrade.
THANKS! you saved me from throwing away my 7700.... I bought this second(?) hand cpu for an old mb and it worked fine for a couple of days, even benchmarks all below 85C until it started freezing dead before boot... it just froze either before post, during post, or after a few seconds at the bios screen when I managed to hit F2 in time... I swapped it for an 7100 i have laying around to modify the config to lower the ratio, fix the cpu core voltage at 1.125, and disable turbo-boost... then back to the 7700 and it FKIN WORKED :D... It's running hotter than usual but at least it's not an expensive paper weight now :D
I like this video, and conclusion you came to. I've had 2 3770k's overheat severely with even minor overclocking/over-volting. I ended up de-lidding the second one, applying liquid metal, and it now stays cool and stable at 4.7 ghz. I think de-lidding is almost mandatory at this point with these 3rd and 4th gen chips, if you want a decent, stable OC.
Bro what is de lidding can you please explain
"ex client" lol! Bryan never pulls any punches. Dude is always honest.
Great video, lots of information. I've fixed lots of computer processors, but that was when they were still made from desecrate components. I had to troubleshoot and repair them to component level, replacing capacitors or resistors or IC's or transistors or diodes. Now that was before micro processors like today's, these were on one to six circuit board cards connected by via a BUS. A lot different than what you are doing, but at that time it was the in thing to do, lol...
I have an i7-920. Sometimes in summer it started to have weird crashes. Turned out it was overheating to 99°C. I completely de-dusted everything inside the computer and it has been running smoothly since that. It is easy to overlook heating problem possibilities until you face one first time.
A computer should be cleaned and thermal paste changed about every third year I would say. Depends of course on the environment. There is software like HWinfo64 that can help you keep track of your computer and thermals. Thermal Paste always gets hard with time. That and dust would probably be one of the two major reasons why laptops tend to die rather soon (3-6 years).
I've never had a CPU go bad. In my experience, it's usually bent motherboard pins. That can be fixed as well with a steady hand and lots of patience.
I had a Ryzen 7 2700X that was broken. What a pain that was.
Recently I did a build with a 4770K. OC'ed to 4.4GHz on a Corsair H110, it maxed out at 79c during Aida64 (after some minutes). I was pretty sure it's been delidded by previous owner (Sometimes you get pleasant surprises when hustling 2nd hand hardware. :)
Another way is to see how quickly the temp rises after the stress test startup. If it instant goes to max temp, is the connection between die and IHS is the limiting factor.
I had that once, where I replaced every single component of a computer. In the end I even replaced the case. It would run for a week, even in my office, then just hang. Of course the customer just assumed I was a lousy computer technician. It was very hard to convince them that this was an incredibly unusual situation.
Cool video. It is a two for one. You also show how to destroy memory by installing it while power is running through the motherboard as well as handling the memory stick on the memory chips rather than the edges of the DIMM as shown in this video.
My recent issue with 4th gen i7 4770 is it never gets heated up 😂😂😂28 to 30 c only. My system shuts down after 30 min Or an hour. Never starts after that. May start after a few hours. After watching your video, and changing CPU ratios it worked. On stress test the temperature was 70-c.b At idling 35-c. The system worked for 5 hours now, today. Thanks buddy
Don't know if anyone's pointed out but, next delid, use some nail varnish on the tiny resistors next to the die. The liquid metal can migrate and short them, which is not good.
Other than that, great little 'how to' Brian. [thumbsup]
I've been building and repairing computers since 2008 and I've only come across 1 bad CPU, which was a dying Core i7-2700k. It needed extra voltage to boot into the system without crashing. One of the contacts (Lands) on the bottom of the 2700k was black, not gold. On some ebay listings, I've seen "Parts Only" processors that were missing capacitors on the bottom of the CPU. So it does pay to visually inspect the CPU.
I didn't need to run the P95 method because I had a working 2600k. In fact, I've used Prime95 Blend to diagnose DRAM problems. Sometimes with the DRAM and CPU stability problems, a thread on the Prime95 test will stop running. With the memory I can swap and add sticks in order to find the bad stick.
