Fixing Will's Problematic PC (Troubleshooting Tips)
Vložit
- čas přidán 8. 06. 2024
- Will is having problems with his 13900K gaming rig, so Adam helps him troubleshoot the problem (and provides tips along the way).
Oodle statement: www.radgametools.com/oodleint...
Watch this next: • Debate: Does Windows S...
Subscribe to our podcast The Full Nerd: • The Full Nerd
=============
Follow PCWorld:
Website: www.pcworld.com
X: www.x.com/pcworld
CZcams: czcams.com/users/pcworld?sub_c...
=============
This video is NOT sponsored. Some links may contain affiliate links, which means if you buy something PCWorld will receive a small commission.
Timecodes:
00:00 - Intro
01:10 - The Hardware & The Problem
03:59 - Potential Software Fixes
05:30 - Hardware Swapping Plan
07:09 - Default State
08:56 - DRAM Swap
14:41 - PSU Swap
19:17 - CPU Swap
25:24 - Day One Recap
27:57 - One Week Later
28:47 - Troubleshooting Tips
#pc #pcgaming #cpu - Věda a technologie
"Previously it's been running at 1.6V" well there's your answer.
The fact that this all starts from basically the factory overclock of the CPU even when running at Intel's specs is very interesting and also troubling.
it's not running at intel's specs, it's running at the mobo's specs
16:04 changing psu and not changing the cables can fry components. Gamers nexus did a video about that just yesterday. Evga messed it up, saying to the customer he could just swap the psu. Don't be lazy and always change the cables!
To be fair, that used to be good advice. It's just EVGA, completely dropped the ball.
Yeah, at least for companies that don't manufacture the PSUs themselves that is solid advice.
I would make an exception here as they pointed out. Same make and model just higher wattage. Had they switched brand entirely then I would agree with this
@@kevincampbell989 IIRC that was the problem with EVGA. Same wattage and model but the pin-outs changed between early and later batches.
@@jihadjoe The issue with EVGA was that they didn't communicate the changes to the customer - support dropped the ball or they weren't informed either.
On the topic of memory errors: Don't just try individual sticks, also try the individual memory channels or even slots if you have more than one per channel. That way you can really isolate the problem.
And memtest86+ booted from usb, run for 8hr. Throw some OCCT testing in there as well
The worst part of this hobby is the more you spend the more complicated the build gets, and that just makes diagnosing things more difficult.
Definitely. That itch to chase frames is an unhealthy curse. I've sort of learned over the years to just be happy with anything around 1080p/60fps.
I've had zero issues with my new computer. The secret, wait 1 year after launch for all BIOS issues to be fixed, so I just now got the AM5 7959X3D Cpu, then buy high end corsair dominator titanium ram, which is high quality and hand tested. There you go, zero issues, oh and no unstable overclocks
Asus MCE and ai overclocking probably cooked that cpu.
Yup, turn off Multi-core Enhancements in the BIOS.
@@Hellwalker855 Yup and Asus are not the only ones doing that. Gigabyte and MSI also have similar default settings.
I don't understand how they are looking at these frequencies, voltages and temperatures while at a gaming load and seem to think everything is fine
These troubleshooting videos are always helpful!
Best practice for stability issues with an Intel machine with 14700K / 14900K: Go into BIOS and turn in one step higher LLC (Load Line Calibration). If that doesn't help, put it back do default but raise the VCore with an + Adaptive Offset. Start with 0.02V. That should help to fix this Problem, if it's CPU related. Some Chips are poor quality (silicon lottery) and need more Vcore than they get from the mainboard by default. If if's memory related: Try the Software "Karhu" with enabled Option "CPU Cache". Best I found for finding DDR5 stability issues. Sometimes the XMP doesn't work for your CPU because the needed VDDQ, VPP, SA and/or IMC Voltages don't match with the XMP profile.
Also those cpu temps while gaming are way too high. If it was doing a multi core benchmark 100c would make sense but definitely not normal during gaming. Those voltages are way too high. The mobo probably killed the last cpu and its gonna kill that one too if its not tamed.
My thoughts exactly. 100*c in games is huge red flag.
Some P-Cores jump to 6Ghz and so the Voltage goes crazy up to 1.5V (sometimes even higher). If you lost the silicon lottery, that voltage/temp in Gaming on P-Cores is not so nice if it hits 96-100°C for a split second or so but it can happen if the game is using just a few cores. That's my experience with a good binned 13900K @ 6Ghz (up to 1,49V) and a poor binned 14900K@6Ghz (up tp 1.57V), both on the same mainboard.
@@BarisYener This was not a split second, The cpu was at 100*C the entire game footage. Clearly a issue with cooler/ mounting or bios settings.
