Anyone know of a CPU testing program.

  • Thread starter Thread starter C J.
  • Start date Start date
C

C J.

Hi all.

I've been trying to narrow down a problem with recurrent file corruption,
which keeps occurring on my system drive. I clean installed windows this
last time around, and I restored data folders and files I had in my old
installation from a backup file on another drive.

So far:

I've tested the harddrive with HDtune, and with the drive manufacturer's
test utility. The drive is within normal operating parameters. So I ran
Chkdsk /r /f and it found invalid data errors in the System Volume
Information Folder. I deleted all but most recent restore points - and then
ran SFC /Scannow to check for damaged core files.

I've also tested both 512MB DDR DIMMS on my system with Memtest86,
separately and together - and despite being two different clock speeds (one
is PC2700 333 mhz, and other is PC2100 266 mhz ) both are made by Xerox,
have the same form factor (are 512MB, have 2 ranks, 4 banks ) and seem to
test fine as well.

I'm getting concerned about possible CPU damage, because of the mixed clock
speed of the Memory Cards. Apparently, I've been running it this way for a
few years, and hadn't noticed the difference in type or speed - until I
removed, and reseated each card.

According to Intel - my board should only use DDR DIMM cards with a 266mhz
speed. Higher speed DIMMs were untested by Intel for this particular board,
which is no longer supported.
 
C said:
Hi all.

I've been trying to narrow down a problem with recurrent file corruption,
which keeps occurring on my system drive. I clean installed windows this
last time around, and I restored data folders and files I had in my old
installation from a backup file on another drive.

So far:

I've tested the harddrive with HDtune, and with the drive manufacturer's
test utility. The drive is within normal operating parameters. So I ran
Chkdsk /r /f and it found invalid data errors in the System Volume
Information Folder. I deleted all but most recent restore points - and then
ran SFC /Scannow to check for damaged core files.

I've also tested both 512MB DDR DIMMS on my system with Memtest86,
separately and together - and despite being two different clock speeds (one
is PC2700 333 mhz, and other is PC2100 266 mhz ) both are made by Xerox,
have the same form factor (are 512MB, have 2 ranks, 4 banks ) and seem to
test fine as well.

I'm getting concerned about possible CPU damage, because of the mixed clock
speed of the Memory Cards. Apparently, I've been running it this way for a
few years, and hadn't noticed the difference in type or speed - until I
removed, and reseated each card.

According to Intel - my board should only use DDR DIMM cards with a 266mhz
speed. Higher speed DIMMs were untested by Intel for this particular board,
which is no longer supported.
Click to expand...

The difference in RAM speed is not an issue. If one module is PC2700 and one
is PC2100, they are both run at PC2100. The machine does not attempt to run
them asynchronous with respect to one another. They "march together" like
good soldiers, never stepping on each others toes. So there is no worry
there about wear and tear.

The first thing I'd physically examine, is the cables used for passing
data to the drive. For ribbon cables, you should be using 80 wire cables.
80 wire cables and 40 wire cables, both have 40 pin connectors on the end.
The 80 wire cables connect the ground signal to the extra 40 wires. The
extra ground provides a cleaner (less crosstalk) environment for the
data signals.

If you are using round cables, modern ones are pretty good, as people have
figured out how to make them. The first round cables were a mixed bag, as
some were naively rolled up, causing crosstalk.

If the hard drive is SATA, the first generation cables don't hold very well
to the connectors. Sometimes you get a flaky connection, but you'd know it,
because the drive would disappear, the machine would freeze, and you'd get
a hint that something nasty had happened. The packets on the SATA cable
are protected by CRC, but CRC only has the ability to detect a limited
number of error patterns, so errors can still get through. If there
were errors, the drive performance would be slow and erratic, as failed
operations were retried. So, again, you might get a hint that something
wasn't right.

There are occasionally problems with particular storage controllers, and
knowing what was controlling the drive, might make it easier to look
up the performance history of the controller.

Another parameter that affects drives is temperature. If the drive was in
a sheltered spot, and didn't get some air flow, if the case was hot inside
and not well ventilated, that could contribute to problems for the drive.
But if that were the case, the drive could well have died by now, rather
than just having the odd error.

The disk manufacturer's diagnostic, available for download from their site,
can give some indication of how healthy the drive itself is. But the test
would not distinguish problem types, and if the problem was cables, the
diagnostic might not be able to say to you "change the cables". It is not
always possible with a diagnostic, to point at one specific problem.

