Tech Support Guy banner
Status
Not open for further replies.

Crashing to black screen or freezing pc

3K views 18 replies 2 participants last post by  Winny1 
#1 ·
Hello all - just joined the forum and added some system info to my profile.

Story so far:

All seemed well until trying to render video projects using Pinnacle Studio 11.1, when I would get random freezes or black screens with the pc being unresponsive, requiring power down (5 seconds holding the button) or reset.

After some time on Studio related forums I bought more RAM recommended from Crucial. No difference. I bought a new video card. No difference.

I defragged, updated drivers, updated bios. no difference.

I then started experimenting with Prime95 torture test, Memtest86+, PassMark's Burn-In Test and Windows Memory Diagnostic with the following results:

Memtest86+ completed one pass ok. Have not yet left it running for longer.

Prime95 runs overnight with no errors when testing large FFT's so far. It produces the black screen crash overnight on small FFT testing but was running ok for over 2 hours to start with.

PassMark Burn-In shareware only runs for 15mins before stopping. Occasionally this crashes or freezes.

WMD always fails on the Stride6 test. If I vary the RAM (original 512MB, new 1GB, or total 1.5GB) it will fail at diffent points during Stride6. Only once did it complete Stride6 (when running with 512MB RAM) but then failed on the 2nd pass. Reading other forums on Stride6 failures makes me think it could be a PSU, CPU or mobo issue.

Took the case off my PSU last night and noticed some corrosion and leakage from a capacitor. Bought and installed a new PSU but the Stride6 and Studio failures are still there just the same.

So, as you can see I've battled long and hard. Any ideas? Should I spend yet more money and fit a new CPU? Could it still be a software problem? If a new CPU doesn't fix it and the process of elimination points towards the mobo am I opening a can of worms trying to replace that?

Many thanks!

--Winny
 
See less See more
#2 ·
Given the power supply issues it's possible that there was some damage to the motherboard. It's a shame that you upgraded all those parts before diagnosing the problem. But everyone needs ram so that's not a big deal.

I would run memtest at least 3 full passes before I was happy with it. But given you've crashed the other tools would lead me to think there is still some hardware not happy. Now the errors you had with stride6 were they on the old stick? Did you try removing this and just using the new ram?

Another culprit could be heat, is it getting really hot? I could see this causing your PC to shutdown, possibly crash if certain apps don't want to shut down nicely.
 
#3 ·
Hi - thanks for the quick response.

Stride6 has failed for me with all 3 combinations of RAM (old on its own, new on its own, both).

Temps do seem a bit high but I'm not sure it's the answer. During high use tests like PassMark where just about every part of the pc is working hard, I can get CPU temps of between 68 and 71 deg C. But I have witnessed my system going down at only 58 deg C and watched it run at 70 for an hour...

--W
 
#4 ·
Those temps do seem pretty high. I don't know what kind of hardware (CPU, Video cards, etc) you have so it's hard to tell exactly. I flag all my Athlon processors at over 60 degrees celsius because this concerns me, and only under extremely long peak CPU intervals do I get the warnings at this stage.

But lets assume that's not the problem. Since the RAM doesn't seem to be the problem, you now need to determine if the CPU or the motherboard is the problem. At this point I'd just replace both but thats my preference (strictly for warranty purposes). You also could try replacing the CPU and heat sink assembly first. My heart thinks its probably the motherboard tho.
 
#5 ·
Here are my system specs:

Motherboard: Gigabyte GA-7VT600-1394 KT600
DDR400 400fsb x8AGP.
Athlon XP 3200+, 2.21 GHz
Memory: (original) 512 MB DDR SDRAM PC 2700 (DDR 333)3 Memory recently upgraded to 1.5GB by adding CT12864Z335, 1GB, 184-pin DIMM RAM as recommended by Crucial.
Hard Drive: 120 GB 7200 rpm, 8 MB cache UDMA Hard Disk Graphics: PNY GeForce 7600GS AGP 8x with 256MB RAM
Windows XP SP2


Reading about Stride6 elsewhere I saw frequent reference to the L2 cache. Any way to see if this is the problem? I know the 2 Prime95 tests I have used test RAM and L2 cache differently.