Another guy I know scrimped and saved to his funds to get a Core i7-6700k, which had a problem with one of the memory channels not working. This resulted in 2 of the memory slots not working. At least he was upgrading from a Core i5-6500 which recognized all 4 sticks, so his problem was easy to diagnose. He tried raising the VCCIO, VCCSA, and DRAM voltages but this didn't help him. He even tried reseating his cooler and everything was good.........no bent pins.
There was another time I put a mono-block on my system, and the uneven mounting pressure caused one of the DRAM channels to quit working properly. All I did here was I reseated the mono-block, and everything was fine after that. So it pays to make sure the cooler was mounted properly.
These days when I delid my processors, I use liquid electrical tape on the contacts that are next to the die. This will keep the liquid metal compound from shorting those contacts and killing the CPU. Some people like using clear fingernail polish here, but I like using the liquid electrical tape better.
My 6600k randomly freezes the entire computer (it had nothing to do with OC or C-States, if it didn't freeze PC, it would run any test). I have been looking for a problem for a long time, blaming RAM, motherboard, power supply and disks. Until now, I had no idea that the CPU has more states than two "works" or "doesn't work". Well, man learns his whole life;)
There is something perversely enjoyable about watching someone track down what is causing a computer issues, jay did a few of these videos a few months ago but sadly stopped.
I'm not sure why, but despite the exact same architecture and only 100mhz speed difference I've never come across a Haswell xeon that's had the same issues with temperature, even spite of the fact many where used in much higher duty cycle loads with smaller cases/coolers than there consumer breathern. Could it have something to do with crystallization of the TIM with heat cycles rather than actual drying, or perhaps that these CPU's never saw attempts at being overclocked? Or may there have been a difference in the binning/selection criteria? Either way, they are a pretty cheap/widely available part still, so if your buying used and your not getting a K skew, I'd compare prices.
You are onto something there, I noticed with some of the mining experiments I did back over a year ago, that some GPUs would be fine* to just keep on running, but take them down for a week then try booting them back up and you would have to change the thermal paste. Something to do with cycling from hot to cold?
umm I'm gonna say voltage is the answer? my E3-1246 v3 needs only 0.89v to maintain 3.5ghz and 1.01 for 3.9ghz "turbo" speed, the 4770 I got rid of needed 1.19v at 3.5ghz and 1.22 at 3.9ghz... both kinda won the silicone lottery, and tested on the same boards. the 4770 needed more volts, hence ran hotter than the Xeon at the same speed...
Haswell E (HEDT) and Haswell EP (Xeon) use solder TIM not thermal paste.
@@C-S-J True, however the Xeon E3-12xxv3 of the LGA1150 socket is in fact thermal paste, not solder, and are the directly comparable parts to the i7 4770 of the video referenced in the video... and yet doesn't seem to face the same age related overheating or instability as the consumer i5 and i7 chips.
This is why I'm speculating about either binning differences of the core itself or effects the different use cases have on the longevity of the paste. And the conventional logic that more use and hotter running would correlate to drier paste and higher temperatures later in life doesn't seem to hold up with my experiences on the platform.
Xeon's are better binned parts for Enterprise systems that need power efficiency and stability. So of course it makes sense when a Xeon counterpart from the same architecture runs cooler and with less voltage than their desktop counterparts. But yes, Haswell was a very hot architecture in general (even hotter than Ivy Bridge) compared to Sandy Bridge, and most overclockers were better off with the 2600K due to it's best performance without the hot temperatures since it was soldered. That may be why a lot of people never upgraded from their 2600K until Coffee Lake came around. This was fixed when Skylake came out with the 6700 being quite cool when you're not bumping up the voltage, just to be undone once the 7700K came out due to the high clock speeds and obviously the poor choice from Intel to use TIM on enthusiast parts. Ivy Bridge and Haswell overclocking ARE NOT the same procedure when compared to Sandy Bridge. The 22nm Tri-Gate transistors are much more sensitive to voltage than the 32nm transistors we saw with Sandy Bridge. This can prove to be a pain when overclocking since you may need to bump the voltage to reach higher clockspeeds, but that may cause the heat to ramp up with it. And we also know that voltage does cause a linear increase in heat either.
