[SOLVED] Trying to diagnose potential hardware issues causing system instability

[LynneL]

Hey all, so I was going to buy someone's old desktop they built around 2016; he reset the OS and gave it to me to test out. The first problem was when I went to install a Realtek audio driver and it crashed on restart. I chalked this up to Windows Update messing something up while applying updates without me noticing (including trying to upgrade to 1903). From then on it would crash continually until I uninstalled the driver and rolled back the updates. It still had problems though, so eventually I just reinstalled the OS.
Up to this point, I had been assuming this was entirely a software issue, since there was evidently no instability before this, but then it crashed after I went to login post-reinstall. So now I am thinking hardware issue.
To test that, I tried memtest86. The results were that it never showed an "error", but would randomly freeze. I've included some photos of where it froze:

https://imgur.com/a/GRe2bLY

I couldn't see any pattern in how it froze. I thought maybe it had been jostled in the dude's car on the way to my place, so, not wanting to mess with the internals too much myself, the next day I asked him to come reseat stuff. After he did, memtest ran without problem for over an hour, getting to pass 2. But then it crashed. Running it again after that, it seemed to crash earlier the next time. Isolating the different RAM didn't seem to change anything. I considered maybe a temperature issue, given that it seemed to get a little better if you let it alone for a bit, but I've never seen the temperature go too high at any point. Here's some photos of it post-reseating stuff:

https://imgur.com/a/QP7q1to

First two are the ones directly after reseating, last two are two I ran consecutively today.
Most of the time it's on CPU 0, but I think that's a result of how quickly it crashes, since I've tried different options for cpu in memtest, and none of them seemed to change things in a positive way. Should have documented that more, but in any case if there are specific options I should try I can try them.
So basically that's where I am at with this, and I'm not 100% on how to proceed. I'm worried it's maybe it's a cpu issue, but I've also heard of motherboard stuff causing issues with memtest. Let me know if there's any other information I can provide that would be helpful, and any advice on how to get more useful information here or ideas on the problem would be greatly appreciated, thanks.

Here are the specs:
OS: Windows 10 Pro
OS Installed on: Samsung SSD 840 EVO 120GB
CPU: AMD FX9590
CPU Cooling: Corsair water cooling
GPU: AMD Radeon R9 290X Sapphire Tri-X
MOBO: ASUS Sabertooth 990FX
RAM: 16GB Crucial Ballistix Tactical DDR3-1866
PSU: Thermaltake Smart M850W 80 Plus Bronze
Sound Card: Creative Sound Blaster Z

 

[COLGeek]

The FX-9XXX series is possibly the worst CPU series in AMD CPU history. They tend to run VERY hot, use a ton of power, and generate a lot of heat.

You should run away from this system before spending a penny on it.

If you already took possession of it, underclocking the CPU to FX-8350 speeds has helped others with similar issues. Don't even try to OC this rig. Your PSU is of mediocre quality and that too will contribute to system issues. The 290X is also power hungry and generates a lot of heat.

 

[USAFRet]

he reset the OS and gave it to me to test out.

Give it back.

 

[LynneL]

The FX-9XXX series is possibly the worst CPU series in AMD CPU history. They tend to run VERY hot, use a ton of power, and generate a lot of heat.

You should run away from this system before spending a penny on it.

If you already took possession of it, underclocking the CPU to FX-8350 speeds has helped others with similar issues. Don't even try to OC this rig. Your PSU is of mediocre quality and that too will contribute to system issues. The 290X is also power hungry and generates a lot of heat.

Hm, well I'll keep that in mind. Honestly I'm just embarrassed this happened while I had it, since I have no reason to believe there was instability previous to this. I don't understand what I would have done to have caused it, but given the chronology I'll just be happy to figure out the problem, and then I'll worry about this stuff since I would feel pretty bad to just give this guy back his computer like "sorry it's broke".
Anyway, regarding changing the clock speed I was thinking about that but wasn't initially sure how to do it. I found the option in the BIOS though, and I set it down to around 4GHz. Running the memtest with that now, so I'll report back on whether or not it freezes again.
I'm curious since you mention heat stuff. I don't know a lot about OCing, but can that cause issues outside of those caused by stuff getting too hot? Like I said, I've never seen actual bad temperature readings from this thing, but is there still reason to suspect that temperature could be a big issue?