Edit: Seems the small FFT Prime95 torture test really focusses on the L2 cache, and the large FFT on the RAM and CPU RAM interface (Northbridge?).

I have left the large FFT test running today, it will have done 20 hours if still going when I get home. Maybe if it's ok then that means the Northbridge is ok (we think the RAM is ok anyway) and so it might be L2 cache)...
 
#7 ·
Hi,

I was editing my previous post whilst you were posting.

Is the benefit of me using CPU-Z to learn what the cache sizes are, or does it do some sort of health check?

I'll give it a go tonight if it looks useful.

It would be great to be able to isolate this problem - been driving me nuts for weeks on end.
 
#10 ·
Ok - the good news is that Prime95 ran for over 18 hours with no problem on the large FFT. So I'm hoping this implies a CPU issue...

I have run CPU-Z and have more information than I know what to do with. In the txt and html reports I can see reference to the caches but how to tell if it has found anything wrong?

Cheers,

--W
 
#11 ·
If the caches look fine and they are as advertised then stress testing the CPU would be in order. Prime has done that so at this stage it doesn't look like a CPU problem.

Check with Intel/AMD web site for actual spec's for your CPU and compare numbers. Any deviants should be noted.
 
#13 ·
If the numbers match then your ok, if something doesn't match it could indicate a problem.

I mean if you know you've got a 2.4Ghz cpu, and it's testing at 2.0 somethings up.

If the L2 cache is not reporting, that would certainly indicate a failure.

As for the reports, they are really more for debuggers, for instance if you are having a lot of crashes, you can refer to this to find out exactly what address points to what hardware (driver) that is causing the problem. Thats a poor example but I just wanted to get the point across.

Now if you find .dmp files and can do some cross referencing you may find out that all your problems point to one particular driver but the driver isn't the problem, then you can see if the hardware is it.
 
#15 ·
The AMD guys thought instability was due to CPU temperature, so recommended reseating the heatsink with fresh thermal paste. I did this last night - took the heatsink off, cleaned up the copper face and the CPU with Q-tips and iso-propyl alcohol (camera lens cleaner), applied a small dab of Alaska silver thermal paste, refitted everything and turned on.

My immediate thought was how cool the temps were, I gave a quick blast of the PassMark Burn-in test program with all features on it peaked at about 51 deg C. Similarly with Prime95. I have seen temps up to 71 C before.

I then ran Windows Memory Diagnostic and it completed 2 full passes, whizzing through the Stride6 test with ease.

I then set up Studio 11.1 to trundle along for a couple of hours doing video rendering of a file which always caused crashing or freezing before hand. Unfortunately it got about two-thirds through before crashing, and since then it will not pass Stride6 in WMD again either.

So, interesting results from all this. Could the thermal paste have moved/run/altered since initial application so the old errors returned. It seems a coincidence for WMD to have been passed straight after applying the new paste for it not to be related to the problem?
 
#17 ·
I removed the heat sink and CPU again. Perhaps I had put a little too much on, there was a small bead of excess along one edge of the CPU. I cleaned up and applied less paste. CPU temp on start-up was now a very high 65 deg C so I suspected I'd not put enough on. I ran WMD anyway and got my familiar failure on the first pass through the Stride6 test (the same as before I removed the heatsink the first time).

Removed components again and sure enough the paste only covered about two-thirds of the CPU. Cleaned up and applied more than last time, and possibly less than the first time but it's hard to tell. Fired up the pc and starting temp a much healthier 42. Ran WMD and it failed Stride6 but only on the third pass through so it is looking increasingly like a temperature related stability problem.

So now my question is: Should amount of thermal paste be this critical, or has my CPU degraded in some way so I should buy another?

Many thanks for any help once again.

--W
 
Status
Not open for further replies.
You have insufficient privileges to reply here.
Top