Intel's specs for the 13900k has the Tjunc as 100c, so I have no clue why temp wasn't considered as a major indicator here. I could imagine that running 100c consistently for a few months ain't gonna do the CPU any favors right? I mean, normal driving in a stock vehicle shouldn't overheat your engine to maximum operating temp within 5 minutes of hitting the road... and if it did, you have some huge issues that would manifest very quickly
Yup. And since he's running SVID at Intel Fail Safe for no good reason on a new CPU, the new one will also degrade like the first one.
You should always validate a new RAM configuration with software like Memtest86 before booting Windows. If a memory stick is flaky it can corrupt the OS.
Memtest86+ is great for troubleshooting generally. It's the first thing I try if there's any system instability, blue screens etc. That way you take the OS and drivers etc out of the equation.
I can tell that you’ve been pushing your CPU too hard with the 1.6V voltage. No CPU should be given such a high voltage! My 13900K, running at its stock 5.5 GHz, only draws 1.28 - 1.312V, and I’ve had it for over a year now without any issues.
1.18 here
Does this bios even do 1.6v?
That seems like alot.
Ive seen 1.6v ONLY in "extreme" power overclocking profiles with a maximum of 1.612v And is usually the max package voltage and is usually only used to prevent power spike bottleneck when adjusting per core voltages.
@@Mr.No.Tv.Community it's very hard to see it, but the 1.6v is at time 24:45.
I'm glad Wil & Adam were able to troubleshoot and fix the problem. I have a similar system (13900k & rtx 4090) so it was interesting to see this diagnostic process.
What is plugged in next to the ram on Will's computer? At first I thought it was 4 sticks of ram but it looks like it's a dual stick kit and something else, what's that other thing?
dust buildup easy to gloss over but cleaning every so often a must, especially if case doesnt have great dust filtration
I'm just here to hear Will and Adam beat each other up 😂
the issue is with PSU it doesn't cover those max watts for the total consumption
get the ekwb 6nm torque wrench/driver. perfect for many installs.
i have another approach on the Unreal thingy and has to do with games and cores amount and how those core are use by games. (my own therory) what if turn off the cores to match other cpus as the game maker say as example? and i have this on 13900k 14900k.
Im interested in that little fan above the ram. Does it make a noticeable difference to memory temps?
I have a ram fan like that and yes it lowers the ram temps a lot, about 10 degrees.
I have a different problem it's a little bit off topic. I have a windows 10 home PC Some of the windows updates are deposited in a protected Recovery partition When this partition gets full no more of this type of windows updates can be installed and because this is a protected partition I don't know how to expand the partition. Is there a way to fix this problem without just getting a bigger boot drive? If there is I would greatly appreciate a step by step instruction video I am fairly computer literate but not so much when it comes to working with partitions especially protected partitions. Also I would like to know is there a way to keep this from happening again in the future like maybe having these updates going to a different location. Is this an common problem or is it just me.
I've never had to reinstall Windows for anything. I ran Windows 7 from launch till 10 came out and just upgraded the same installation on the same SSD from 2010 and it ran fine. Hardware changed maybe twice in that time.
Im curious why the motherboard wasn't ever suspected as the culprit? They can and do fail. Swapping out the cpu might be a temp fix. If the issue comes back look at the mobo. Hopefully it stays fixed.
Pro tip to PCWorld to determine the right time to hit DEL to get into the bios (instead of slamming it all the time) 😁
As with the RAM RGB is your friend here - the moment your keyboard RGB lights up - press DEL - voila you are in UEFI. 😉
I know Will doesn't have a 360 AIO, but 100C on a KS part when CPU utlization was under 10% in a game like Fortnite is amazing.
could it be the cpu needed reseating?
I would put the original cpu back to verify that.
Usually bad seating equals bent pins lol
Maybe bad memory controller on the cpu or bad pin contact? Should have have warranty on it still.
16:00 Well . . . I shouldn't just blindly swap PSUs this way, a customer had an EVGA power supply, he did an RMA, EVGA said to just ship the PSU itself and leave all the wires how they are because it is a direct replacement, same model, same wattage, same everything! He got his RMA unit, he installed it in his pc, and would you know, the guy lost 22tb or something of storage, all his ssds and hdds were fried! Never assume that psus ALWAYS have the same rails . . .
Great video, but I have to ask why you didn’t swap the old CPU back in to see if you could replicate the problem. There are CPU related issues that are not caused by a “bad” CPU. These include marginal package seating causing thermal or mechanically intermittent contact on one or more pins of the socket or thermal issues due to bad paste or cooler install.
