HELP - how do I start narrowing down where's the problem?

Velokki

Velokki

Velokki

Associate
Joined
22 Nov 2014
Posts
14
Hi guys, really appreciate anyone who contributes to solving this problem, it's driving me nuts.

TL;DR version: I ran Prime 95 for 2 minutes and my rig crashed. Restarted, ran again, it lasted for 20 minutes but crashed again. How do I narrow down what component is causing the crashes?

Longer version:

I've been having problems where the my PC just freezes and I get a loud buzzing sound from the speakers, like this. It has also frozen a couple of times without making the sound. I've had this setup since 2011 (with the exception that I changed a GTX 570 to 970 in 2014) and even though it's clearly made to be a gaming rig, I've not really played games with it at all! I remember that the rig froze a couple of times in Bioshock Infinite and Tomb Raider back when I had the GTX 570. This discouraged my gaming even further, and I thought that my GPU was just giving up so I decided to treat myself to a GTX 970. Bioshock Infinite also crashed on the GTX 970 which made me seriously doubt that GPU was ever the problem.

Well, now lately I've been having a lot of freezing. Mostly when watching full screen videos; Youtube, flash videos or desktop VLC. But it's also happened during just desktop work, browsing and music stuff (Guitar Pro or Cubase running).

I googled and saw that Prime95 would be a good way to test system stability. Well, like you read in the beginning, I've only been able to run it twice for short periods of time in blend mode until the system crashed. Now my question is, how do I start narrowing down what components are causing the crashes?

It is notable that I've had problems with USB ports - it seems as if sometimes the USB ports didn't power up or just work in general. I remember having to always unplug and plug in the USB keyboard during startup, though a simple system format solved that problem. But for ex. now that the system crashed during Prime95, I saw that the mouse had lost its led lights - so the power was clearly cut off for it.

I've been thinking the motherboard's the problem, but that's just 100% guessing. I seriously have no knowledge of what's causing the freezing.

My rig consists of:
AsRock Z68 Extreme3 Gen3
Corsair 650W 650AX 80+ Gold
Galax GTX 970
Intel Core i7 2600K
G.Skill 16GB (4x4GB) Sniper Low Voltage, DDR3 1600MHz
Samsung 128GB 830 Series SSD
Samsung 1TB SpinPoint F3 3.5", SATAII, 32MB, 7200RPM

The components are put together in a Fractal Design Define R3 case with the Thermalright TrueSpirit cooler and some Noctua fans. There's also a Samsung Blu-Ray Combo attached.

I want to get this solved ASAP, I really appreciate anyone who helps me here. This problem is stressing me out and hindering my creativity with music because I can't trust that the system is stable during guitar recording, drum programming etc!

Cheers!
 
1. Have you tested with one ram stick in, then swap and try the other one.
2. Have you checked HDD health status.
3. Is your CPU overclocked (assuming not), try to downclock it if the first two options do not improve the situation.

NOTE - Ensure you have CPU Temp checks running whilst using PRIME. (afterburner is good: https://gaming.msi.com/features/afterburner).
 
I've not overclocked, ever.

What's the best way to check HDD health?

Also, do I need to configure anything when taking out RAM one by one and rebooting? Or does it detect automatically how much RAM there is?
 
Ok, thanks. Will do that as soon as I can.

Any other diagnostics or stress tests I should do?
And when I try each of the RAM sticks, one at a time, should I always run Prime95 then?

I'm quite the beginner with these!
 
Yes - stress test each time with prime95 and monitor cpu temps as well...

Let us know what happens. If there is still no change then someone with more experience will have to help you diagnose the CPU problem.
 
Hello good people!

I've been very busy/lazy.... Took some time, but I finally did Memtest86 for the whole of last night. No problems whatsoever, complete test went fine many times and 0 errors. It was all sticks in, haven't yet tested with other programs, but honestly, is there a need if Memtest86 already did the complete test a couple of times in the span of over 12 hours?

CrystalDiskInfo provided faultless results.

I should note, that when I originally posted this thread, I ran Prime95 for 5 minutes and the computer crashed. Ran it a gain for 20 minutes until the computer crashed.

I think I should run Prime again, but would you already say based on this info, that the CPU's out of luck? Is there any info I can provide you guys to help diagnosing when I run it again today? (I do feel kind of silly now that I think of it, that I already didn't deem the CPU to be the problem, since Prime95 certainly did fail)

Thanks a lot!
 
Last edited:
Sounds like your CPU might be overheating, install RealTemp and check your CPU aint getting too hot, it should be staying below 90 degrees really at stock speeds while stress testing.

If its getting too hot just check your cooler aint clogged up with dust, if its a water cooler feel it and make sure its vibrating so you know the pumps running. My cpu cooler sometimes gets unseated when im moving my PC so its worth reseating the CPU cooler also if your still getting heat problems after you've cleared the dust. Might even be worth changing the thermal paste on there.

If it hasnt been overheating im not sure whats causing it as CPUs dont normally just fail for no reason, i would maybe checkinto your software, update everything run a virus check etc.
 