As for thermal cycling, that's a lot worse on a CPU and GPU than if you ran it constantly at the same temperature. Large temperature spikes can prove detrimental to hardware. It's why mining GPUs are usually actually in better condition than a GPU used for gaming because a GPU used for gaming will be thermal cycled A LOT between hot and cold temps. And there's miners that actually ramp the fans up and undervolt their GPUs also to keep low temps. Thermal cycling can cause PCBs to warp over time and solder to crack. In fact, the Xbox 360 had this issue where it would get very hot under load, cool off when not in use, and over time this caused the solder that held the GPU in the socket to crack and cause the GPU to stop working, and thus the red ring of death would occur.
i usually just use IPDT from intel and when it says it's good, i just trust them. but if something else goes wrong, it usually is very accurate what's a bad processor or not. takes five minutes.
Well, prime95 is nice, but usually takes a lot of time before you can encouter an error. What is somewhat better to test CPU/RAM is Intel burn test which uses linpack.
Usually it is enough to run 10 loops with maximum ram to find if the computer is stable or not :)
Filip Doležal just throw in call of duty 4 blackout run zombies it’s best stress test yet trust me got my i7 9700k at 5.2 ghz at 1.325 volts its amazing !!! Stays in 50-65 while gaming on just 120mm aio in push pull and fan over vrms!!!!
you can only truly test stability with a mix of stuff. start with the IPDT, then stuff like aida64, then intel burn test, then prime with FMA3, then real-world crunchers like seti@home.
i have found that everything would be fine, until seti@home (lunatics avx client) would show it to be unstable. if you are looking for an unconditionally stable oc, you need to have a large set of synthetics and some real world stuff.
my latest stability finder is Y-Cruncher (also does prime calculations) but seems to find errors faster than Prime95.
AVX generates more heat, and you may need to set up an AVX Offset in the bios, to downclock the CPU by -1 or -2 on the multiplier when AVX instructions get hammered like that.
@@mrlithium69 I think AVX offset was introduced later. I have the Gigabyte Z97 flagship board for my 4790K, the GA-Z97X-Gaming G1-WIFI-BK, and it doesn't offer that. I'm pretty sure that ASSRock didn't either. Not sure about Asus, as my first modern Asus board is a Strix Z370-I (prior to that, my previous Asus was a 286 board).
never in 25 years have I ever had a faulty or bad CPU. Bad ram Bad ram slots on the mobo bad mobo faulty caps, bad onboard NICs.
trust me its living hell ,lol
For testing CPU only use Small FFTs in Prime95 NOT the blend test, and use HWMonitor for temperature monitoring when Prime95 runs at the background,
Prime95 will give errors if the CPU is faulty or unstable overclocking.
For memory testing use memtest
you da MAN B....always cool stuff with your channel.
A faulty CPU rare? Nahhhh, at the store I work for, I even had clients, who had faulty power lines that caused problems and resulted in random shutdowns under load. Or a faulty case that also caused random shutdowns. Yes a case, not sure what it was, but it was gone after we changed the case (, after we had changed everything else. The client thankfully was a very understanding one).
Anyway, great video.
@【】_【】 Not sure what you mean by that? If you want to imply that I lied to the client, it was him who suggested the cause. It even was fixed after he installed a UPS.
I had the same problems with an Engineering sample cpu, it is running mostly fine now without changing much of the settings.
I did change out the motherboard battery and that increased the booting stability a bit.
Really enjoyed that. Analysis and repair.
The bad thing about these CPUs overheating is that over the long term will last shorter and will have more probabilities to die sooner. So that brings me to the next question: and how about the alarming high temperatures found in gaming laptops? These usually get to a heat level that is borderline the maxium temperature that these dies can handle.
