Curious RAM behavior: A study.

[lhgpoobaa]

My workhorse Epoxy 8K3A+ (KT333 chipset) has 1 stick of XMS3200 Cas2.5 512Mb Corsair ram in slot 1.
The Ram and the system bus are running synchronously at 167Mhz.

When i first got the mobo and ram, I found something curious with Memory timings: Turbo memory timings were fine at stock voltage up to around 170fsb, but in order to run 1T command rate required 2.6V or better. Even down at 133Mhz.
Stability was determined by multiple MEMTEST86 passes.

Fastfoward a number of months to last week, when my system started displaying unusual buggy
behavior and the odd BSOD.
Initially i thought it was other Overclocking factors at work, but last night i ran memtest to just make sure it wasnt the ram.
Low and behold, BINGO. memory errors showing up within half a minute.

Doing some more experimentation last night it seems that my memory has become "worn down",
and it won't run bug free with the 1T command rate, at ANY voltage or Mhz.
Can still run all other timings at Turbo though.

How very curious.

Anyway its good to know i found the source of the glitchy behavior.


<b>Americans generally do the right thing, after first exhausting all the available alternatives - Winston Churchill</b>
<A HREF="http://members.iinet.net.au/~lhgpoobaa/images" target="_new">My Images!</A>

 

[marneus]

At least U got errors, overnight one of my crucial memory chips went from 256mb to 128mb... the case was closed & I hadnt fiddled with the bios recently... no overclocking either...

but hey, it was still under warranty...

Hmmm, wonder if I can get a valid page fault ???
these invalid ones are far too commonplace...

 

[lhgpoobaa]

True enough... and the fix was easy, just slip it back to 2T command rate.

My old Kingmax went like yours did. 2x256 sticks of PC150. I booted up one day and lost some 32mb of ram.
I blame the dumb asus a7v133 mobo... made the DEFAULT ram voltage 3.56v! proly burnt it out.


<b>Americans generally do the right thing, after first exhausting all the available alternatives - Winston Churchill</b>
<A HREF="http://members.iinet.net.au/~lhgpoobaa/images" target="_new">My Images!</A>

 

[marneus]

3.5v... gee, I think the phrase 'smokin' would be apt ;-)


Hmmm, wonder if I can get a valid page fault ???
these invalid ones are far too commonplace...

 

[papasmurf]

I have had the same problem with my bros old stick of samsung 2100, the thing would run at 166mhz cas 2 no errors at default voltage bug free then it started to have errors and I had to gradually increase the voltage each month lol it finally died last month but it served me well and did one hell of a job for a stick of cas 2.5 2100 to be able to run at cas 2 3200. I just wonder why it died, I didn't bring the voltage that high and I had some heat spreaders on it.

Treat your body like a $600 car. God didn't intend it to last so use it. Run it into the ground!

 

[lhgpoobaa]

Yep. Was a nasty board.
3.56v for the sdram and 1.80v for the Vcore.
THATS how asus boards achieve stability.

<b>When life hands you lemons, ask for a bottle of tequila and salt</b>
<A HREF="http://members.iinet.net.au/~lhgpoobaa/images" target="_new">My Images!</A>

 

[tRiXtA]

i have seen pins burned off ram modules that cause the physical amount of ram to chasnge... its weird eh.

<b>The pentium 2 I once adored is dead! Long live my barton savagely ripping 1561 fps in soldier of fortune!!! </b>

 

[EvilHammie]

*squeak*
[anyone know how important 1T command rate is?]

<b>Friend of Fredi and his Cables. </b>