Give it back.

Yeah I mean I get the sentiment but see above I guess. We talked about it, and while I haven't bought it from him he is away for a couple of weeks so we agreed I'd try to figure out what's up and we'll talk again when he's back. I potentially shouldn't have gone into the situation here in the first place, but I wanted to give full context to indicate that this is both not a new system but also one I'm not that familiar with.

EDIT: I'm probably going to go to sleep soon, but so far memtest has run for about 50 minutes without freezing which is pretty good. While it's crashed after running for longer than this before, it's definitely a good start. Assuming that this fixes it, how should I understand the problem itself? That is, what specifically comes from the cpu running too fast that causes the instability?

 

[Rogue Leader]

EDIT: I'm probably going to go to sleep soon, but so far memtest has run for about 50 minutes without freezing which is pretty good. While it's crashed after running for longer than this before, it's definitely a good start. Assuming that this fixes it, how should I understand the problem itself? That is, what specifically comes from the cpu running too fast that causes the instability?

The CPU overheats the VRMs on the motherboard which are what provides it with power. The few existing successful AMD FX-9590 Based systems have active cooling around the CPU focused on those VRMS (as in a fan pointing at it). Also it tends to beat up on your power supply as well, and being that the one thats in this one isn't very good that doesn't help the case.

Our of curiosity whats the price on this machine? I too agree its a major no buy, Horrible CPU that gets beat in gaming by i3's from a few years ago. Just trying to get it stable may lead you to tearing your hair out.

 

[LynneL]

Well the thing is still going at 8 hours in; working on finishing pass 4 of 4. I think that's only with CPU 0 though, and if it does it for each then theoretically it's about to try to run for 8*7=56 more hours. Not sure that I should let it do that given the problem seems to have been identified.

The CPU overheats the VRMs on the motherboard which are what provides it with power. The few existing successful AMD FX-9590 Based systems have active cooling around the CPU focused on those VRMS (as in a fan pointing at it). Also it tends to beat up on your power supply as well, and being that the one thats in this one isn't very good that doesn't help the case.

Our of curiosity whats the price on this machine? I too agree its a major no buy, Horrible CPU that gets beat in gaming by i3's from a few years ago. Just trying to get it stable may lead you to tearing your hair out.

I read a little about it before bed, and apparently it can also do with the voltage? Something about the transistors not flipping fast enough at high clock speeds. I guess the argument for overheating-based explanations in this case would be that the temperature monitors aren't in the right places to "see" the components that are overheating? His cooling setup seems pretty okay on the places the temperature sensors are present. I'm wondering if maybe he had these settings changed, either the clock rate, or the voltage, or something via software in Windows and when he reset it we ended up in this less tenable state.
Anyway, for the pricing it was set at 400 Euros "or best offer" and while we haven't taken the sound card out yet I was going to replace that with a DAC I have for 30 Euro off the total price. Honestly, the plan was to do some haggling and based on looking up the components and other context I thought I could get it down to 300-330 euros. I forgot that I'm literally incapable of haggling, though, so we just discussed the sound card thing and landed at 370. Oh, small thing but if we're talking pricing I should mention that I left out the two HDDs it has (4TB and 1.5TB).
Some context is that I'm working abroad right now, and am a bit overwhelmed with a lot of stuff at the moment so a desktop that seemed to be a slight upgrade from my one back home for a not-astronomical price that I don't have to take the time to figure out and build was pretty enticing. So long as the thing runs, anyway. Literally, I can't even get packages to this apartment and I've had stuff stolen when I tried to ship it to a friend's place. It's a real mess, so idk it backfired here but I'm in general more willing to pay for the comfort of not having to think about stuff like this lately.

 

[COLGeek]

This system, even with the additional HDDs, is not work €370. Seriously, you should reconsider altogether and just tell friend you changed your mind after consultation.

 

[LynneL]

This system, even with the additional HDDs, is not work €370. Seriously, you should reconsider altogether and just tell friend you changed your mind after consultation.