My 4790 is on the warm side too, but since it's not giving problems and as you say it's still somewhat expensive, I'm not going to risk delidding. But maybe some day.
@@webbie7503 I have the same cpu. But it gets really hot. (50 - 55 C idle) (70 - 85 when gaming) I dont know if it is the cooler or the cpu whos causing the issue. The cooler isn't great i think its a spire sigor iv, but anyways i just wondered what cpu cooler you have i need tips :D
Oh my god i need that background now STAT!
I had a customer bring me a Dell optiplex 9020 they bought from the off lease section. The computer would not boot and the warranty had been 90 days and they were at 100 days. So they just told me to get them a used 9020 board and I5 to replace their I7 4790. At that time , nobody had listed an issue with the power switch. I still have not read the issue and now I know this issue. I still have the board and chip I pulled but I did discover that the power supply was blown. Since I did that as a favor to them, I no longer had any extra parts around for checking stuff. My guess was the power supply blew hard and took the board and chip.
There was an issue with the 8 pin power supply they used having a very high failure rate. Anyone else disagree with the assessment ? The power supply did not have any power and when their used same Dell 8 pin power supply came in, I plugged in the I7-4790 board and did not get any fans or lights or beeps. I was unable to check the CPU without old parts, The customer was using the computer daily so, I did not pull the I5 and take a chance on damaging their board in any way. So chance the I7-4790 is alive ????? I vote it is dead because of the definitely dead PSU and Board. PSU had the reputation of popping as it died with a high failure rate.
Good content Bryan thanks man. Never ran into this issue with my 4770 as it ran cold. Thanks for the tips tho 👍🏼🙏🏼
Over 2 years now I've gone through 3 sets of RAM, a 1070 and 2070 GPU, an SSD and NVME.2 storage for the OS, and reinstalled windows several times.
Yet my system still crashes semi-regularly. Sometimes daily. Sometimes weekly. Raaaaaarely I can get 3 weeks without a crash.
I'm confident I have either a bad CPU. Why? Because while installing my AIO cooler I was trying to squeeze it on suuuuuper tight with a LOT of pressure, before I realized I could loosen the screws to get the AIO mount on, with ease.
And you know how AMD's PGA sockets are.
Im definitely a budget shopper so going with a ZALMAN cooler i don't mind. Plus i always used them. I found the biggest one that supports up to 300watts and stuck that on my 8700K.
I got a Gigabyte Aorus Gaming 7 cause of it's high end VRM components and after watching Der8auer's video on it, I had to get it. Plus it looks good with the ZALMAN 9900MAX and a Sapphire Nitro+ RX 480 8GB OC (RX 580) with it's back plate almost perfectly matching the design on the board.
All in i built a z370 monster for only $670. I see PC's prebuilt going for over $1500 with similar specs. I can say im happy with the choice to go z370 and not the i9 9900K route. I might upgrade to that in the future being that the MB can handle it. (8700K $280, ZALMAN 9900max $20, Z370 Aorus Gaming 7 $150, 16GB Gskill 3200MHZ ram $100, 650watt Corsair PS with case $120) had the rest of the parts SSD's and HDD's. $670 is what i would've had in just the 9900K with a proper IO and MB.
10:15 that is because 1080p is too low for +22' monitors, 24' 1440p is the sweet spot. 4k is overkill and no sense
I've been building systems since the mid 90s and the only new CPU I ever had bad out of box was a Ryxen 5 1500x, it had the dreaded segfault bug. It took me months of diagnosis and fighting with AMD to get an RMA because the serial number on my CPU was not supposed to be an affected model.
Beware of Ryzen 1000 series parts floating around on the second hand market, you may well buy a defective CPU and be plagued by random inexplicable crashes.
12v, 5v, and 3.3v are way too high, PSU has some issues...
Nah, won't make any difference.
They're well within tollerances. The ATX spec gives a +/- 5% limit. 12.6v max, 5.25v max, 3.465v max.
Theres 10% wiggle room on ATX voltage specs. And only 12v is relevant here, which further gets converted both in the motherboard VRMs and in this case on the CPU FIVR.
