On Tue, 29 Jul 2003 13:15:39 +0100, "Mark \(UK\)"
<(E-Mail Removed)> wrote:
>> There's little to no gain running the memory bus faster than the
>> FSB... if you have the FSB that high, then it's an area to look at as
>> cause for the instability, but otherwise try setting the memory bus to
>> synchrous setting with the FSB.
>
>My FSB is set to default (think thats 266MHz) for my MB and cpu (athlon xp
>1700+).
Oh, well certainly the system won't post if you up the memory bus to
400MHz, since it only does "+33", which would be running at 333Mhz.
You might even consider backing the memory bus down to synchronous
setting to the FSB as I mentioned previously, to have both running at
266.
>
>> Try moving the module in the 2nd slot, to the 3rd.
>
>Ok i've just moved it, not sure how it could help though, whats the theory?
>Am i right to keep the new bigger module in the 'higher' slot? Config is now
>DDR1: 512MB, DDR2: Empty, DDR3: 256MB.
The board should have termination resistors after the last memory
slot, but even so the memory bus "seems" to usually be cleaner with
the memory at both "ends" of it, leaving any empty slots in the
middle. Granted, it's usually not a large difference, but sometimes
enough. As you mentioned in another post, changing the memory from
"Turbo" probably helped as much or more.... always, always, always
return the board to default settings before trying to troubleshoot
memory. If the board defaulted to "turbo", it shouldn't have, is a
bios "bug".
>Will run like this for an hour or so until next crash...
It would take ages to determine memory instability in this way, and
you may corrupt data too... better to use memtest86 (linked
previously) to test for a few hours. Personally I am more agressive
than that, I up the FSB and memory bus speed (keeping within the
ceiling speed of the CPU of course) because I want to know what speed
causes errors, I WANT it to err... so I know how much of a margin I
have... If it only errs at >180Mhz for example, you have a pretty fair
margin at 166MHz.
>
>> Check the BIOS to verify that it's set to use "SPD" timings for the
>> memory.
>
>Can't find that setting. The only memory ssettings present are the bus speed
>and the voltage. Do you think +0.1 voltage may help? i know that can help
>overclockers stabilise their system. None of my stuff has ever been
>overclocked by the way.
It shouldn't be needed. If you can't find SPD setting option, then do
you have an "auto" or "manual" choice, where you get to manually
choose the memory timings? Letting it run in "Auto", or however that
particular bios words it, to not be choosing the values yourself, is
the SPD setting.
>> http://www.memtest86.com
>>
>> I would also remove the old module and test only the new, to verify
>
>Thats the next port of call.
It's an important step... but since the system seems stable with both,
odds are pretty good the new module is ok, yet perhaps it (or the old
module) has too little margin to allow running both modules at the
same time. In general the more memory you use, the better the memory
needs be, given the same motherboard.
>Its Micron ram from Crucial.com, they sell that stuff on reliability so i'm
>sure they'd take it back it was faulty.
>
A module can test perfrect fine yet the motherboard can be the
culprit, esecially when running multiple modules. It could easily be
that the older module is of lesser capabilites, that a pair of the
newer module(s) would work better, but again the motherboard may be
the weakest link, it may not matter which modules unless they're quite
over-spec for the application.
Dave
Similar Threads
