PC reboots with Cinebench, Premiere Pro & spider-man remastered

[DSP]

Hi everyone, I had previously posted about a similar issue but I didn't want to revive a dead post which at the time I thought I had resolved. Apologies in advance, this is going to be a long and messy post, please do let me know if anything needs clarifying!

The problem
I have been using my build (specs below) for quite some time now with no issues, however, recently I started doing some editing in Premiere Pro and my PC started to reboot whenever I was rendering or playing back rendered video. This reminded me of my previous issue where I could not run Cinebench multi-core tests - as soon as I hit start, the PC would reboot. No BSOD or error messages. I previously resolved the Cinebench issue by updating my chipset drivers. When I had the issue with Premiere Pro the other day, I decided to try Cinebench again, and lo and behold, the same issue is back. I also started to notice this issue in Spider-Man Remastered. I cannot not play the game because of it. As soon as I get past the first cut-scene of the game, bam, reboot with no BSOD.

I have checked Event Viewer on each occasion, and all I can find is a critical Kernel-Power event ID 41 task category (63).

All other games and high and low load activities except the above, I never have this issue. I was running a curve optimizer with negative 5 on all cores, +100MHz CPU boost override and PBO limits on Auto when I first encountered the issue. So I figured the problem may be these settings, so resetting BIOS to defaults was one of my first go-to's. I also kept a check on temps and everything seems within "normal" ranges for Zen 3 as far as I can tell. Usually ranging from 50C to 80C under max load, depending on whether I have custom BIOS settings or not of course. Including my specs, what I've tried so far and some findings/things to consider below:

Specs

Mobo: MSI B450M Mortar
CPU: Ryzen 9 5900x
CPU Cooler: NZXT Kraken X62 AIO Water Cooling Unit - 280mm rad (behind the rad in pull config)
GPU: MSI RTX 3090 Ventus 3X 24GB GDDR6X
RAM: Kingston Fury Renegade DDR4 3600MHz CL16 32GB Kit (2x16GB) - XMP Profile 1
PSU: Corsair RM 850 W 80+ Gold Fully Modular ATX PSU initially (note - I have now upgraded to Corsair RM1000x as I was convinced my issue was with my PSU, but that did not resolve the issue)
Storage: WD S3750 1TB NVMe SSD with Heatsink
Case: NZXT H400i mATX

What I've Tried So Far
I will outline further below the scenarios in which I was able to run Cinebench either fully or for a few minutes, but first, a list of all of the different things I have tried.

• Checking Event Viewer for clues (critical Kernel-Power event ID 41, TC 63)
• Removing undervolt/OCs, resetting BIOS settings to default
• Downloading the Cinebench app from the Microsoft Store as opposed to from the Maxon site
• Closing all other apps when running Cinebench, Premiere Pro etc.
• Disabling antivirus etc.
• Trying to watch for signals using HWInfo - what happens just before the PC reboots? [I can't see anything unusual - too fast!]
• Adjusting fan curves in NZXT CAM
• Updating BIOS to the latest version
• Uninstalling old chipset drivers and installing new chipset drivers
• Testing RAM for errors with memtest86 - no errors found on 2 passes
• Disabling memory fast boot
• Disabling fast startup in Windows
• Changing PSU from RM850 to RM1000x (most of my tests were run on the new RM1000x)
• Testing with PBO on Auto, PBO disabled etc.
• Testing with custom PBO limits such as 142 PPT, 90 TDC, 140 EDC / 185 PPT, 125 TDC, 170 EDC
• Disabling Global C-State Control
• Testing settings like Power Supply Idle Control on auto, low and typical current
• Trying to lower SoC voltage from 1100 (default) to 1000
• Lowering both VDDG voltages to 0.950v and VDDP to 0.9v
• Testing load-line calibration on a variety of settings including auto, mode 1, mode 2, mode 3 etc.
• Testing with RAM on XMP profile 1, 2 and default (2400MHz)
• Testing different Ryzen Master recommendations (tried auto overclock, PBO enabled + curve optimizer for -29 on all cores etc).