And all the critical parts of the computer generate the required voltages locally (CPU, GPU, memory etc) so even 10% higher voltages won't hurt a bit
I've only ever seen one CPU failure, was a real pain to diagnose. it was an AMD A-series APU. The computer would just shut off randomly, ruled out heat related issues and swapped the power supply and nothing changed, it would just shut down a couple times a day. I pulled the GPU and used the onboard graphics and had artifacts even in the BIOS screen. Apparently the GPU portion of it had failed. Replaced it with the best CPU I could get for that socket which was I think a shitty Athlon X4, re-installed the discrete graphics card and everything was fine after that.
This was a pretty awesome video. I wish I lived near ya to buy a PC from you one day and chat about tech!
Good to see you use Prime95 too Brian since it puts a more demanding load on the CPU. I would not worry too much about Haswell CPUs. My Haswell is fine. My guess it was abused before you got it because my 4790k has been overclocked to 4.6ghz for years now and no thermal problems
Also, warm for 6AM? Wisconsin says screw you, the realfeel was -55C outside with wind chill just the other day
I've had a faulty cpu with that kind of freeze thing before. Also replaced everything before trying an alternative cpu. Very frustrating.
I’ve got the 4770K in my rig, had it for 5 years now and it’s still holding up just fine
And would probably in the next 10 years, if not more. If you got a cheap motherboard it will die sooner.
Koeras it’s a Lenovo motherboard it’s not that bad
@@themattyshow4026 Lenova does not manufacturer Z87 motherboards as far I know (?)
Koeras it’s a prebuilt Erazer X510
@@themattyshow4026 They use OEM parts of one the major manufacturers.
It would be cool to have a community forum or something like that... damn even a discord channel would be cool!
The only bad CPU I have ever gotten was a Ivy Bridge Xeon. Maybe, it was my board that could make it work but it was returned thankfully.
Wd40 gang
"lmao stupid sticky WD40 boys, dust will settle on your parts FOREVER!"
*this post was brought to you by Isopropyl Alcohol Gang*
even Bryan admitted that wd40 wasn't really good for pc parts because it's too oily xd
@@Brandon-uy1uv Lol i know, but down under we just call all types of spray wd40, its a weird thing. The aussie export, multipurpose stuff is the best thing going.
I had a problem with a CPU i7 6700 for months! It always had bluescreen with the WHEA uncorrectable error. We noticed it would bluescreen when idle or doing light tasks but never on load. Long story short we fixed it by putting voltage to 1.35V.
hey m8 i've a question for the 'Yes man' ^^ can you explain why you use prime95 over aida64 ?
Keep your awesome work, i really enjoy your happiness and that you always keep smiling !
always very helpful.I would have loved to see this vid before I burn my I7 3770.should have get it back where I got it
1:02 you made me double check that my GPU wasn't playing up!
in about 20 years, I have only ever known 2 CPUs die, one was an AMD K62-350, WAAAAY back.... I don't remember much about the cause, and another was more recent, it was during a re-build of an i5-2400, built into new case and it just stopped working, swapped everything out, even put everything on a new board and was still nothing - swapped out the CPU and it all came back to life...
I do however believe the CPU took one of the RAM channels out on the board, as only 3 of the 4 slots will work now...
But also recently (2017) I built a 7700k system, and I ended up returning it TWICE to Intel as temps were not under control, were hitting low-mid 90's in less than 2 minutes when gaming...
