Idle BSODs on 3700x

[ReD-EyeD]

I'm in dire need of a solution or direction to what to do with my problem. Sorry if it's a long post, but I'm trying to do my best providing as much info as possible.

I'm getting random BSODs during idle or low load stuff like watching youtube or fast scrolling pages. It can happen randomly from a few times a day to once per three day.
The system is: Ryzen 3700x, Asus tuf x470-plus gaming, RAM Kingston HyperX HX432C18FB2K2/16 32gb (https://www.kingston.com/dataSheets/HX432C18FB2K2_16.pdf), Thermaltake Smart 750w 85+ gold, gtx 1070

Bsods range a lot, but most of them are IRQL_NOT_LESS_OR_EQUAL, KERNEL_SECURITY_CHECK_FAILURE or DRIVER_OVERRAN_STACK_BUFFER specifying ntoskrnl.exe, ntkrnlmp.exe or amdppm

What I already did:

• Reinstalled Windows twice (one bsod happened during one of the installs)
• Temps are never higher than 80 for cpu and 75 for gpu
• Using the latest bios
• Using the latest chipset drivers (tried both versions from AMD and from Asus sites)
• Reinstalled and tried multiple versions of GPU drivers, used DDU
• Played with ram voltages a bit
• Calculated Ryzen RAM timings and tried them, instant bsod
• Tried XMP off and on
• Tried different RAM speeds
• Updated and reinstalled different versions of network and sound cards
• Ran chkdsk, sfc scan, driver verifier and other windows tools
• Ran memtest86+ a few times overnight. 2 times no errors in 10 hours (at least 4 passes) and 1 time a small amount of errors on the 5th pass

Here is the latest minidump: https://drive.google.com/open?id=1RJZE2Y7-H3Daa6G1TnSFGke_VfJHTEZP

Feel-craft:
Almost never crashes on load or during heavier games. Goes through Cinebench, RealBench and other lighter benchmarks.
Prime95 shows error 5 seconds into the test saying this: FATAL ERROR: Rounding was 0.5, expected less than 0.4 Hardware failure detected, consult stress.txt file. and 2-3 threads stops doing the test. Don't know if it's connected or not.
I may be wrong here, but somehow it felt more stable on 2133 ram than on the default, but still did bsod on idle on the next day. Default ram voltage is 1.2, but I tried going to 1.35 and it also felt more stable, but once again I'm not sure if it's true as bsods are very random in nature.

I don't know what to do now. I need a tech Dr. House to solve this. Please, ask questions if I missed something important to mention.

 

[eidairaman1]

Bsods at random on idle has been ram

 

[ReD-EyeD]

Bsods at random on idle has been ram

Can this be connected to prime95 errors and some lighter games crashing (as in closing instantly without any error)?

 

[neatfeatguy]

Could a RAM issue.

I try to do the most basic TSing first before messing around with other settings. Sometimes simple is the best way to go about it.

First, I'd reset the BIOS to default settings. After that, I'd test one stick of RAM by itself and see how things run. Run the system for a while. If you get a BSOD, then swap RAM sticks and test the other one by itself. If you don't get a BSOD after a long period of testing, then swap RAM sticks and run the other one by itself to see if it throws out any BSODs or not.

 

[birdie]

Looks like a memory issue to me. Run memtest86 for an hour from a USB flash drive and see if it's OK.

Also try reseating your RAM sticks.

 

[eidairaman1]

Can this be connected to prime95 errors and some lighter games crashing (as in closing instantly without any error)?

Yes but to sit idle and crash says ram.

Ive experienced this before all the way back to the xp days.

Could a RAM issue.

I try to do the most basic TSing first before messing around with other settings. Sometimes simple is the best way to go about it.

First, I'd reset the BIOS to default settings. After that, I'd test one stick of RAM by itself and see how things run. Run the system for a while. If you get a BSOD, then swap RAM sticks and test the other one by itself. If you don't get a BSOD after a long period of testing, then swap RAM sticks and run the other one by itself to see if it throws out any BSODs or not.

If it wasn't crashing I'd have him update the bios.

 

[ReD-EyeD]

Looks like a memory issue to me. Run memtest86 for an hour from a USB flash drive and see if it's OK.

Also try reseating your RAM sticks.

I did ran memtest86+ from an USB as I mentioned before. Two times it went fine for 4 passes and another time there were errors on the 5th one.

http://imgur.com/lZo5Dx5

I'm also concerned about prime95 instantly turning off 1-2 cores before them reaching 100% load

If it wasn't crashing I'd have him update the bios.

