Windows 7 - windows 7x64 pc is locking up

I am using a Windows 7 x64 Pro pc for file and
print sharing for a 8 user group.  The pc has a
Intel i7-2600k cpu with a Gigabyte z68xp-ud5h
motherboard and 16 Gb of ram.

I am using a AMD 6790 video card that had trouble
with "streaking" on another pc with 2 monitors.

The pc is randomly locking up after periods of
high disk access.  Could it be the video board ?
Should I throw the video board in the pond ?

Thanks,
Lynn

Any warning or error in System and Application event logs?

No.

Thanks,
Lynn

Hi, Lynn.

Any chance of a heat issue?  That's what always comes to my mind when I hear
of a failure after some operating time.  Are your disks internal or
external?  Adequately cooled?  (Dust bunnies cleaned out recently?)

RC
--
R. C. White, CPA
San Marcos, TX
rc@grandecom.net
Microsoft Windows MVP (2002-2010)
Windows Live Mail 2011 (Build 15.4.3555.0308) in Win7 Ultimate x64 SP1




No.

Thanks,
Lynn

I do not know anything about that card, but you could probably tune
its "acceleration" down or off, or make Windows use a vanilla video=20
driver, or disable Aero.  That'll release some memory, anyway,
and I suppose the card /could/ have a memory management issue.

But disk activity or network activity in normal network use may=20
be triggering the crash.

it is a lot of memory for a server, so is it someone's PC?

You could use a disk test tool - you may have one in BIOS.
It could be a disk erase tool, so read the instructions
and/or make sure of your system backup.  SystemRescueCD
is free, good for tools, less good for instructions.

Or, carefully unplug and then re-connect the network card /and/=20
the hard disk at every point in the system where unplugging
is possible.  Oh, and investigate whether the PC power supply=20
has enough watts for all of your hardware.  If there are USB=20
peripherals, try a separate powered hub, to lighten the load. =20
If the electricity supply is sub-standard, buffer it with an=20
uninterruptible power supply that also makes the supply stable.
Maybe do not power the monitor fr=89m the PC.

Server is in 72 F room.  Case interior is cool
with case fan running at medium speed.  The PC
was fully rebuilt last week.

Lynn

16 GB of ram is cheap, only $90:

http://www.amazon.com/Corsair-Vengeance-Channel-Memory-CMZ16GX3M4A1600C9B/dp/B004RFBIUU/

I swapped the video board for a AMD 6850 HD.  It
started streaking on me while I installed ACT! 2012
and I decided that was bad sign.

Yes, I have a UPS.

Lynn

Some UPSs are worse than wall power, especially any modified sine unit.

If you "rebuilt" the PC and added more RAM, I wonder if you need to
adjust ram timing. Maybe the automatic settings are too fast.

I have run into RAM issues where it matters how the DIMMs are built, i.e.
what size memory chips are used. That is, some DIMMs are comprised of
higher density chips which in turn loads down the memory bus less.

More RAM is always good, even if it means slower bus timing. My next
system will have 32Gbytes. All the better for virtual machines.

I have got 5 other PCs with the same motherboard
and ram combo.  Of course, none are a file server.

Thanks,
Lynn

Well there is always the ram test. Takes forever though.

Maybe it would be quicker to swap the RAM into a different machine.

I had a PC crashing which turned out to be, I think, a RAM module
returning some bits correct and some wrong, and somehow the EFI test
reported that the RAM was okay.  However, a program on SystemRescueCD
and available separately, named Memtest, Memtest86, or MemTest86+ -
I think the third is a fork of the second - found the bad module.
It was designed to report single bad bytes or bits in memory (and then
you just set your PC to not use those memory locations!), so 2 GB
playing up all at once was quite indigestible.

I also once, long ago, helped a colleague at work with their home
computer, and with the boss's consent, by swapping their memory
modules with those in a workplace PC.  Since both computers worked
afterwards, the boss was happy to let the memory stay swapped.
The specification of the memory modules was the same, so I do not know
why there was a problem, but the other interesting features, as far as
I remember, were that (similarly?) the operation of cache memory meant
that the BIOS test did not detect a fault, and/or data in RAM was
corrupted if you left it several seconds between writing and reading,
so I suppose you say that it was not refreshing right - that could be
why it passed tests, it had not got corrupted yet.

This was back around the days when well-informed burglars would break
into offices and pull memory modules out of desktop PCs - they just
took the RAM.  I accept that this is a while ago, but, even so, you are
using a well-specified PC for a job that I believe could be done by
a Raspberry Pi.