Did u get any post or beep with your dead cpu.....m currently in a bind
thx man, good benchmark aida64 i have a i7 4770 and temperature stays around 70-75 degrees for 5 minutes test
Great video on a lightly covered topic. Ive had a couple cpus appear faulty due to thermal paste getting on the contact pads, even a tiny amount that is barely visible can cause issues. Same thing happened to a video card as well
I had a i7 4770 that I delidded, I didn't seal the IHS to the CPU little board. I used Arctic MX4 both between CPU die and IHS and between IHS and cooler. It worked fine for about one year and a half.. until last week when it simply refused to boot showing nothing on the screen. I took it out, I re-placed it in socket, and now it works again. Before delidding I had 70 Celsius in idle.
i recently brought the i7 4770k and i can honestly say ive not had any issues at all.. mind you i upgraded mine originally from a i5 4460 cpu that id had for years.
i'm still running the i7 3770k, oc'ed to 4,7ghz and still running strong after 7 years... i was getting high 90° at 4,2Ghz and i delidded it immediatly after getting it, and now it's sitting in the low 80 at 4,7Ghz, glad i chosen the i7 instead of the fx8350 since the fx nowadays is really weak compared to the i7
Tough break, kinda sux you aren't certain if some of the hardware is what you actually supplied...
Q: Do you have any sweet spots for 3770 non-k cpu's?
I delidded mine a few days ago with great temp success running at 4.644ghz with 1.235 volts via 43x108 peaking at 66c under stress.
Anything I can do to improve, is that my lottery sweet spot?
Great vid as always, I know a dude that bought one of your builds and am always surprised by the bang for buck with whatever he demands of it, you sure know how to tune!
i mean it's a testament to Intel that we look everywhere else first before blaming the CPU. back in the 90s, Asus was like that for their MBs and Kingston ram had a lifetime warranty for a reason.
I have a i7 4770k and have no idea why the pc keeps just shutting off.
Sometimes the pc is just idleing and it shuts off. Just watch low stress content and boom off.
Thank you for letting me know about it's thermal grease. It's come down to the motherboard and CPU.
Retesting the ram one more time before I delid the cpu.
Other option would be to buy a motherboard first and see if that works fine. That would be sad though had an amazing formula VI motherboard.
Going to just buy new set of ram. Worst case scenario I end up with extra ram for my system.
Do let me know what would you recommend I do?. I have changed the power supply, tested both the rams, added a new cpu cooler with new grease, tested it on unraid still turns off, remove all hdds and tested it still turns off, removed most usb plugs, changed grease on gfx card which made it 12c cooler understress.
Ram test just turned off was at 92%.
Temps are generally really good on the cpu. Below 37 max 50 to 60.
Will try Prime95 tomorrow.
Wheres our daily tech yes lovin man? We need it
I had some boot-looping problems with my 4770k (and or mobo?) a couple years ago, and I still occasionally run into issues especially when overclocking. Without a doubt the most frustrating thing to fix. I could get it to boot with only 2 sticks of ram after clearing cmos but it would only boot once (it would start looping after shutting down or restarting), so I had to clear cmos after each shutdown (or oddly enough change which ram sticks I had inserted). I finally fixed it after messing around in the bios by setting all voltages from the default "auto" to "normal" with the exception of vcore and ram voltage which I set manually. After that it had no issues even with 4 sticks (it wasn't vcore or ram voltages that were the issue since those were the first I tried experimenting with). If I were to guess, there must be some sort of auto-voltage-calibration thing that's not working properly but I have no idea whether its a problem with the motherboard or cpu, but it must be a hardware issue since updating/switching bios didn't help. I've been wanting to sell my 4770k and upgrade to ryzen but I'm afraid someone will buy it and cant get it to work and I'll have a situation similar to yours. :/
did you change the bios battery anytime?
Not many people know this but if you place the faulty CPU at the large end of a celestial grade telescope you can spot the microscopic bug that is causing the CPU to fail. At this point a pair of surgical tweezers can be used to carefully extract it, at which point the CPU should work fine again.
i had a faulty i3-7100 on a cheap computer I built for someone. It would BSOD shortly after getting into windows. I replaced everything, motherboard, power supply, memory etc.I finally tried a 7700k cpu and it worked absolutely fine. The CPU was almost brand new so RMA through intel, they asked a bunch of questions and had me mail in to verify faulty. Sure enough it was and they sent me a brand new in the box cpu, even got another cpu fan.