To test the CPU, you can go to mersenne.org and get a copy of Prime95.
In the menu, there is an option called Torture Test. The program carries
out a mathematical calculation (an FFT) with a known answer. The program
can detect rounding errors or complete failure of the calculation.
People who overclock, use Prime95 Torture Test as a stability test for
their overclock. If you can run it for hours on end, without a reported
problem, then chances are your processor has nothing to do with the
corruption.

So I'd start by looking at the cabling. And even changing the cabling, if
it looks like it has been physically abused (bent, torn, connectors not
gripping cable fully).

Paul
 
Test programs don't and never have found those type of problems. This has
been true of PC's and mammoth water cooled central processors. The only way
to find the problem is to start replacing hardware. Hard drives and memory
are cheap. I would start with the hard drive as it sounds like it is on the
way out. The memory will only run at the mother board speed and shouldn't be
a problem unless it is slower than the board is rated for.If you are worried
about them remove one and only run the other.
 
C J. said:
Hi all.

I've been trying to narrow down a problem with recurrent file corruption,
which keeps occurring on my system drive. I clean installed windows this
last time around, and I restored data folders and files I had in my old
installation from a backup file on another drive.
Click to expand...
[snipped for brevity]

Somehow, I don't think the CPU is your problem. CPU problems don't manifest
themselves as corrupted files. A CPU problem generally causes the machine
to freeze as the processor just stops executing instructions, though it is
possible for some corruption to occur, it is usually a lot more evident than
just a corrupt file. Similarly memory problems tend to manifest themselves
in a similar manner.

In your case, I would be inclined to say that the data is being corrupted on
its way to or from the disk drive itself or possibly within the disk drive.
The first suspects would have to be the disk drive cable or possibly even
the disk itself (but you seem to have eliminated the drive - at least for
now). Disk drive cables can become faulty (intermittent connections) or the
cable may not be quite up to the job of carrying the data signals and
maintaining their integrity (some older cables are particularly guilty
here). Also under suspicion (and less obvious) is the power connection to
the drive. The 4 pin connector on parallel ATA drives can becomes loose
with even moderate connection and disconnection. A momentary break can
corrupt files easily (SATA drives are less vulnerable because multiple
connections carry each power rail, plus the connector is a better design -
but I don't think you have such an animal). An under-rated power supply can
also be at fault here.

Thus a good start is to check the power connections and power supply rating.
If they are OK then replace the drive lead with a good quality replacement
80 conductor cable. Ensure that if you are using a single drive that it is
connected to the end of the cable and not the middle connector (the
unterminated stub will compromise the integrity of the signal due to
reflections). Check that the hard drive isn't getting too hot (upper
temperature limits vary but 55 °C is usually tops). SpeedFan can check
this. And if that doesn't solve the problem, try a replacement drive.
They're cheap these days (or borrow one).
 
M.I.5¾ said:
C J. said:
Hi all.

I've been trying to narrow down a problem with recurrent file
corruption, which keeps occurring on my system drive. I clean
installed windows this last time around, and I restored data folders
and files I had in my old installation from a backup file on another
drive.
Click to expand...
[snipped for brevity]

Somehow, I don't think the CPU is your problem. CPU problems don't
manifest themselves as corrupted files. A CPU problem generally
causes the machine to freeze as the processor just stops executing
instructions, though it is possible for some corruption to occur, it
is usually a lot more evident than just a corrupt file. Similarly
memory problems tend to manifest themselves in a similar manner.

In your case, I would be inclined to say that the data is being
corrupted on its way to or from the disk drive itself or possibly
within the disk drive. The first suspects would have to be the disk
drive cable or possibly even the disk itself (but you seem to have
eliminated the drive - at least for now). Disk drive cables can
become faulty (intermittent connections) or the cable may not be
quite up to the job of carrying the data signals and maintaining
their integrity (some older cables are particularly guilty here). Also
under suspicion (and less obvious) is the power connection to
the drive. The 4 pin connector on parallel ATA drives can becomes
loose with even moderate connection and disconnection. A momentary
break can corrupt files easily (SATA drives are less vulnerable
because multiple connections carry each power rail, plus the
connector is a better design - but I don't think you have such an
animal). An under-rated power supply can also be at fault here.
Thus a good start is to check the power connections and power supply
rating. If they are OK then replace the drive lead with a good
quality replacement 80 conductor cable. Ensure that if you are using
a single drive that it is connected to the end of the cable and not
the middle connector (the unterminated stub will compromise the
integrity of the signal due to reflections). Check that the hard
drive isn't getting too hot (upper temperature limits vary but 55 °C
is usually tops). SpeedFan can check this. And if that doesn't
solve the problem, try a replacement drive. They're cheap these days
(or borrow one).
Click to expand...

