Question BSOD/random restarts, but not overheating ?

[pixelparty]

Whenever I play Baldur’s Gate 3 on my PC I experienced computer restarts. At first, just a black screen with an instantaneous restart. Now I got a blue screen with the error message WHEA_UNCORRECTABLE_ERROR.
Checked event viewer, what I see the most when this happens is Event 46 (A fatal hardware error has occurred. Component: memory). I have checked my RAM with MemTest86 and passed. Drivers and BIOS are up to date. XMP is not enabled. Next I analyzed mini dump files with WinDbg and I didn’t see anything that explained the crashing except something about AMD (example: FAILURE_BUCKET_ID: 0x124_0_AuthenticAMD MEMORY_ UNKNOWN_FATAL_IMAGE_AuthenticAMD.5ys).

My PC is a prebuilt: iBUYPOWER Gaming Desktop - SlateMR289a - AMD Ryzen 7 5800X - 16GB
3200 Memory - NVIDIA RTX 3070
1TB NVMe SSD - Black

Bought in January of this year.

I had swapped out the AIO cooler with a Corsair almost immediately due to their bad reputation with theirs failing. I have always used Ryzen Master to optimize my cores, which I think ends up slightly undervolting my CPU (-30?) to get cooler temps (35 idle, 60-70 at most while gaming). I also monitored my CPU and GPU temps while playing to rule overheating out. If I stop optimizing my cores, my temp seems to idle at 69. I’m kind of scared to even run BG3 with idle temps that high.

My question is, has anyone had experience with this error that might be able to point me in the right direction of what could be wrong? Or could the problem be the undervolting? My power supply? CPU failing already? I never experienced this issue while playing other games before so I’m unsure how to pinpoint it.

 

[Darkbreeze]

You do realize that pretty much EVERYBODY has been having issues with BG3 right?

Have you updated to the latest installer or installed the patches that are available?

Is this the only game this happens on, or is it just the only game you've been trying to play so you're not sure?

What are your FULL hardware specifications including EXACT power supply model?

 

[pixelparty]

You do realize that pretty much EVERYBODY has been having issues with BG3 right?

Have you updated to the latest installer or installed the patches that are available?

Is this the only game this happens on, or is it just the only game you've been trying to play so you're not sure?

What are your FULL hardware specifications including EXACT power supply model?

You do realize that pretty much EVERYBODY has been having issues with BG3 right?

Have you updated to the latest installer or installed the patches that are available?

Is this the only game this happens on, or is it just the only game you've been trying to play so you're not sure?

What are your FULL hardware specifications including EXACT power supply model?

No - my friends and boyfriend have not had any issues with the game but I just want to rule out something being wrong with my PC vs the game causing issues specifically Atleast I would know it’s not my fault lol

Yes, all up to date.

So far this is the only game it’s happened with but I tend to play games not as graphically intensive on PC (sims, high on life, etc) as I prefer console gaming for most games so I don’t have a huge list of games to test it with.

Here are the specs: https://www.bhphotovideo.com/c/product/1720818-REG/ibuypower_slatemr_289a_gaming_desktop.html/specs

Again, this is a prebuilt so I’m not sure the EXACT PSU model since I believe iBuyPower has their own “brand” they use in their builds

 

[Aeacus]

Again, this is a prebuilt so I’m not sure the EXACT PSU model since I believe iBuyPower has their own “brand” they use in their builds

Well, that you need to find out since it is critical in terms of troubleshooting.

Also, ibuypower doesn't have their "own brand". What they mostly use, are low/crap quality PSUs. And that - is an issue.

If nothing else, you're looking into PC disassembly. Meaning that you either remove the back side panel to see the PSU label. Or if not, screw the PSU loose, to take it out, just to read the label on it.

No - my friends and boyfriend have not had any issues with the game

Just because few people doesn't have the issue with the game, doesn't mean there is no issue with the game at all. Baulder's Gate 3 has optimization issues and internet is filled with people having issues with the game. Unless you're implying that they all lie? Whereby issue is with something else.

To rule out game issue, you need to try with 2nd game. E.g CyberPunk 2077 would do. If you don't personally own it, does your BF has it on their Steam account? Since if he does, he could share his Steam library with you, where you can download, install and run any of his games. As long as he doesn't play any at that time.