Your return policy should be no returns. Or no returns on swapped parts
My first pc was a Q9550, I picked the parts but had the store build it. Once arrived it worked but not great and had higher temps and seemed to thermal throttle. In transit the cpu cooler must of been shaking because the thermal paste was cured with airgaps on all but the very center of the cpu. Cleaned it up and new paste and it was running normal. This system also had the OCV-Reaper 1066ddr2 that would cause issues on stock volatages or if placed in certain slot combinations with bootloop.
On a side note to whoever reads this comment: nowadays every cooler out there will come with some sort of pre-applied thermal compound which most of the time is $0.10/g crap (some coolers do come with decent stuff so do your research if you can). I recommend everyone to scrap that white stuff off, get yourself some decent thermal compound like Kryonaut or MX-5 (if you want more longevity but decreased performance over Kryonaut) and replace it, your temps will probably be about 5C lower. Plus it will probably last longer unless if you're constantly running 80C+. Kryonaut's break-down point is 80C, MX-5 *should* last longer in a high temp scenario but will not offer the same level of performance. There are more expensive solutions that might endure longer than Kryonaut in a high temp scenario but I personally think that by the time it starts losing effectiveness, your system will also need some cleaning, so it's not hard to also repaste at that point.
On GPUs I highly recommend to test it on day 1 and then straight up repasting with Kryonaut once you're sure it's working fine. You will get better performance after the repaste, specially on Nvidia GPUs with the way GPU Boost works.
In my case, I use a slim gaming laptop as my daily, my CPU (i7-8750H) runs 80C+ under load and these high temps lead to the thermal paste to break down/dry out faster than normal (I use Kryonaut). So I just replace the thermal paste whenever it's time to open my laptop to clean dust build up, which I probably do every 3 months or so. Every slim laptop I've seen out there running an 8750H will inevitably CPU thermal throttle at some point due to design limitations, so every degree counts if you want max performance and longevity.
My Ryzen 2700X was a faulty CPU. The impossible happened. Wasn't hard to identify it/ It was just unstable. System was freezing whatever the memory settings, whatever the voltage. I managed to keep it stable at 4.2 GHz and 1.34v (yes the chip didn't crash at around 1.3v in prime95: "golden chip") with 16GB of cl14 3200 memory. XFR2 and PBO2 didn't work at all. System hang up while booting. After a few weeks nothing worked. Even 3.7 GHz with 1.4v and cl18 2133. I tried different hardware. Wasn't the powersupply either. And my Crosshair VII went through RMA too, they sent it back, so by conclusion it wasn't the motherboard either.
Well RMA just gave me a new Ryzen 2700X. Been working fine in the last 6 months. And probably will just do fine in the next 5 to 10 years.
@13:57, we can clearly see a "gap" in the glue on the PCB.
May that be contributing to the thermal paste actually drying off ? i.e. the humidity just escaping ?
If so, what would be a good test we could come up with, without delidding, to test the glue between heat spreader and PCB ?
I once had an AMD system shutting down due to 'overheating'. I replaced the cooler a couple of times and the thermal paste a couple of times. Eventually, I became convinced that the motherboard 'thought' that the system was overheating and shut down. A new motherboard solved the problem.
I see a blast from my past with that mobo, seen the word RANGER and i got goosebumps. I had the A88X ROG Crossblade Ranger paired up with a A10 7870k Black Edition and x4 4gb Corsair DDR3 vengeance Pro Editions at 2400Mhz AND X2 ASUS STRIX GTX 970S BOTH WITH nzxt g12 bracket AND corsair H110I V2 280MM RADS AIO ON THEM, SADLY THE PC WAS LOST TO A FIRE IN MY FLAT, WAS MY VERY FIRST REAL BIG PC BUILD, CASE WAS THE HUGE PHANTEK ENTHOO PRIMO SE EDITION, NOW I JUST HAVE LOTS OF PICS TO LOOK AT, WOULD SERIOUSLY LOVE TO BE ABLE TO BUILD IT AGAIN BUT ,OMEY VERY TIGHT, SORRY PPL, FOR GO ON ONE, MEANT A LOT TO ME
I've had a broken 4790k for nearly a year now. I'm gonna try this.
