Which to rma the memory or the motherboard?

Plec

Plec

Plec

Soldato
Joined
19 Apr 2003
Posts
13,567
OK, I had faulty gfx card, xfx 4850 xxx, that I have returned and since replaced with the one below in my sig which is working fine. This was causing lockups in games and furmark and, if you’re interested, it’s well documented in this [thread=17997995]thread[/thread].

While trouble shooting the card I noticed that the memory would become corrupt and stop programs from installing and sure enough when running memtest I had thousands of errors. However, this was always cured by resetting the CMOS - memtest would run fine until a systems crash (which I wrongly thought was the cause of the memory becoming corrupt).

I assumed, wrongly as it turns out, that the re-occurring corrupting memory was a result of the system crashing all the time due to the faulty gfx card – but this doesn't appear to be the case. As since installing the new perfectly working gfx card (tested) I’m getting re-occurring memory problems without the system crashes. But, again, after a CMOS reset I can get memtest to run fine for a time until, for some reason, it becomes corrupt again.

Basically, my memory over a period of time becomes corrupt but can be cured with a CMOS reset - until it decides at a whim to corrupt again. So is this the motherboard memory controller or the memory itself? (I have tried the memory in slots 1 and 3 and also 2 and 4 – I get the same problem)

I’m going to be testing it throughout the day – turning the machine on and off, unplugging it, running apps, benchmarking etc – all the while running memtest in between. I would be very grateful for any advice, experience or links to other threads that I may have missed.

I really could do with some sage advice over this bizarre problem as I’m totally demoralised with this build. I’m trying very hard not to get stressed about it as it’s only a bit of hardware but I’m getting very annoyed about the time I’ve given over to it already with regard to the gfx card. But now I’ve found that I have yet more problems I can feel my blood pressure rising…

Totally demoralised.

Plec
 
Have you set the memory timings or are they on Auto? If Auto then set them within the BIOS and see if it improves things.
 
Tried a single DIMM at a time? It may be one of your sticks playing up.

If it only happens with one, not the other, it's the memory. If it happens with both it's most likely the memory controller - the chances of you getting two duff sticks is fairly low unless it's a bad batch.
 
IAMATEAF, it's on auto but the settings are correct for the memory - but i'll try it out.

Thanks for the reply.

Plec
 
Last edited:
Thanks for the reply Audigex, i'll try that over the coming week after i've exhausted the limited amount of tests i've got to do so far. At the moment it's behaving itself but i'll have better idea after a couple of days. (i have tried single dimm but not over a long period of time - just a memtest after a cmos reset.)

Sound advice, though, and it should eliminate any doubt as to what the cause is - but like everything with trouble shooting components it's very time consuming - but should be worth the effort. (I wish i had a second rig for testing dodgy components.)

Thanks again for the reply.

Plec
 
Last edited:
Plec, memory doesn't corrupt in the way you are thinking, once it goes its gone. It doesn't suddenly start working again once its properly borked.

There are other factors that can cause memtest and windows errors like you are describing and they are *Thermal* issues with either the Memory sticks/Northbridge, also there is a possibility that the voltage isn't being set correctly for either the memory or northbridge?

Seems a bit strange the problems you are having, my first guess is that something is maybe overheating? :confused:
 
Update: It ran absolutely fine all day yesterday after a CMOS reset. I turned it on and off, installed software, ran benchmarks went into the BIOS etc. However, after being switched off all night I booted up the system with memtest and I’m now getting over 3000 errors in the first 3% of testing. I am so confused…

Thanks for the reply Big.Wayne.

The fact that i can get the system running again after CMOS reset is the thing that is confusing me too. Like you, I thought that if the memory was faulty then it’s faulty - no matter what you then try to do to correct it. (Unless obviously the BIOS settings were set incorrectly etc.)

I’m going to try manual settings and upping the memory voltage even higher and see what happens. After that I’ll be trying the one stick method in each dimm but all this is consuming time, effort and patience I am fast running out of. I may just rma the memory and get it tested – let someone else rule it out for me.

Plec
 
(Please see above post too - thanks again Big.Wayne - it may be voltage related in some guise.)

OK, this may be a bit more definitive:

This time i did not reset the CMOS instead i loaded optimized defaults in the BIOS set the memroy voltage at 1.9v (as usual) and lo-and-behold the system is running memtest again. So I don’t need to keep resetting the CMOS – so it may be cold boot voltage problem?

So it may be a cold boot issue, maybe - it's early days. However, i have had parts in other computers that required the computer to be fully warmed up before they operated properly. I may need to up the voltage to 2.0v on the memory - but i'll see if the rig runs smoothly from now on this morning. I'll turn it off for an hour this afternoon and see if i can replicate the problem when its booted again.

Any experienced advice would be appreciated.

Plec
 
what mem are you running?