Another option, almost equal of value, would be GPU bench. Namely Unigine Superposition,
link: https://benchmark.unigine.com/superposition

Run 1080p Extreme preset, since this will stress your system. Look if you get same symptoms as with Baulder's Gate 3.

might be able to point me in the right direction of what could be wrong?

At current moment, and without any further info regarding troubleshooting; 3 main suspects;
1. Game issue.
2. RAM issue.
3. PSU issue.

 

[pixelparty]

Well, that you need to find out since it is critical in terms of troubleshooting.

Also, ibuypower doesn't have their "own brand". What they mostly use, are low/crap quality PSUs. And that - is an issue.

If nothing else, you're looking into PC disassembly. Meaning that you either remove the back side panel to see the PSU label. Or if not, screw the PSU loose, to take it out, just to read the label on it.


Just because few people doesn't have the issue with the game, doesn't mean there is no issue with the game at all. Baulder's Gate 3 has optimization issues and internet is filled with people having issues with the game. Unless you're implying that they all lie? Whereby issue is with something else.

To rule out game issue, you need to try with 2nd game. E.g CyberPunk 2077 would do. If you don't personally own it, does your BF has it on their Steam account? Since if he does, he could share his Steam library with you, where you can download, install and run any of his games. As long as he doesn't play any at that time.

Another option, almost equal of value, would be GPU bench. Namely Unigine Superposition,
link: https://benchmark.unigine.com/superposition

Run 1080p Extreme preset, since this will stress your system. Look if you get same symptoms as with Baulder's Gate 3.


At current moment, and without any further info regarding troubleshooting; 3 main suspects;
1. Game issue.
2. RAM issue.
3. PSU issue.

I was considering upgrading the PSU since you are right, IBP components tend to be crap and I’d rather replace it with something I trust.

And no, not implying anyone is lying. I was not aware of the massive amounts of complaints from other players with the game. I guess I didn’t expect a game to crash my PC entirely like this - maybe a crash to desktop but not a hard crash. That is why I was worried it was more of a hardware issue until I browsed Reddit to see the mayhem it’s apparently causing and a lot of people are having the same hard crash issues. Makes me feel a little better.

Yes, we share a Steam library but we have Cyberpunk on console. I’ll look for another graphically intense game to run the test with or do the benchmark after work.

Does memtest not completely rule out RAM issues, btw? I also had uploaded my mini dump .dmp files on another site and someone had mentioned that a CPU hardware error is causing the crashes, and that it’s happening when reading data from the CPU’s L1 cache. Any thoughts on this?

 

[Aeacus]

Does memtest not completely rule out RAM issues, btw?

It does, IF you let it run all 13 tests and 4 full passes.

One full pass for 16GB takes ~4h. So, all 4 passes ~16h. Did you ran your memtest86 that long? I don't think so.

I also had uploaded my mini dump .dmp files on another site and someone had mentioned that a CPU hardware error is causing the crashes, and that it’s happening when reading data from the CPU’s L1 cache. Any thoughts on this?

1. There are no minidumps for me to read.
2. I don't have the know-how on how to read those. For that, we have different people.
3. Minidumps doesn't tell what the issue is exactly, they only give you further avenues what to look more closely.

Also, if the CPU would be the issue, you'd see BSoD when in Win, even at idle, and not when running specific game. IF your PC even POSTs and boots to OS, that is.

I was considering upgrading the PSU

To what, if i may ask?

 

[Misgar]

I encountered an instance recently where MemTest86+ gave a clean bill of health, when I installed XMP 3200MT/s DDR4 RAM in an AMD 2600X system (specced at 2,933 max).

I ran MemTest86+ for two complete passes at 3,200MT/s, which took 3 hours 45 minutes. No errors were reported by MemTest86+, but as soon as I started using the computer normally, it black screened without warning after 15 minutes and rebooted into Windows.

I was not running any processor intensive tasks when it crashed, so I loosened the CL (CAS) timings at 3,200MT/s by a couple of clock cycles. This fixed the instability problem.