Here’s what I’d tell that customer. Good luck finding a tech as knowledgeable and experienced as this guy who has half the business sense to get you an actual good deal on a nice system.
Generally it's a pretty good idea to run a Temperature Monitoring Program while doing this to make sure your not pushing the CPU Temperature outside of it's recommended operating limits because a overheating CPU can cause this exact problem........
Bought my 4770 when it launched, about year ago temps on it got horid, delided it and put on new thermal paste, didin't bother to seal heat spreader, temps are now great and cpu still going strong.
I too have had the same issue with a faulty 7700k. Also seen a few 4770/4670 and a few 6600k.
Just got off work, right on time for a video.
Is the i5 4690k in the same boat for the thermal paste?? Or did they fix it going from haswell to devils canyon?? Awesome video, had no idea something like this could be fixed relatively easy :)
My Devils Canyon is still fine after years of being overclocked
Yes, the TIM for that CPU can exhibit the same issues.
The only 'faulty cpu' encounter I've had was a i7 860, which had a faulty memory controller so half the motherboard ram slots wouldn't work (I tried a different cpu and the all of the slots did work, so I was sure it was the i7 860)
@Curtis Albrecht well in my case it was a motherboard with 2 slots, and yes actually only the slot closest to the cpu didn't work (so channel a)
I have an i5 760 with the same problem, and I saw a lot of lynnfield processor with a faulty memory controller
Wow that's bad. I think I have a faulty X399 board that acts the same. The only defect/weak CPU memory controllers I've seen are Ryzen (My Ryzen 1700 can only do 2666 reliably, though that's technically not defective since anything over 2133 is "overclocking", its still REALLY weak!).
My Threadripper 1950x does great with quad channel 3600 (plus some overclocking on top of that!), BUT.... If its cold, around freezing, (which sometimes it is in my room), it can't run full speeds. At roughly 0c, the Threadripper starts dropping memory connection, and 32GB turns into 24, and then 16... And then you need to drop it down to 2133 just to boot, then run some loads on it to heat it up so the rest of the memory detects. I'm 99% sure the motherboard is faulty based on how it just drops sticks of RAM completely when its cold. The lowering of stable memory speeds however is normal, and a known fact for Ryzen based processors acting up when cold (so don't run dry ice or phase cooling on them, but liquid nitrogen gets cold enough it starts working again for whatever reason). This memory issue was a huge pain in the butt when I was testing my LN2 cooler with regular ice (so the coldest it could possibly get was above 0c since there was melted water at the bottom of the cooler, still had these issues, so it can happen in the morning on any cooler). I'm very certain that its a bad board though, the CPU seems to bench fine at 4.15 Ghz on a 120mm AIO and 3600+ Mhz quad channel RAM most of the time, it only seems to crash when its doing nothing in the desktop (or 4x Handbrake encodes at once, but that could be a software conflict, though I normally run 2x at once...)
I have a 4770 and I'm now wondering if my CPU has gone bad. Temps are fine, usually 30's when idle, 50-60c when gaming, but 2 times in the last 3 months my PC has done a slow freeze and lock up. When this happened, I was playing a game on one screen with Chrome open on the other with task manager and Afterburner background. One time the game froze 1st, then Chrome. Other time it was reversed. After those 2 froze, the task bar would slowly become unresponsive. Afterburner freezes. Any other programs open would also slowly freeze. Task manager is always the last to freeze, but clicking End Task on frozen programs does nothing. Then it's frozen too. After about 5min or so, the mouse and keyboard stop responding. Only 3 things I haven't swapped out is CPU, main SSD and MB.
I have an i7 4770 and it still works fine.
JOERIcorn its not a specific type of cpu that doesnt work. Its a rare thing and it can be any cpu
Ayyy Mr yes, I got a question for you m8. I own a delided 3770k (LM) oc at 4.5ghz with 1.31v. Temps reach 89c with aida64 fpu. Is it a problem or not?
Doesn't the Machine Check handler get tripped when a fault is encountered inside the CPU?