Check your temperatures while running Prime95. I would seriously consider checking that your heatsink is mounted correctly and possibly replacing the thermal paste while you're there.
 
The rig endured 30 minutes of Small FFTs, but when the temps were stably staying around the 65 celsius mark, I got impatient and put blend mode on, in hopes to create more heat and get to the torture zone faster. The computer crashed around 10 mins of running the blend mode. Temps were around 65-70 region. I have a gut feeling of it being the power supply or MOBO, since the CPU temps seem to be very tolerable. I do also have the True Spirit cooler that seems to be properly in place, too, which probably helps quite a lot to keep the temps low like that.

What do you think? Any diagnostics I can provide you?

Thanks guys!
 
mickyflinn said:
What bios version are you running ?
Click to expand...

Bios%20ver.png
 
Seems like your cpu temps are fine, maybe try running Prime and a GPU bench at the same time, that should push your PSU pretty hard and crash immediately if your PSU is failing.

I've gotta admit that i agree its more likely to be the motherboard, i've had loads of motherboards fail over the years but only a couple of PSUs.

I would try reflashing the bios just to make sure theres nothing wrong there.
 
I've still yet to test the PSU with GPU + CPU test simultaneously, will report when done.

I feel like the motherboard's caused me trouble for a long time with all the USB port power issues, random behavior etc.

I'd like to just buy a new motherboard but literally no one seems to make 'em anymore! Is there a shop selling older mobos or what'd be a good place to get one? I'd definitely like a new one with warranty, though, since the failure rate of motherboards seems to be quite big and don't wanna end up with another mobo that's already failed with the previous owner.

I just wouldn't wanna ditch much cash at the moment for a new cpu + mobo, since they do cost quite a lot. The problem is also that the 2600K is still so damn powerful, and gains of upgrading a CPU would be quite minimal (correct me if I'm wrong).

Whaddya reckon?
If I upgraded my mobo + cpu, would that make my DDR sticks obsolete or could I run them with newer DDR4 mobos as well?

Thanks again, you guys are incredibly helpful!
 
There are DDR3 LGA 1151 motherboards on OcUK. But if I was upgrading, I'd go DDR4 rather than hinder the new CPUs performance with slower RAM for the sake of saving a bit of money.

Save up a little more money (easier said than done at this time of year!) and get a DDR4 board.
 
Yeah, I read that there's consensus that there's simply no sense in going with old DDR3 on a new board. I just wouldn't want to invest in PC parts right now, but I guess life is expensive, haha.

But you know what is curious? I've been running Prime95 Blend for 10 hours now overnight and computer is still going strong without any problems.
cpu%20temps%202.png


I seriously hate the sporadicity of this! Sometimes the PC functions fine and sometimes it just crashes every now and then. Yesterday it could take 10 minutes of Blend, now it could take 10 hours without problems.

I'm gonna do the PSU test today. But other than that, I seriously don't know what's wrong! If that's okay, then probably mobo, right?
 
I would guess yeah, its hard to narrow down random faults that dont always repeat them selves. I havent used Prime95 in years since it tended to fry haswell chips, when your setting up the test are you inputting how much ram it uses? Im not sure if that was an option or not, i know on realbench its an option, if you set it too high and you have other system tasks using memory it can cause a crash when its trying to use more mem than whats available.

I dont think thats the problem since ya remember it crashing games too but i thought i better mention it since ya said it was only crashing one of the tests before.

Its a pain getting hold of them older boards, when my mobo has failed in teh past i tended to use it as an excuses to upgrade tbh, you could always sell the CPU and ram and put the money towards the new setup if ya did that, 2600ks are always in demand still since theyre a nice chip.
 
Tried Prime95 + Furmark simultaneously. First run lasted 20-30 mins until crash. Second time it ran fine for 40 mins, but then I just stopped testing, thinking that no meaningful test data would be gained - it's gonna crash sometime, maybe 1 hour, maybe 10 hours, but anyway, it wouldn't bring the specific issue into light. CPU seems fine I guess, RAM sticks stress tested multiple times, GPU is fine, Power doesn't seem to cut out and it's a quality Corsair with 80+ Gold rating and a 7 year warranty. So it's definitely got to be the low-quality AsRock (ArseFu*k) motherboard. Not gonna do that mistake ever again and save on the motherboard, let alone not properly stress test it in the beginning.

Gonna buy a quality motherboard next time around, maybe Asus TUF series or something and before I think it's allright, I'll f'in dip it to lava, freeze it in a freezer, run stress tests for a year and then deem it reliable if it hasn't failed, haha.

I guess I'll just go with this semi-unreliable setup for now until I see what AMD Zen and Kaby Lakes are about. Preferably I'd want a reliable rig for the next 5-10 years (I've had this setup for 5 yrs now, and could easily go next 3-5 if this would just work!), though my GTX 970 will have to be updated for VR down the line.

Are there any particularly trustworthy brands of Mobos with long warranties?
 
You must log in or register to reply here.
Back
Top Bottom