I've tried so many different settings that I have probably missed a fair bit above, and I didn't take note of all outcomes either. However, in a few situations (see below), I was able to run Cinebench fully and render/watch video playback in Premiere Pro, and even play a little Spider-Man Remastered, but I re-encountered the crashes later. These did seem kind of random though. It feels like either something in my BIOS settings or perhaps drivers is causing some issue? It's driving me insane at this point so I'd really appreciate any thoughts you might have.

What seemed to help (but not resolve) the issue

• Default BIOS settings, PBO disabled, no XMP profile on RAM (2400MHz default) - Cinebench completed 1 full 10 minute multi core test, I started another one for good measure and 2 minutes in I again got a crash but this was better than not being able to run at all
• PBO enabled in BIOS with custom PBO limits: PPT 160, TDC 115, EDC 140 and curve optimizer on -28 for all cores and a +50 MHz boost clock override , RAM on XMP profile 1 -> ran Cinebench fully once, but Premiere Pro still crashed

I'm sorry for this mess of a post and I realise all this info is probably muddying the waters more than it's helping but I'd really appreciate your help folks! TIA

[Jon-Slow]

I would suggest you reset the BIOS to default again, bring your voltage down safely below the default, then bring down your core clocks down to the lowest possible. Do a test. And if it didn't reboot, then repeat the process and increase your clock and voltage little by little until you find your red line. I would suggest a Full Cinebench r23 for the first run and a 2min run for any following test. No need to run full r23 for every single time. For the final test where you think the red line is, do 2 full tests at least. then do a full 1 hour or more of OCCT.

 

These stuff are usually easier and more straighforward with Intel CPUs for me. But should be similar in results with yours. Chances are your processor is unfortunetly just a bad chip and you've got unlucky. So if you can find that red line, then you can stay below it. If you can't find that line then your issue runs deeper.

 

Worth mentioning that this could also be caused by your motherboard failing to correctly deliver to your processor.

 

Maybe someone else could help you better if they have a similar experience with the same board+cpu.

[Jon-Slow]

Also what I've just thought of right now, Your cooler might not be seated correctly on your processor too. Just check for that as well. shot in the dark, but might be the cause.

[DSP]

1 hour ago, Jon-Slow said:

I would suggest you reset the BIOS to default again, bring your voltage down safely below the default, then bring down your core clocks down to the lowest possible. Do a test. And if it didn't reboot, then repeat the process and increase your clock and voltage little by little until you find your red line.

Thanks for the suggestions! Can I just ask which setting you're referring to in the BIOS? Is it just the CPU core voltage? I'm still relatively new to a lot of these settings, but my understanding was that curve optimiser does something similar by basically reducing the voltage under load. That being said, the default voltage for the CPU when idling remains steady over 1.4v from what I can see, so I also wonder if this issue is somehow caused by the fluctuations in voltage. But that's a completely uneducated guess so I am sure I'll be corrected lol. 

 

1 hour ago, Jon-Slow said:

Chances are your processor is unfortunetly just a bad chip and you've got unlucky. So if you can find that red line, then you can stay below it. If you can't find that line then your issue runs deeper.

 

Worth mentioning that this could also be caused by your motherboard failing to correctly deliver to your processor.

I should say that this is the second 5900x that I got in my system - the first one I had to return due to similar issues with a little more instability, so I do lean towards what you've mentioned in that this may be something to do with the motherboard and the CPU's compatibility. They are supposed to be compatible with the latest BIOS, but perhaps it's not a great match?

 

1 hour ago, Jon-Slow said:

Also what I've just thought of right now, Your cooler might not be seated correctly on your processor too. Just check for that as well. shot in the dark, but might be the cause.

Definitely a shot in the dark, but I'll check this too and report back. Thanks again

[Beerzerker]

1 hour ago, DSP said:

Thanks for the suggestions! Can I just ask which setting you're referring to in the BIOS? Is it just the CPU core voltage? I'm still relatively new to a lot of these settings, but my understanding was that curve optimiser does something similar by basically reducing the voltage under load. That being said, the default voltage for the CPU when idling remains steady over 1.4v from what I can see, so I also wonder if this issue is somehow caused by the fluctuations in voltage. But that's a completely uneducated guess so I am sure I'll be corrected lol. 