I'd always assumed that if RAM passes several complete MemTest86+ runs, it's probably stable, but this time I was wrong.

 

[Aeacus]

(specced at 2,933 max).

I ran MemTest86+ for two complete passes at 3,200MT/s

So, RAM is specced to 3200 Mhz, you tested 3200 Mhz but when in OS, RAM runs at 2933 Mhz? If so, you do realize that different frequency results in different behavior, right?

I'd always assumed that if RAM passes several complete MemTest86+ runs, it's probably stable, but this time I was wrong.

Even faulty RAM passes the tests, but what indicates faulty RAM are errors generated during the pass.

I once ran the memtest86 with my old AMD build to see how it fares. Within 5 mins, it started spewing all kinds of errors, but memtest86 continued on doing it's tests.
Since i don't use the system as of late and it has DDR3 RAM in it, i haven't gotten around replacing the RAM. Build still boots to OS though (got to love the durability of DDR3).

 

[pixelparty]

It does, IF you let it run all 13 tests and 4 full passes.

One full pass for 16GB takes ~4h. So, all 4 passes ~16h. Did you ran your memtest86 that long? I don't think so.


1. There are no minidumps for me to read.
2. I don't have the know-how on how to read those. For that, we have different people.
3. Minidumps doesn't tell what the issue is exactly, they only give you further avenues what to look more closely.

Also, if the CPU would be the issue, you'd see BSoD when in Win, even at idle, and not when running specific game. IF your PC even POSTs and boots to OS, that is.


To what, if i may ask?

Heh, I ran one full pass and aborted halfway through pass 2 before moving on (took about 30 min) since I had read 99% of errors will show during the first pass (?) but I’ll redo for all 4 passes and let it run overnight properly.

Definitely don’t see a BDOD on Idle or ran into any hardware issues within the time I’ve even had the computer so I think I can safely rule out the CPU in that case.

I was going to upgrade to this one specifically. Also considering the 850w gold version.

https://www.bestbuy.com/site/corsair-rme-series-rm750e-80-plus-gold-fully-modular-low-noise-atx-3-0-and-pcie-5-0-power-supply-black/6537989.p?skuId=6537989

 

[Aeacus]

Also considering the 850w gold version.

PSU's efficiency does not equal to build quality. Just because PSU has 80+ Gold efficiency, doesn't mean it has to be good. Heck, there are 80+ Titanium PSUs out there (highest efficiency there is), that are complete crap. Somehow, Silverstone managed to produce such unit.

Corsair RM750e is decent, but better look towards Corsair RMi or RMx, since those are better quality. Or Seasonic Focus/PRIME.
(The 3 PCs i have, are powered by two PRIME TX-650 and one Focus PX-550 units. Full specs with pics in my sig.)

 

[pixelparty]

PSU's efficiency does not equal to build quality. Just because PSU has 80+ Gold efficiency, doesn't mean it has to be good. Heck, there are 80+ Titanium PSUs out there (highest efficiency there is), that are complete crap. Somehow, Silverstone managed to produce such unit.

Corsair RM750e is decent, but better look towards Corsair RMi or RMx, since those are better quality. Or Seasonic Focus/PRIME.
(The 3 PCs i have, are powered by two PRIME TX-650 and one Focus PX-550 units. Full specs with pics in my sig.)

Thanks! The reason I mention gold is because I was looking at a Corsair bronze unit but BF said gold is preferable because it’s more stable than bronze? But as you said, sometimes that doesn’t equal quality.

I believe I was looking at the RMx version (https://www.bestbuy.com/site/corsai...R5GwpJfBzq_kiIxAoWQaAg5GEALw_wcB&gclsrc=aw.ds) but I wasn’t sure if 850w was “overkill” even though I know the PSU will only draw the power that it needs and you can’t technically go wrong with more.

 

[Aeacus]

because it’s more stable than bronze?

That is complete nonsense.

PSU's efficiency has essentially 0 to do with how "stable" the PSU is. Also, there are no such thing as "stable" or "unstable" PSU. What there are, are build quality. Starting from complete crap (aka doorstop/fire hazard/time bomb etc), low quality, mediocre quality, good quality and great quality.