Yeah I might; like I said now that it seems to be more stable I'll start considering the higher level "should I get this at all" question. I do very much appreciate this advice, but just to be clear the reason I'm not like "yeah I def' won't get it" is that it's pretty much this or nothing since I have no plans to figure out a build myself right now. At best I'd just be waiting for another desktop to show up on the little marketplace deal my work has. I'm over here for a little over another year, so really just something usable for that period that I can resell on the marketplace at the end would be nice. It's been a while since I've been like really into gaming, so I was expecting less "running the latest AAA games at max settings on 4k" and more "reasonable cpu for Ableton and will let me finish DOOM at comparable settings to my old desktop".
Additionally, I think the events here kind of imply a price renegotiation, though that's dependent on my actually bringing it up which is a lot easier to theorize about than to actually do.
Out of curiosity, say the thing runs alright at 4GHz, in your estimation what would it be worth? I mean at a hyperbolic minimum it's worth, e.g., the price of the SSD, or the case, etc. I tried to price out some of the stuff myself by comparing benchmarks and things back when I was first looking at it, and allowing for some upmarking I didn't feel like I was getting taken on a complete ride, so to speak.

 

[LynneL]

Ah, it seems I may have spoken too soon on the stability front. The test I ran it with (the one it does automatically if you don't go into the configure part) uses CPU 0 by default. It ran for about ten hours to completion without problem, but since it wasn't using all the CPUs I decided to try the "parallel" option in memtest just to be sure. Selecting this, it crashes into a restart (instead of freezing) pretty quickly. Additionally, trying to select CPU 1 instead of CPU 0 for the single CPU option it does not even start. Instead it immediately freezes. This is all still running it at 4GHz. It doesn't seem to be a temperature thing in either case (almost certainly not in the latter).
So, should I worry about this? From what I'm reading I'm thinking it might just be a problem with how memtest86 works, rather than something actually indicating real system instability given how well it performed otherwise. Certainly I'd be surprised if it was a RAM thing at this point, but on the end of the CPU to make sure it's fine is there anything else I should try? I've seen mention of "prime95" and "OCCT" for CPU testing so, while I have to focus on other stuff right this moment, looking into those is my current idea for next steps.

 

[Rogue Leader]

I can't stress enough, especially in the case where you're looking for a slight upgrade that "just works", and all the problems you've seen so far, there is never a scenario where this will be a reliable system. And for 370? There HAS to be something else used you can find.

 

[LynneL]

I can't stress enough, especially in the case where you're looking for a slight upgrade that "just works", and all the problems you've seen so far, there is never a scenario where this will be a reliable system. And for 370? There HAS to be something else used you can find.

Yeah, I mean something else might show up on the marketplace. It's a definite possibility, since it's a relatively large workplace (which I guess is why we have a "marketplace") and has people that are into this stuff. Most of the stuff I see on there is out of my price range (there's one on there for 850 euro right now), so it would really just be a matter of waiting and hoping something in the same range comes up.
I don't think it's totally fair to discount the possibility that things might be okay with the lower clock rate, and in general until I know the final state I can get this thing in it's hard to make any kind of definitive on what I'm going to do with it. It was evidently working okay for him, so while it might be a bad deal, if I give it back to the dude worse than it was before he brought it over I feel like that's a way worse deal.

 

[Rogue Leader]

I don't think it's totally fair to discount the possibility that things might be okay with the lower clock rate, and in general until I know the final state I can get this thing in it's hard to make any kind of definitive on what I'm going to do with it. It was evidently working okay for him, so while it might be a bad deal, if I give it back to the dude worse than it was before he brought it over I feel like that's a way worse deal.

Based on our collective experience with this processor, trust me, its fair. I'd bet money the system has always had issues and he just ignored them. You'd be surprised what people deal with and treat it as normal.

I agree with you in trying to get it to work right, but theres only so much you can do. If he didn't hand it to you with it downclocked in the BIOS already, thats not how he did it. Personally I would tell him "hey from the minute I got it, it has been nothing but problems" and outline the steps you took, and let him deal with it.