 

I should say that this is the second 5900x that I got in my system - the first one I had to return due to similar issues with a little more instability, so I do lean towards what you've mentioned in that this may be something to do with the motherboard and the CPU's compatibility. They are supposed to be compatible with the latest BIOS, but perhaps it's not a great match?

 

Definitely a shot in the dark, but I'll check this too and report back. Thanks again

For stock speeds and Ryzen, 1.4v's is way too high and I'd have to say it's heating up under load.

I say that because I can run any Ryzen chip I have (Including my 3950x) at or even a bit beyond stock with a little as 1.27v's without issue and some can go lower without issue as well related to voltage used at stock speeds. 

Undervolting itself won't cause alot of problems if you don't overdo it (Go too far down) so what I can suggest is this:

Go into the BIOS and manually set the chip for stock speed and CPU voltage about 1.27v's.
That means stuff like "Offset" for CPU voltage aren't used because if you don't understand what the system is doing when you use offset or you don't figure up the final voltage correctly when using offset, you'll most likely have problems.

Next, do a few CB runs and so on, monitor temps and see where they wind up under load, not at idle using a monitoring utility to see what temps it's reaching.

The way to know if more or less CPU voltage is needed is simple - If it starts crashing and then it does so more quickly each time you try it back off the voltage a little and retest.

If it's more or less taking the same amount of time to crash, maybe a small bump UP in voltage will do but note a bump (Incremental adjustment) means just that - A bump, not a jump in voltage.
If you see higher temps regardless that means to check your thermal paste (TIM) for proper amount/application and make damned sure you didn't leave the plastic film on the bottom of the cooling block by accident when you check it.

I do know though 1.4v's is just too much, even if on water in some cases for Ryzen chips and you're running a larger cored 5900x.

[DSP]

Thanks for the reply @Beerzerker!

 

So I'm back after a bunch of testing, and it's kinda strange. I'm pretty sure I tested with BIOS defaults as well as with and without PBO enabled etc before, but I guess there must have been some mystery setting that wasn't cleared or something because I seem to have pinpointed the issue. Or maybe given the fact that I tried so many settings, I might have missed something at some point.

 

Basically, it seems that when I set PBO to Advanced in BIOS, regardless of whether I enable curve optimizer or a boost overclock or not, I get the reboots. For instance, setting PBO to Advanced, and having everything on auto, alongside curve optimizer disabled and boost overclock disabled (know this is pointless but did for testing purposes) gives me the reboot. Curve optimizer negative values of even 1 =  reboot. No curve optimizer but boost of 50mhz for instance, also reboot. PBO enabled as opposed to Advanced = no reboot! 

 

That being said, 

8 hours ago, Beerzerker said:

Go into the BIOS and manually set the chip for stock speed and CPU voltage about 1.27v's.
That means stuff like "Offset" for CPU voltage aren't used because if you don't understand what the system is doing when you use offset or you don't figure up the final voltage correctly when using offset, you'll most likely have problems.

I tried this (i.e. CPU voltage override at 1.27v) without setting the CPU at a specific speed and PBO disabled, and the CPU was performing below stock clocks (around 3.4-3.5 GHz). I guess that's expected unless I force the CPU to the stock speed, right? 

 

I will also test the voltage override with the chip at stock speed, however, am I right in thinking that in that scenario, the CPU will definitely not go over 3.7GHz which I believe is the stock speed? Or will it still try and boost further up if there's headroom?

 

With the rest of my testing, it seems like I am getting the best results with PBO Enabled, RAM on XMP 1 and a CPU voltage offset of -0.1v from a CB score and clock speed perspective (boosts up to 4.4-4.5GHz). 

 

With all of this said, I still would love to know if you have any inkling as to why PBO advanced settings seem to give me the reboots regardless? I wonder if it's the sudden drop from 1.4+ volts on the CPU down to 1.1v or thereabouts (on heavy load) with the curve optimizer etc? I'm not very proficient with undervolting/overclocking so apologies if that makes no sense lol.

 

Even if that was the case, shouldn't PBO on Advanced with no curve optimizer and no core boost still be stable? Am I just tripping or does something seem a bit strange there? Also wondering if there may be any scenario where I could run a stable curve optimizer. Maybe custom PBO limits for PPT/EDC etc? I did test a bunch of those though and couldn't seem to get it stable.

