G
Guest
Guest
Archived from groups: alt.comp.hardware.overclocking.amd,comp.sys.ibm.pc.hardware.chips,alt.comp.hardware,alt.comp.hardware.pc-homebuilt (More info?)
On Sun, 17 Oct 2004 10:22:09 -0700, "JAD" <Kapasitor@coldmail.com> wrote:
>Memtest86+ is actually a good, if not the best, software based
>> >>memory tester. It is certainly a very good confidence check
>
>
>false sense of security
Snipping to create context you can argue with is a nasty Usenet habit.
>But how much data corruption might have occurred before Windows
>> throws up a BSOD? I can't really afford to run the risk of corrupt
>> data just because I want to use Windows to test my memory in the way
>> you mention.
>
>
>if there were horrible things wrong with your memory you will know it
>long before you test with SW testers.
>
>BSOD tell the tale when they come up. If there are memory registers in
>the 'cause' line you will know its memory related.
Again...RUBBISH!
>if you are that concerned about memory errors...use ECC memory
>only....
Once "validated" through testing, diagnostics and a few days of running
with an OS, modern memory modules are generally good for years of reliable
use...IME. ECC has its place of course... which is slightly different from
initial checking/validation.
>the only way to know for certain is to remove the stick and take it to
>a hardware tester.
<sigh>We all have to start somewhere when dealing with a newly built system
- memtest86 does a reasonable job and serves its purpose. It's simply part
of a proactive approach to system integrity at time of build and
installation.
This is simple stuff: we have a mbrd from one supplier, a CPU form another
and memory modules from a 3rd, all supplied with power from a 4th. Why
anyone would want to argue with the concept of running some diagnostics as
a first step beats me... possibly you rely on Dell to do that for you??ô_ô
> There are memory controllers and many other things
>indirectly controlling memory, if this is the area of concern,
>software checkers are not up to snuff, and may give you memory errors
>that are not chip related, or NOT tell you and you replace mods for no
>reason.
Hmm, seems to contradict what you just said about registers and memory
errors.<shrug> Then again, it depends on the system - AMD64 systems have
no "other things" between the CPU package and the memory, other than the
mbrd traces and sockets.
Rgds, George Macdonald
"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??
On Sun, 17 Oct 2004 10:22:09 -0700, "JAD" <Kapasitor@coldmail.com> wrote:
>Memtest86+ is actually a good, if not the best, software based
>> >>memory tester. It is certainly a very good confidence check
>
>
>false sense of security
Snipping to create context you can argue with is a nasty Usenet habit.
>But how much data corruption might have occurred before Windows
>> throws up a BSOD? I can't really afford to run the risk of corrupt
>> data just because I want to use Windows to test my memory in the way
>> you mention.
>
>
>if there were horrible things wrong with your memory you will know it
>long before you test with SW testers.
>
>BSOD tell the tale when they come up. If there are memory registers in
>the 'cause' line you will know its memory related.
Again...RUBBISH!
>if you are that concerned about memory errors...use ECC memory
>only....
Once "validated" through testing, diagnostics and a few days of running
with an OS, modern memory modules are generally good for years of reliable
use...IME. ECC has its place of course... which is slightly different from
initial checking/validation.
>the only way to know for certain is to remove the stick and take it to
>a hardware tester.
<sigh>We all have to start somewhere when dealing with a newly built system
- memtest86 does a reasonable job and serves its purpose. It's simply part
of a proactive approach to system integrity at time of build and
installation.
This is simple stuff: we have a mbrd from one supplier, a CPU form another
and memory modules from a 3rd, all supplied with power from a 4th. Why
anyone would want to argue with the concept of running some diagnostics as
a first step beats me... possibly you rely on Dell to do that for you??ô_ô
> There are memory controllers and many other things
>indirectly controlling memory, if this is the area of concern,
>software checkers are not up to snuff, and may give you memory errors
>that are not chip related, or NOT tell you and you replace mods for no
>reason.
Hmm, seems to contradict what you just said about registers and memory
errors.<shrug> Then again, it depends on the system - AMD64 systems have
no "other things" between the CPU package and the memory, other than the
mbrd traces and sockets.
Rgds, George Macdonald
"Just because they're paranoid doesn't mean you're not psychotic" - Who, me??