Re: HUGE problem with A7V8X-LA and data corruption issues! (Hewlett Packard System)

In article <r-2dnXPGtYpkOUrZnZ2dnUVZ_vKdnZ2d@xxxxxxxxxxxxx>, Rob
<robc@xxxxxxxxxxxxxxxxxxxxxx> wrote:

ByTor wrote:
Specs on the board:

Hewlett Packard Pavilion a720n

http://h10025.www1.hp.com/ewfrf/wc/document?lc=en&cc=us&dlc=en&product=
443068&docname=c00069442

Honestly I am at my wits end in troubleshooting this issue..... ;0(

I will try & explain what I've done so far but it has been exhaustive &
may have missed including it here so please bare with me.

The bottom line problem is that data is corrupting on this machine when
I do "network" transfers, or sometimes partition/partition/drive
transfers.....2 examples:

1) I create images to a seperate drive in the machine. If I restore that
image from where it was originally written it's perfect....If I copy
that images to another partition(whether it's the same drive or not) and
restore it the image errors with "corrupt or invalid image."

2) Large file transfers across the network no matter what
drive/partition I copy to it corrupts. But very small file transfers are
fine????? I have 8 computers networked in my home and this one is the
only problem with corruption.

Here's what I've done:

1) Ran memory tests(memtest full 18hrs), removed memory, used different
memory.
2) Changed the RJ45 cable leading to the NIC and used various ports from
my router & 8 port workstation.
3) Disabled the onboard NIC card slapped in a PCI NIC.....Worked great
for awhile than crapped again(I've tried various NICS and the ones I've
tried worked perfectly in all my other machines & no corruption. Even
the one that crapped worked great on another machine....
4) Removed **ALL** hardware (CDRoms, etc) & only left a stick of memory
& one harddrive, did a **CLEAN** install of both Win2K & XP with no
updates on the machine & the data still corrupts.
5) Ran full drive read/write scans on the harddrives present, no issues.
Put these drives in other machines and data transfers were
perfect........

I think I may have covered a good portion of what I done......I am by no
means a rookie at this stuff but this has got me twisted as data tranfer
is vital to me for the use of this machine.........I do use tools to
verify the data which is QuickPAR and/or a SFV verification file.

My warranty is up so calling tech support is useless and personally I
think I'll be fighting with them more with their basic approaches.....I
can't deal with that right now ya know, blamin the OS, I don't have
their original OS crap on it.....Go to start/programs, dbl click
this.....Arrrrrrrrrrrgh!!!!!!

I would REALLY REALLY appreciate anyones help on this.......And yes I
know proprietary blows but this machine was an exceptional deal......I
always build my own.........





Could be the CPU is getting a little too hot on the longer data
transfers. Download Prime 95 and run the "torture test" mode. It
should fail out pretty quick if that's the case. Maybe you could get
away with cleaning the HSF assembly and putting a fresh coat of grease
on it. Just for the heck of it I'd try a different Video card too!
(might be some memory leak going on there)
Rob

First off, if you were contemplating replacing the motherboard,
the nearest Asus equivalent might be A7V400-MX SE. That board
doesn't have a Firewire chip, and the sound solution uses a
different chip, but the Northbridge is KM400A and Southbridge
8237.

In addition to Rob's test, you could also try setting up a
RAMdisk in main memory. Then transfer files over the network
and into the RAMdisk. Now, my problem with doing this on
a system here, is I really couldn't make the RAMdisk very
big, as I only had 1GB of memory in the system. The RAMdisk
avoids using the 8237, and allows you to test the network
interface (either Southbridge based, or one of your plugin
NICs). That might help distinguish between a disk based
problem and a NIC problem.

The problem could be almost anywhere. One thing you could
try, is to disable the Firewire chip, and repeat some of
your file transfer tests again. I would want to examine
which devices share physical interrupt lines and see if
you can work up a theory. (Devices assigned the same
IRQ, are candidates for sharing the same physical interrupt
line, and that is one way to validate any IRQ information
that might have been offered in the manual. Devices on
different physical wires, should not be assigned the
same IRQ.)

When using the 8237 to host disks and do network transfers,
that didn't use the PCI bus, as near as I can tell. Using
your NIC card uses the PCI bus, which would be one difference.
Since the plugin NIC solved the problem for a while,
your problem probably isn't caused by the PCI bus. The
Firewire chip sits on the PCI bus.

I think a few more test cases and it is probably time
to swap in another motherboard. It doesn't have to be
the A7V400-MX and you could select something else if
you wanted - I note you're using a non-HP setup on
the boot disk, so that means you are free to use
whatever hardware you want.

One thing that makes me nervous about microATX boards, is
the use of a Northbridge that has built-in graphics. Some
chipsets have stability issues when built-in graphics
are used, and the system memory or FSB are cranked to the max.
One cure for this is to use a separate
graphics card, and that allows the graphics core in the
Northbridge to be shut down. But if that was an issue,
chances are Prime95 would fall over in seconds.

Do the chips run cool ? Does the Northbridge heatsink
look like it has any thermal paste, to help make contact
with the heatsink ?

If the symptoms had shown early enough, you might have
got the warranty to do some of the work for you.

HTH,
Paul
.

Relevant Pages