BSODs probably caused by hardware

[Pete301]

Hey,

I have been having problems recently with my setup and wanted to pick peoples brains as to what might be causing these blue screens that affect my computer more and more frequently.

My rig is on my profile. Now, I have been having blue screens quite consistently over the past 6 months while using windows 7 64 bit. I had put it down to some iLok drivers that I required to run some software (uninstalling them wasn't really a long-term option for me) so I just dealt with the blue screens as they were not very frequent. However I recently had another go at debugging my computer and I uninstalled these drivers, a combination of this, installing new gfx drivers and reinstalling some other software caused quite a big mess on my windows install. It came to the point where I'd try to boot up windows and it came up with PFN_LIST_CORRUPT blue screens. So I decided to just reinstall windows.

Now that I had a clean install of windows I thought to myself "let's try and debug this BSOD issue once and for all". I didn't install any new drivers other than graphics drivers, and the only software I had running was Steam that relinked itself as it is on a separate hard drive. I was now confident that I wouldn't get a blue screen, however after about half an hour of running the machine it blue screened with exactly the same error code that I had been getting before all of this. This leads me to believe that it could be a hardware problem.

Onto the next stage, I ran Memtest86+ on both my sticks of ram for 4/5 passes, and it all came through without errors. Reseating the ram didn't solve the blue screens. I have tried using just one stick of ram, and so far that works fine, using either stick of ram seems to be fine. So I thought maybe it's something with the dual channel memory management or something, so I tried sticking in both of the sticks but just on 1 channel, that still caused a BSOD.

Through some googling I saw that upping the voltage a bit on the ram can solve some issues with it, but I don't want to do that in case it is the ram that is faulty and then I can't RMA it. I haven't yet reseated the CPU to see if that fixes anything as I have no thermal paste handy. I also haven't had a chance to test if it's the GFX card that is failing.

I was wondering whether it could be the power supply causing the problem because it is about 5-7 years old, it might be that, it's a 550W Xclio Greatpower PSU.

Thanks,
Pete

Sorry for the long post.

[kompukare]

A BSOD is just an (unrecoverable) error message and without you telling us what the error message is I can't see anyone being able to help you!

If you weren't able to write the numbers down you should turn off the auto-restart option in Windows Control Panel\System and Security\System > Advanced System settings >Advanced > Startup and Recovery Settings > uncheck automatic restart

Also if there are anything in %Windir%\minidump you can download nirsoft's BlueScreenView:
http://www.nirsoft.net/utils/blue_screen_view.html

[Pete301]

Ahh yes, sorry about that, forgot to include the errors.

The stop codes I have been getting are (in order of most common):

0x0000003b - ntoskrnl.exe
0x0000001a - fileinfo.sys
0x0000001e - ataport.sys

I also have the dumps if that will help?

[kompukare]

Hm, they seem all over the place!

Actually looking up PFN_LIST_CORRUPT gets 0x0000004E which seems to be memory but not as in hardware more like a bad driver poking around where it shouldn't:

http://www.tomshardware.co.uk/forum/...orrupt#9554367

so while that usually is from a bad driver, bad hardware might cause that error too I guess. 4/5 passes isn't a lot for Memtest86+, overnight would be better. Also Memtest is not the only tester out there but I think you are best off trying Memtest86+ again to see if you can get an error or trying to replicate the problem somehow.

Rather suspect that Xclio only provides 'great power' in name though although it seems they're made by CWT so not the worst OEM. If you can't replicate the problem with MemTest86+ then try a memory stress test in Windows while also stressing the GPU and maybe the HDD to see if the PSU is to blame. Otherwise you can just try swapping RAM around but without a know good spare it's hard to be sure.

[Pete301]

Ok, I'll try the memtest86+ tonight, to check the ram. In the meantime, how do I perform a memory stress test that I can perform in windows?

[kompukare]

there's HCI memtest

http://hcidesign.com/memtest/

without the pro version you might have to run more than one instance though. The likes of OCCT might be useful too since it has a PSU test (basically tries to load everything while monitoring the voltages AFAIR)

[Domestic_Ginger]

Prime blend maybe. IBT maybe?

HDtune for HDD is a quick one.

Your ram settings and voltage are well within spec?

I had a real killer last night; after windows update and a little bios tweak I restarted to be greated with a flashing cursor. went into bios reset changes; same again. tried restarting and on and off at wall. No joy and I was starting to get a bit pissed; I then realised my sansa fuze was still plugged into USB port. Went back into the bios for boot priority and lo and behold my pc was trying to boot from my mp3 player unplugged my mp3 player and all was fine again!

[Pete301]

Ok, I just ran the HCI memtest, I ran two instances of it at once with one stick installed and it found in total 90+ errors. Obviously this is not normal, I am just about to check the other stick of ram now.

[Pete301]

Ok, I ran the second stick to 75%, again with two instances running and no errors were found. I put both DIMMs in the same slot for testing to make it fair. So does this mean I have a faulty stick of ram? Would it be wise to still run the MemTest86+ overnight? Shall I leave both sticks in for the overnight test, or test one tonight and one tomorrow?

Also, I always had the ram speed and voltage within spec. the only thing that I changed was when I noticed that it was running at 1333 instead of 1600. I then went into the BIOS and changed it to XMP which is correct for my ram AFAIK (however I have changed it back to 1333 now to try and stabilise things, surely that wouldn't be a problem?)

[kompukare]

Might be worthwhile running MemTest86+ and taking a photo in case corsair don't want to RMA (I assume they have lifetime warranty).

It is possible that the slots on mobo are a factor too. Did you try the 'bad' stick in the same slot as the good one?

You could also try underclocking the RAM to see if you still get the errors although an RMA might be the easiest solution.

[Pete301]

Ok cool. They do have a lifetime warranty, and yea I tried the bad stick in the same slot as the good one. I also have underclocked the RAM. Should I just MemTest just the bad stick and send it to corsair?

[Domestic_Ginger]

Ask them if you can keep one.

I suppose if your pc needs the ram you'll have to tell them. If its a matched pair then they might want both though:/

[Kumagoro]

Check your boot drive for bad sectors as well.

I like hdat 2 on hirens boot CD, use the most powerful test.

[Pete301]

I ran memtest86 overnight on the bad stick of ram and it came up with nearly 300 errors over 10 passes. I then checked my good stick of ram for the most part of this morning and it came back with no errors over 8 passes. Both sticks were inserted into the same slot on the motherboard when I ran the tests. So I guess it looks like it's the ram. I will do some HDD tests as well but I have sent a support ticket to corsair asking if I can just RMA one of the sticks (yes they came as a matched pair). Waiting to hear from them now. Thanks for all the help guys

[Pete301]

Just an update...

I asked to RMA the faulty stick of RAM back to Corsair, only to find out that I would need to send both sticks to Holland. This means that it'd cost about £27 to get it delivered and tracked, for which I could buy a new stick of RAM anyway and wouldn't have to have any down-time on my PC. Now I am in the process of asking Scan if the RAM is eligible to be sent back to them for a replacement, which would turn out significantly cheaper than sending to Corsair.

If I can't manage to send my RAM to Scan then I will probably end up buying a new stick. Are there any suggestions that you can offer in terms of matching the ram with the existing one in my machine?

Thanks

[Kumagoro]

Surely it can't be that expensive. International signed for is only about an extra 6 quid so I would have thought a tenner would be more likely.