There is always more you could do. In this case that would have meant two additional CPU swaps including cleaning and repasting, plus testing, so another 30 minutes to an hour only to confirm what is already very VERY likely.
So after we wrapped up filming we threw the old CPU into a fresh new system (still had the same problem) and are in the process of doing some tests on it and his original system for a potential follow up video.
-Adam
Intel failsafe cranks up voltage increasing heat. Use to get crashes like this in games .I’m running 14700kf on Ak620 air cooler. ASUS Tuf gaming z790 . and 6000 Corsair ram .I enforced all limits in bios .manually set wattage limit to 253wat for long and short term. XMP 2 to use the rams timings without asus micro optimizations. Using typical scenario power profile I’m bios (not failsafe). I downclocked my p cores to 5.3 all core and left the E cores alone at 4.3 . Then undervolted by .050 . My goal was to tame it with air cooler. 5.3P/4.3Ecore all core all the time 1.25/1.26 under load with mid 80s under heavy core usage gaming normally 60-70 most games and take a 10-15 min Cinbench to get in 90s with zero throttle and margin of error scores over stock . Could probably do better but considering I’m on air I’m happy with performance , temps and stability now :)
Over 1.4v on this cpu is way too high, even with OC
Not sure wht default
voltages are so high...
CPU degradation
Can you guys get the cameraman a tripod for the camera. I was getting sea sick during the video haha very good troubleshooting steps video. Keep it up guys.
*monopod. Does 99% the work but only needs 1% the floor space. Great for places that aren't primarily built as shooting locations.
I noticed he mentioned he updated:
Intel Management Engine.
My question is WHY?
Also does he know what Intel ME is?
Also is Intel ME even in use and if not Why is it enabled.
Most companies dont use intel ME for system monitoring and management and if its not in use it should be disabled.
Intel management engine is Out of band management similar to lights out capability on servers and provides low level functionality over remote access that can be compromised if it is not in use and not configured and not specifically disabled.
Replying to myself to add a side-note: Also why are system integrators and OEMs still shipping with intel ME ENABLED by default when its not widely used and adopted and It is a target for exploitation by threat actors?
I've personally ran into problems with hardware when ME wasn't updated (live on stream in fact). Most of the time those updates seem to be packaged with BIOS updates, but it's something I always check when having CPU related problems (update, update, update).
-Adam
What is the ILM, what does it mean?
It stands for integrated lever mechanism and basically it's what holds the CPU in place.
-Adam
I had an issue with Windows corrupting files and random crashing with my 13900k. Found out Windows was only using 16 of the 24 cores after a Windows update. This fix is under System Configuration > Boot Advanced Options > uncheck box with number of cores.
Can 13900k be retired to Intel as defective?
1.6V default that is your issue
CPU
my 13900K is at something like 35-45°C playing fortnite, they dont know enough to know what the spec settings are in the bios to set them, also they have been messing with settings in the bios that they have no idea what they do or how they work, they need to reset to defaults, those voltages are so far out of spec its crazy and for sure will degrade the cpu I just checked and when playing fortnite my VR VOUT was from 1.27v to 1.38v my 13900k is garbage bin as well
Wow. It has been many, many years since a CPU has been the culprit and affecting only specific SKUs. Amazing Oodle caught this and was able to pinpoint the issue. Instead of fearing backlash, they presented their test cases, solutions and looks like your test shows merit to their theory. Meanwhile, shame on Intel here: these i9 K-series CPUs are extremely expensive and shouldn't have any systemic failures like these.
Intel is flying too close to the sun and they just don't seem to learn.
It's simultaneously Intel's issue and not Intel's issue. They have a sane spec and PL1/2 that doesn't degrade the cpu rapidly, but board manufacturers don't follow it. The problem is that Intel allows board manufacturers to do whatever they decide since their spec is more of a recommendation than a requirement. Usually most raptorlake cpus get a 100mhz increase from spec and unlimited PL1/2 on most boards. EX: 14900K spec is 5.6ghz all core and 253w PL1/2, but most boards set 5.7 all core and 4095w PL1/2
It's basically the equivalent of AMD just allowing board manufacturers to have custom PBO settings by default if they desire.
@@nepnep6894 Did you watch the video? You seemed to have missed that they already ruled out ASUS' power draw shenanigans. See 5:13
It wasn't too-high PL1 / PL2 / MCE enabled. The CPU was crashing under Intel's "sane spec" with a stock PL1, stock PL2, MCE disabled, SVID safety on.
That's what is quite unusual here: even following *Intel's spec*, his i9-13900K was crashing.
I saw some messages that those 13 gen Intel CPUs can degrade, so after some time (especially if running lots of benchmarks) CPUs unable to hold stock frequencies even using stock voltages.