Efficiency only shows how much power PSU draws from the wall and how much of it is wasted as excess heat.
Full chart here: https://en.wikipedia.org/wiki/80_Plus#Efficiency_level_certifications

) but I wasn’t sure if 850w was “overkill” even though I know the PSU will only draw the power that it needs and you can’t technically go wrong with more.

Having 100W more than suggested isn't overkill. But if you go too over the top, e.g 1kW and higher, you'd have two cons:
* beefier PSU costs more
* less efficiency

PSU is most efficient when the load on it is between 50% and 80% of it's max rated wattage.

So, for e.g. 400W load on 650W unit (e.g Seasonic PRIME TX-650 80+ Titanium), would be 61.5%, meaning that the PSU is 96% efficient, where only 4% of power is wasted as excess heat. Whereby PSU draws 416W from the wall, gives 400W to components and wastes only 16W as excess heat.
Same 400W load on 1kW PSU (e.g Seasonic PRIME TX-1000 80+ Titanium) would mean that the PSU is 94% efficient, since load on PSU would be 40%. Whereby PSU draws 424W from the wall, gives 400W to components and wastes 24W as excess heat.
But if you'd get 1kW 80+ Gold PSU, which is the norm today (e.g Seasonic PRIME GX-1000 80+ Gold), would mean that on 400W load, PSU is 90% efficient. In this case, PSU pulls 440W from wall, gives 400W to components and wastes 40W as excess heat.

All-in-all, RM850x is better choice than RM750e, since RMx has 10 years of warranty and better capacitors in it, than RMe with 7 years of warranty and lesser quality components in it.

 

[Misgar]

To clarify my statement about 2933 vs 3200MT/s in an old system I refurbished recently.

From what I've read, the AMD 2600X officially supports up to DDR-2933.

I'd just swapped out an A6-9500 dual core APU for a second hand 2600X. I wanted to see if the replacement CPU would work with some spare DDR4-3200, instead of the DDR-2666 I used with the A6-9500.

Unsurprisingly, the CAS timing needed a small tweak to make it stable at 3,200MT/s.

This computer is part of a collection of old machines I keep for testing software. I like to check out new software on something that doesn't matter if things go wrong.

Restore Points don't always work for me, if I need to roll things back. Hyper-V images are all very well for checking new apps, but some software needs direct access to the hardware, ruling out VMs.

Reminds me a bit of regression testing new software builds at work. Sometimes they go wrong and on rare occasions, kill the hardware test platform.

I'd rather my 3800X and 7950X rigs amongst others stayed uncorrupted.

 

[pixelparty]

That is complete nonsense.

PSU's efficiency has essentially 0 to do with how "stable" the PSU is. Also, there are no such thing as "stable" or "unstable" PSU. What there are, are build quality. Starting from complete crap (aka doorstop/fire hazard/time bomb etc), low quality, mediocre quality, good quality and great quality.

Efficiency only shows how much power PSU draws from the wall and how much of it is wasted as excess heat.
Full chart here: https://en.wikipedia.org/wiki/80_Plus#Efficiency_level_certifications


Having 100W more than suggested isn't overkill. But if you go too over the top, e.g 1kW and higher, you'd have two cons:
* beefier PSU costs more
* less efficiency

PSU is most efficient when the load on it is between 50% and 80% of it's max rated wattage.

So, for e.g. 400W load on 650W unit (e.g Seasonic PRIME TX-650 80+ Titanium), would be 61.5%, meaning that the PSU is 96% efficient, where only 4% of power is wasted as excess heat. Whereby PSU draws 416W from the wall, gives 400W to components and wastes only 16W as excess heat.
Same 400W load on 1kW PSU (e.g Seasonic PRIME TX-1000 80+ Titanium) would mean that the PSU is 94% efficient, since load on PSU would be 40%. Whereby PSU draws 424W from the wall, gives 400W to components and wastes 24W as excess heat.
But if you'd get 1kW 80+ Gold PSU, which is the norm today (e.g Seasonic PRIME GX-1000 80+ Gold), would mean that on 400W load, PSU is 90% efficient. In this case, PSU pulls 440W from wall, gives 400W to components and wastes 40W as excess heat.