I have the latest bios already

 

[eidairaman1]

I did ran memtest86+ from an USB as I mentioned before. Two times it went fine for 4 passes and another time there were errors on the 5th one.

http://imgur.com/lZo5Dx5

I'm also concerned about prime95 instantly turning off 1-2 cores before them reaching 100% load

Replace the ram

 

[birdie]

If you had errors at least once it indicates your memory is faulty. Increasing RAM voltage might or might not help you. Try safer frequencies/timings anyways. Start with something basic, like 2666MHz.

In a perfect world memtest86 must produce no errors if you run it for 24 hours straight (that's rarely necessary as a single pass usually reveals something is wrong).

In theory your CPU might be faulty instead but I've no idea how to test it. You could probably check it using OCCT Perestroika. I've never had CPU errors in my entire life, so again I've no idea.

 

[ReD-EyeD]

@eidairaman1 @birdie
Sadly, it's quite hard to get new ram during this quarantine.
How likely it's a ram problem? I read some things about early ryzens having problems with some of their cores or something. Also wondered about mobo not being properly optimized for gen3 and other things. And if it's ram then what should I aim for?

 

[R0H1T]

Are you undervolting your RAM? I usually run them like that when I'm not overlocking the memory, or both at times

 

[ReD-EyeD]

Are you undervolting your RAM? I usually run them like that when I'm not overlocking the memory, or both at times

No, 1.2 is it's default. I did try upping it to 1.35 and it still bsoded

 

[kingkongtol]

i have suggestion, based on my aorus b450 pro, r5 3600, vgen tsunami-x 32GB kit @3600, and rx vega 64, i also got random bsod at first, tried your lists already, updated windows and still got bsod, another day when i unplugged my xbox controller and never plug it again, i realize bsod never appear, maybe you can check which usb device triggered the bsod or another hardware, sometimes monitoring software can be a bsod trigger
with overclocked ram, you may re-tune your voltage, especially your soc voltage, until you got stable system with lowest voltage, then add 1 level higher voltage for stability

edit: this amd bsod thing makes me crazy, i must swap all my mechanical hdd with ssd

 

[ReD-EyeD]

i have suggestion, based on my aorus b450 pro, r5 3600, vgen tsunami-x 32GB kit @3600, and rx vega 64, i also got random bsod at first, tried your lists already, updated windows and still got bsod, another day when i unplugged my xbox controller and never plug it again, i realize bsod never appear, maybe you can check which usb device triggered the bsod or another hardware, sometimes monitoring software can be a bsod trigger
with overclocked ram, you may re-tune your voltage, especially your soc voltage, until you got stable system with lowest voltage, then add 1 level higher voltage for stability

Thank you for the suggestion, but I doubt that it's a problem here. All I have now is razer mouse and keyboard which I already downgraded from synapsis 3 to 2 as a test and all bsods point to other stuff like ntoskrnl, ntkrnlmp or amdppm.

 

[R0H1T]

No, 1.2 is it's default. I did try upping it to 1.35 and it still bsoded

Perhaps I missed it but did the system run stable at default setting i.e. everything at default including CPU, RAM, Windows basically anything that can be tweaked?

 

[kingkongtol]

Thank you for the suggestion, but I doubt that it's a problem here. All I have now is razer mouse and keyboard which I already downgraded from synapsis 3 to 2 as a test and all bsods point to other stuff like ntoskrnl, ntkrnlmp or amdppm.

yeah, i didnt think it was my xbox controller, but bsods never happen again after i unplug it

edit: you can also try set power mode in windows to performance, and always update your windows, its free

 

[ReD-EyeD]

Perhaps I missed it but did the system run stable at default setting i.e. everything at default including CPU, RAM, Windows basically anything that can be tweaked?

If the system was stable at defaults I wouldn't had this much headache. For now it seems a bit more stable at lower ram speed than on absolute default, but I still can't run prime95.

 

[neatfeatguy]

Thank you for the suggestion, but I doubt that it's a problem here. All I have now is razer mouse and keyboard which I already downgraded from synapsis 3 to 2 as a test and all bsods point to other stuff like ntoskrnl, ntkrnlmp or amdppm.

It's most likely not the problem, but I absolutely hate the Synapse software for Razer. The software was buggy and caused other issues when I used one of their mouses (that actually completely failed around the time the warranty ended) I had to uninstall and clean the registry of the Synapse software to eliminate the issues I was running into.