Nope it is probably E-Cores fault try affinity when gaming to P-Cores only and leave E-Cores to system management and background apps E-Cores are shit cores
at least this fixed my crashing / sttuter issues with my 13700k
I dislike DDR5.
Issues with every PC i've used with DDR5 other than OEM laptops from a 12600k and 13900k, to a 7950X with 192GB of 4800, and an 8700G with 48GB of 6400(but works great with 8000).
I have two kits of 6400 that were super unstable on my 8700G, so i tried my OC kit of 8000C38 and it works perfectly no tuning required whatsoever just enable XMP and go.
What seems to have been the problem is that my 8700G was trying to run FCLK in 1:1 while at 6400(FCLK/MCLK/UCLK at 3200)
When i used the 8000 kit the MCLK stayed in gear1 jumping up to 4000Mhz/8000MT/s but the FCLK and UCLK dropped to 2:1 at 2000. Much more stable, but also much more performance
What may have been happening is that FCLK/UCLK were switching back and forth from gear1 and gear2, or possibly wasnt running in a clean ratio, say 2400Mhz for 3200Mhz/6400MT/s RAM
MOBO manufacturers wanting to get intel even worse in the eyes of consumers.
OC...
BIOS update helped me with my RAM to be stable on AM5 with EXPO.
1 and 3 or 2 and 4. Performance left on the table. Unless it’s a 2 dim and my old eyes are misleading me.
Yep, it looks to be a Mini ATX with only two ram slots.
It's a 2 dimm board
It's a 2 DIMM board: rog.asus.com/us/motherboards/rog-maximus/rog-maximus-z790-apex-model/
-Adam
It looks like this board has a weird expansion card for an SSD that looks like a DIMM, that's what you're seeing beside the 2 memory slots
nic
I would have also checked the p1 and p2 state.
1.6v lmao 😂
I've had zero issues with my new computer. The secret, wait 1 year after launch for all BIOS issues to be fixed, so I just now got the AM5 7959X3D Cpu, then buy high end corsair dominator titanium ram, which is high quality and hand tested. There you go, zero issues, oh and no unstable overclocks
Running XMP needs tweaking of VDDQ / VCCSA its not just plug and play, Do you know how many hundreds of hours of karhu it took to get my 2x32GB DDR 4 100% Stable at 4133 GEAR1 CL1 with all sub timings tightened, to test just run the ram at 4800 and see if it's fixed, then if it is you have to fix VDDQ/VCCSA for 6000 then test with Karhu.. not fortnite lol Every CPU needs different voltages for the IMC for X Memory Frequency all you have done by swapping the CPU is got a better IMC that runs the ram at 6000 at whatever the asus motherboard is throwing at VDDQ/VCCSA it may still be unstable at 6000 with the 13900KS but just fortnite stable..
Should get better cpu than doing over clocking.
Cpu above 80 degrees core temp is probably what killed it. Heat is a pc killer.
There are too many issues with this specific build to comment, but from multiple outlets, and what we've been seeing "out in the wild" Intel and AMD have tightened their binning process a little too far for profit. Plainly, there are too many examples of high end designated chips that should have been binned down. Makes me personally want to avoid SKUs like the 14900k, etc. Maybe that's why they launched the 14900KS, at least in theory it will perform as a 14900K should as specified.
What are you doing? The AIO is pulling all of the 13900K heat out of the CPU and blowing back into the case. Put the AIO on top and exhaust the heat out of the case. What a rookie mistake for PCWorld.
unbelievable. that was the one reason to buy Intel.
put a slightly bigger than rice size drop, then use a piece of thick paper and use it to spread it evenly, look at how it is on a AIO, Nice thin layer. your cringe and bad advice, just start replacing everything! wow!
Real note. Now Intel is the desperate one, so they take more risk with the chips for performance sake. Higher clocks, that could degrade/push over the edge lower quality silicon. We don't hear issues with lower end CPUs like the 13700k just the hardest pushed ones.
When AMD was in the losing position we saw the same thing. Bulldozer CPUs, first few gens of Ryzen. All had issues with stability or heat. AMD is in the lead now, so they take less risks with products, making them more stable.
Buy a 7800x3d lol
Reinstall os takes 3-5 minutes
Why are you goobers running >200 watts in Fortnite? You complain about your computer crashing while you are overclocking the CPU? The stock psustained power limit somewhere 125 watts. Just set that for PL! and PL2 and move on. I really find it irritating that being overclocked has become the normalized.
What I want to know is who is paying for the replacement CPUs? If you are buring CPUs then you are incompetent at buidling a computer. There is something seriously wrong with your computer.