 

8 hours ago, Beerzerker said:

For stock speeds and Ryzen, 1.4v's is way too high and I'd have to say it's heating up under load.

I do want to peddle back to this too - definitely not questioning this, but just trying to understand the reasoning you have for saying so. I seem to recall reading on a lot of forums where 1.4v is deemed pretty normal for Ryzen 5900x on idle/low loads and totally fine. Is this something you think is the case with certain motherboards and Ryzen 5000s etc? And shouldn't the BIOS defaults be adjusted as such if that is way too high?

 

Thanks again for all your help

[Beerzerker]

No prob.
It's because the default voltage spec'ed is to guarantee EVERY chip period will function at stock, hence the higher voltage. Some chips are just crap and those take more than good ones do to run at the same speed, that's why they do it.

The way I force my chips to hold a certain speed is to bump up the BCLK by one increment and it holds wherever I set the speed to be, if it's left completely at stock it will change speeds because that's what it's supposed to do by design.

I've PM'ed you a link to a Gamers Nexus vid about how PBO works, it explains it better than I can put into words myself.
Once you understand how it works you'll have an idea of what's going on which helps.

BTW don't forget if you think you've over tweaked it, default settings will get you back into the ballpark of where it should be.

[DSP]

Awesome, thanks so much again. I will definitely take a closer look into the video and see what the optimal settings are for my build. 
 

Just wanted to ask again though as I have this obsessive need to try and understand/make sense of things lol. Do you have any theories at all about why PBO advanced seems to give me the reboots, even if everything is on auto and no CO or clock boost? I would’ve thought that’d essentially be the same as PBO Enabled, no?

[Beerzerker]

Nothing I could mention that the vid woudn't cover.
I'm not big on running/enabling PBO anyway but that's just personal preference on my part.

[DSP]

Totally fair. Thanks again! 

[amaru psyn]

EDIT: it seems i have finally managed to resolve the issue, and without having to disable anything !

my ram was set to a-xmp profile 2, because iirc when i built pc profile 1 wouldn't even post. it does now however, and everything has been hunky dory for days - nary a single spider-crash in more than five hours of play since - not after i overclocked gpu by a couple of hundred mhz, the vram by 100 mhz, and even reduced the voltage a lil !

HUZZAH !  

ps: i didn't even realise that there was much difference between the two profiles, as both have exactly the same cas timings. i guess that's merely the surface level though, considering just how many power setting options there are for ram alone in the bios - variables i've never messed with before, as they're beyond my knowledge

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

ok so i'm in a similar situation, with an identical recurring error code (event ID 41 task category 63), and it's driving me mental 

like you, i'd tried everything i could think of and find via event log critical errors and googling, with no resolution. then today i decided to search for "spider-man remastered pc kernel power error" without including 'rx 6700', and found this page

excited to learn that there seemed to be a quick fix, i went into bios and changed PBO from 'auto', to 'enabled'. then i loaded the game, thwipped and smashed the fuck out of the escaping riker's and raft prisoners, and for at least the next half an hour all seemed sweet...

alas ! sudden black out during last phase of fighting prison escapees - so not at a time there should have been much power fluctuation. dafuuuuuqqqqqq  

the only thing we appear to have in common is an MSI b450 mobo, so at least that likely confirms that the source of the problem is something in the bios - of which i am running the latest version available on the MSI site: AMI BIOS 7C02vH9 2022-08-15

msi b450 tomahawk max II
amd ryzen 3600 
16gb (2x 8gb) g.skill ripjaws v @ 3600mhz cl16-16-16-36
sapphire pulse radeon rx 6700 10gb
corsair rm650x

as per beerseeker, i'm gonna try disabling PBO now, but am not optimistic. so i guess i'm wondering whether the PBO setting fix has endured, and if you've had any issues since you last posted 

?

nb: one thing i have noticed about the bios is if you change overclocking settings under the 'settings' tab, it does nothing. but if you adjust the same settings via the 'overclocking' menu option, it recognises the change/s. hopefully this is an inconsequential coding oversight and not a potential cause of unseen clashing / confusion in the bios settings..

Edited February 11, 2023 by amaru psyn
problem appears to be resolved