The last Geil stuff I bought had a warning sticker on them telling me to set the voltage to 2.2v, dunno how well they would operate at default volts.
 
Thanks for the reply.

Corsair XMS2 4GB DDR2 PC2-6400C5 TwinX (2x2GB) - but it defaults as 1.8v but their website states it should be set at 1.9v and that's what i've always used. I've just experimented with 2.0v but this hasn't made any difference - i may try higher.

I can't get it to self correct itself now via the BIOS so i'm going to have to resort to clearing the CMOS again to getthe memory stable again - so i am definitely going to have to rma a part as i can't keep doing that just to get a stable system.

I'm going to try the BETA BIOS but i'm clutching at straws now. It only. seems, to become corrupt once the machine has been turned off for more than 5 minutes - which stinks of a voltage problem. But is it the memory or the motherboard?

It's easiest to rma the memory i suppose - but i have a sneaking suspicion it may be the MB.

Plec
 
Flashed the BIOS to the latest BETA version - no difference. I'm going to let the system warm up for 10 minutes and see if i can clear the problem via the BIOS and loading optimal defaults. If it doesn't and requires a CMOS reset then i'm going to rma the memory i think as it's the easiest thing to get checked out first.

I may put an rma on the MB just to be on the safe side - otherwise i may run out of my 28 days by the time the memory is tested. (bought from 2 different companies.)

Plec
 
Sounds like the board to me.

If it were the memory it'd be broken full stop, fiddling with the BIOS would do nothing.
 
Plec slow down a bit before you start RMA'ing stuff left right and centre! :p

I'm trying to re-read your posts but its like you drunk too much coffee! :D

I'm not even sure what hardware you got? what is new what is used, how old etc?

Hardware doesn't need to *warm up* to work properly heh . . .
 
HeX - it's looking that way.

I've just tested both sticks independently in one dimm slot. Each stick has passed the mem test - and that was without resetting the cmos before i tried the test.

I have even left the machine turned off for over 30 minutes and then booted up in memtest with each stick independently and again both have passed.

So basically i'm only getting these errors when both sticks run together - and i have tried 1 and 3 and 2 and 4 combinations.

So does this definitively prove that it's definitely the motherboard or could it be that the board and memory combination, in dual mode, just don't get on?

Thanks again for the input guys - it's really appreciated.

Plec
 
Big.Wayne said:
Plec slow down a bit before you start RMA'ing stuff left right and centre! :p

I'm trying to re-read your posts but its like you drunk too much coffee! :D

I'm not even sure what hardware you got? what is new what is used, how old etc?
Click to expand...

Thanks for the reply Big.Wayne - I can imagine they do read that way as i'm cramming in testing and posting inbetween doing my work and now i have kids added to the mix as they've just finished school/nursery. I don't need coffee i need valium.

Anyway, the motherboard, memory, gfx card, and cpu are new (see sig). All the other components are from my old system.

I had to return my original XFX 4850 xxx edition due to it crashing the system all the time and so i presumed the memory errors i was getting were due to that - i was mistaken. (The card was definitley faulty it was tested by OC's and others have had same problem.)

So, basically motherboard, memory, cpu and now replacement gfx card which seems to be running perfectly.

Since my last post i have tried the following:

I've just tested both sticks independently in one dimm slot. Each stick has passed the mem test - and that was without resetting the cmos before i tried the test.

I have even left the machine turned off for over 30 minutes and then booted up in memtest with each stick independently and again both have passed.

So basically i'm only getting these errors when both sticks run together - and i have tried 1 and 3 and 2 and 4 combinations.

So does this definitively prove that it's definitely the motherboard or could it be that the board and memory combination, in dual mode, just don't get on?

Should i try facing it to the north or place it on a Leyline(sp) - it's that tempremental in wouldn't surprise me. :)

Thanks again for the input guys - it's really appreciated.

Plec
 
Last edited:
Ok lets start from the beginning . . .

You bought this new hardware:

  • Intel Dual-Core E5200
  • Gigabyte GA-EP43-S3L
  • Corsair XMS2 4GB DDR2 PC2-6400C5 TwinX
  • Asus 4850 512 Glaciator
And whacked it together and straight away had problems. The first problem you say was a faulty graphics card which you RMA'ed and recieved a new one which worked?

So it seems the new computer hasn't actually worked properly since you built it? (one thing or another?).

A workaround you found to the Memtest error problem was to either reset CMOS or now just load CMOS defaults?

Each stick passes memtest by itself? how long are you running memtest?

If the sticks pass by themselves (if memtest has been left to run enough) but then fail in Dual Channel it sounds like the board needs to be configured correctly, although the board shouldn't really have a problem?

To give yourself an hours peace I would run the system with just one stick of ram and see how you get on, there may still be a problem lurking and waiting to strike.

What PSU are you using and how old is it?
 
You must log in or register to reply here.
Back
Top Bottom