All-in-all, RM850x is better choice than RM750e, since RMx has 10 years of warranty and better capacitors in it, than RMe with 7 years of warranty and lesser quality components in it.

Thank you for all your help! This has been very insightful. I’m going to benchmark later today to see how that fairs and upgrade my PSU anyway with the RMx most likely.

 

[pixelparty]

That is complete nonsense.

PSU's efficiency has essentially 0 to do with how "stable" the PSU is. Also, there are no such thing as "stable" or "unstable" PSU. What there are, are build quality. Starting from complete crap (aka doorstop/fire hazard/time bomb etc), low quality, mediocre quality, good quality and great quality.

Efficiency only shows how much power PSU draws from the wall and how much of it is wasted as excess heat.
Full chart here: https://en.wikipedia.org/wiki/80_Plus#Efficiency_level_certifications


Having 100W more than suggested isn't overkill. But if you go too over the top, e.g 1kW and higher, you'd have two cons:
* beefier PSU costs more
* less efficiency

PSU is most efficient when the load on it is between 50% and 80% of it's max rated wattage.

So, for e.g. 400W load on 650W unit (e.g Seasonic PRIME TX-650 80+ Titanium), would be 61.5%, meaning that the PSU is 96% efficient, where only 4% of power is wasted as excess heat. Whereby PSU draws 416W from the wall, gives 400W to components and wastes only 16W as excess heat.
Same 400W load on 1kW PSU (e.g Seasonic PRIME TX-1000 80+ Titanium) would mean that the PSU is 94% efficient, since load on PSU would be 40%. Whereby PSU draws 424W from the wall, gives 400W to components and wastes 24W as excess heat.
But if you'd get 1kW 80+ Gold PSU, which is the norm today (e.g Seasonic PRIME GX-1000 80+ Gold), would mean that on 400W load, PSU is 90% efficient. In this case, PSU pulls 440W from wall, gives 400W to components and wastes 40W as excess heat.

All-in-all, RM850x is better choice than RM750e, since RMx has 10 years of warranty and better capacitors in it, than RMe with 7 years of warranty and lesser quality components in it.

Small update if you’re still interested lol, ran the benchmark in 1080p Extreme and had no issues whatsoever. Also ran a CPU benchmark with Cinebench and all was well with that too. I’m getting more convinced by the second that it’s just the game.

 

[Aeacus]

I’m getting more convinced by the second that it’s just the game.

Yeah, this much we told you in the very beginning. Still, get a new PSU. Less issues down the line this way.

Oh, missed this earlier;

Heh, I ran one full pass and aborted halfway through pass 2 before moving on (took about 30 min) since I had read 99% of errors will show during the first pass (?)

No.

Test 7
To use all possible data patterns 32 passes are required.

Source: https://www.memtest86.com/tech_individual-test-descr.html

No-one in their right mind would do 32 passes.

But 32 passes is absolute and will tell you if your RAM is sound or not.
2 passes are bare minimum, while 4 passes are considered acceptable.

 

[pixelparty]

Yeah, this much we told you in the very beginning. Still, get a new PSU. Less issues down the line this way.

Oh, missed this earlier;

No.


Source: https://www.memtest86.com/tech_individual-test-descr.html

No-one in their right mind would do 32 passes.

But 32 passes is absolute and will tell you if your RAM is sound or not.
2 passes are bare minimum, while 4 passes are considered acceptable.

So update, I upgraded to the Rm850x and game is still causing shut downs unfortunately.

 

[Aeacus]

game is still causing shut downs unfortunately.

Then it comes down to game issue. Need to wait until devs release optimization hotfix/update.

 

[pixelparty]

Then it comes down to game issue. Need to wait until devs release optimization hotfix/update.

I may have figured it out but I don’t want to speak too soon. But the game began crashing every 10 minutes for me now after updating some other drivers that were recently released (I was able to play for a few hours atleast before a crash). As a last ditch effort, I disabled Ryzen Master’s core optimizer which was undervolting my CPU a bit (I prefer it for the lower temperatures). Haven’t had a single crash yet. I’m guessing the game is so intensive I needed all the juice I could get with my CPU. Temps rose to low 70s (at most 80 for a few seconds in certain areas) but so far so good.