cquirke (MVP Windows shell/user)

On Wed, 25 Jul 2007 05:47:09 -0400, "Rick Rogers" wrote:

>You can, but there is now an onboard memory diagnostic tool. Just type
>'memory diagnostic' into the search box.

I'd use MemTest86, as it is MUCH safer.

The onboard tool works as follows:
- may require Windows to launch
- writes to the installation so that it runs next boot
- starts to run after the next boot
- does a quick test and then stops, unless you "etc."
- writes results to the HD installation's even log

In contrast, MemTest86:
- boots off the CDR
- repeats tests until stopped by the user
- accumulates test results on screen
- does not write to the HD at all

When RAM is bad, all bets are off. Not only can stuff being written
to disk be corrupted, but where they are written on disk can be
corrupted, too (e.g. you can splat over NTFS's MFT, etc.).

Also, most "read disk" calls are one bit-flip away from being "write
disk" calls, so you don't want an OS that requires thousands of read
operations off HD to boot (not to mention all the automatic write
operations; temp, page file, directory access timestamps, etc.)

When BIOS detects a memory parity error, it halts the system, because
it "knows" this is the only safe thing to do.

Similarly, Windows will halt the system when errors are such that
context loss makes it unsafe to continue; BSoDs in 9x, STOP in NT.

MS is either losing this clue, or is treating us with contempt (e.g.
it's more important to cut support calls than to preserve user's HD
contents) when they move from halting on BSoDs to automatically
restarting the system (effectively, hiding these errors).

So when MS designs a RAM tester specifically for use where RAM quality
is suspect, and this process makes liberal use of both read and write
traffic on HD, one wonders if they've completely lost the plot.



>--------------- ----- ---- --- -- - - -
To one who only has a hammer,
everything looks like a nail
>--------------- ----- ---- --- -- - - -