Wow was surprised to see someone had posted to this thread as recently as
this week.

Well, to update what has transpired since last post here: The 80 wire 40 pin
cables are both brandnew. Power Supply, fans check out okay. The overall
system checked out okay - though for the sake of elminating other problems -
I replaced the Memory modules and I swapped out system drives. I had been
running a older 60GB XP Pro system drive that I had for years - up until
yesterday when it started clacking loudly... so I had to put the problematic
300 GB one back in.

Somewhat the same deal was happening to 60GB - as with this 300 GB drive.
The drive was running very good until suddenly windows XP set itself to run
chkntfs, forced a reboot and determined the Drives file System and volume
folder were dirty. About 148 USN and Journal index entries were removed.
Post chkdsk /r - it started clacking and would freeze up system while
running some programs and or browsing internet.

Drive scanned clean of Viruses, or other malwares (slaved it into a laptop
as External USB drive) Only thing done to 60GB drive was windows updates
(hadn't been in over a year and 3 months), and I changed Antivirus and
Firewall programs.

I'm F***ing fed up with this s**t. Only thing I haven't replaced is the
motherboard And I'm inclined to believe its probably about time to build an
entirely new PC.
 
...
I'm F***ing fed up with this s**t. Only thing I haven't replaced is the
motherboard And I'm inclined to believe its probably about time to build an
entirely new PC
Click to expand...

First off, you are trying to fix things by swapping parts. That can
even exponentially complicate the problem. You were on the right
track. But did not finish the job.

OK. I don't even see where you first reviewed event (system) logs
for historical problems stored with the all so important error
numbers. Also Device Manager should be checked (although I expect to
see no problems).

Now what can make all hardware appear intermittent. Same concept in
a house. When multiple doors start jamming, do you plane the doors,
or first confirm the house foundation is intact. Do you immediately
starting fixing things, or first find the problem? In your case,
that foundation is a power supply. Without expensive equipment, or
something simple like a $20 multimeter (even sold in K-mart), nobody
will know if the power supply 'system' is defective or good. Yes, I
said 'system' because power supply is only one component part of a
power 'system'.

Setup computer to consume a maximum load. That means video card
must be doing complex graphics (ie uncompressing and playing a movie),
while reading many files from disk, while reading from CD, while
downloading from the internet, while using the modem, while ... what
else can your system also multitask to? Now you are ready to take
four critical DC voltages on any one orange, red, purple, and yellow
wire from power supply to motherboard. Those numbers must exceed
3.23, 4.87, and 11.7. And then post those numbers here because
numbers also contain other useful information.

Until we have those numbers, then everything in the system is
suspect. Only after a power supply 'system' is exonerated do we move
on to other suspects.

Heat is a diagnostic tool. Fools who never learned how electronic
work want to fix intermittent failures with more fans. Put that
computer inside a 100 degree F room or heat selective components with
a hair dryer on highest setting. Yes, as hot as a hair dryer can be.
These are pig heaven happy temperatures to all working electronics.
But defective or intermittent hardware may fail repeatedly at these
temperatures. Now the computer is ready (warm enough) for Memtst86.
Same applies to disk drive diagnostics. Entire disk subsystem is not
tested if manufacturer diagnostics are not run hot.

Step by step we eliminate usual suspects. Heat is a diagnostic
tool. You had the right idea, but use heat to make those tools
useful. Cold (below 40 degrees) is also a powerful tool but typically
not as good at finding defects.

BTW, this is also why responsible computer manufacturers routinely
provide comprehensive hardware diagnostics for free.

With above numbers and diagnostic reports, then we can move on to
other suspects. Meanwhile, there is no way to avoid the need for that
meter other than to spend $thousands on other test equipment. Power
supply can only be tested when fully taxed - that means multitasking
to all computer peripheral simultaneously. Even a power supply
tester is all but useless. Get the meter. Report those voltages
here.
 
Back
Top