Based on things you've done and the MemTest86+ that eventually had errors - issue right now is pointing to the RAM. You need to pinpoint the problem by either testing each stick of RAM by itself or using completely different RAM. If one stick of the RAM is good and the other is bad, you can run the system with just the one good stick of RAM until you can get it replaced or returned to the manufacture for a replacement set.

 

[R0H1T]

Well if everything's at default then RAM looks like the culprit. You can try an RMA or perhaps get in touch with the manufacturer given the long turnaround times right now.

 

[ReD-EyeD]

It's most likely not the problem, but I absolutely hate the Synapse software for Razer. The software was buggy and caused other issues when I used one of their mouses (that actually completely failed around the time the warranty ended) I had to uninstall and clean the registry of the Synapse software to eliminate the issues I was running into.

Based on things you've done and the MemTest86+ that eventually had errors - issue right now is pointing to the RAM. You need to pinpoint the problem by either testing each stick of RAM by itself or using completely different RAM. If one stick of the RAM is good and the other is bad, you can run the system with just the one good stick of RAM until you can get it replaced or returned to the manufacture for a replacement set.

Oh you don't need to tell me about how bad Synapse is, I absolutely agree. Had millions of problems with it as well.
Thank you for the answers, I'll do what you just said. Should I also run the memtest overnight on a single plank?

 

[neatfeatguy]

Oh you don't need to tell me about how bad Synapse is, I absolutely agree. Had millions of problems with it as well.
Thank you for the answers, I'll do what you just said. Should I also run the memtest overnight on a single plank?

Just use your computer as you normally would with one stick of RAM in it. If you still experience crashes, then swap out for the other stick of RAM and see how that one functions.

 

[eidairaman1]

@eidairaman1 @birdie
Sadly, it's quite hard to get new ram during this quarantine.
How likely it's a ram problem? I read some things about early ryzens having problems with some of their cores or something. Also wondered about mobo not being properly optimized for gen3 and other things. And if it's ram then what should I aim for?

Tell tell sign of a crashing on idle and then you having errors during memory test.

I've experienced the same thing and been down the same road you have this is all I'm going to say on it do what you want

 

[ReD-EyeD]

Just use your computer as you normally would with one stick of RAM in it. If you still experience crashes, then swap out for the other stick of RAM and see how that one functions.

Ok, I will. I'll post here in a few days then with an outcome. Hopefully you guys will be around Thanks again!

 

[John Naylor]

I did ran memtest86+ from an USB as I mentioned before. Two times it went fine for 4 passes and another time there were errors on the 5th one.

I'm also concerned about prime95 instantly turning off 1-2 cores before them reaching 100% load

1. Memory is a fail, replace it.

2. When installing new, check manual for proper slots to be used.

3. Consult the various 3700x RAM troubleshooting threads

4. Skip P95 ... not really useful for modern processors.

5. Run memtest 86+ overnight

6. Run RoG real Bench Benchmark (8 minutes) uhaving only necessary components connected ... UIf it passes run the stress test for 2 - 4 hours depending on what comfort level you want. RoG RB is a multitasking real world application based test that, in my experience, detect errors that 24 hour P95 stable configurations miss.

1. Memory is a fail, replace it.

2. When installing new, check manual for proper slots to be used.

3. Consult the various 3700x RAM troubleshooting threads

4. Skip P95 ... not really useful for modern processors.

5. Run memtest 86+ overnight

6. Run RoG real Bench Benchmark (8 minutes) uhaving only necessary components connected ... UIf it passes run the stress test for 2 - 4 hours depending on what comfort level you want. RoG RB is a multitasking real world application based test that, in my experience, detect errors that 24 hour P95 stable configurations miss.

Note Amazon Prime Deliveries have been running 5 - 11 days ... dunno if it's them or their shipper. So make sure to ask what shipper they using and how long it takes

 

[ReD-EyeD]

@John Naylor thank you for the answer
2. Yeah, had problem with it before. It's A2-B2-A1-B1 for this board
3. There are so many problems though and random solutions like disabling core scaling and such didn't work for me, I tried most of it
6. I ran it for 30 min and didn't had problems, guess I'll try it for 3 hours, but as I said it has never bsoded on heavy load, only during idle or low tasks like youtube

Also what ram should I aim for? Just the QVL list? Because I saw a lot of posts from people with exactly same problem and it didn't help them switching to the QVL ones. Is there some kind of fool-proof option?