System completely unstable, even at 166MHz

[zstryder]

So for about a year I've been running at 200 * 11 - nothing particularly extravagant, as my RAM isn't ideal for any serious OCing. But I was able to push my XP-M a little higher than stock to get some extra oomph.

 

Recently I took my RAM out to help my friend test his RAM in my system using memtest - it was throwing all sorts of errors in his machine and he wanted to make sure it was the RAM and not his new mobo. The RAM tested fine, but as soon as I put my old RAM back in my whole system went nuts. Windows stopped booting, I had trouble posting, etc. Everything that could go wrong, went wrong.

 

I tried loading my old saved BIOS configs through CMOS reloaded - they all failed (it would load, but then the mobo would fail to display or POST). I cleared the CMOS and went back to stock speeds. 166FSB * 11.5 = ~1.9GHz, and also had the RAM running at 166. To my surprise, memtest threw errors at my RAM. After some experimentation using only single sticks and testing different DIMM slots, I came to a pretty bizarre conclusion:

 

If I am using more than one stick of RAM, I need to increase vdimm to at LEAST 3.0V.

 

Why? Good question! Running one stick of RAM, I can leave it at 2.6/2.7V. Using two sticks or all three of my sticks, required upping voltage to 3.0V to get it stable at a measly 166MHz. I never had to do this before when I was running at 200. To me this seems indicative of a serious problem. Does anyone have any idea why this could be?

 

Also, due to all the tinkering I've been doing (and crashing Windows on startup) Windows doesn't even start up any more normally - it will only start up in safe mode. Any clues on how to fix that? I'd rather not have to reformat unless necessary.

 

I'm just trying to get my system stable again, preferrably at 200 but if that's not possible, I'd settle for 166.

[zstryder]

Actually, that's a lie. Now my system is freezing on memtest at 3.0V for the RAM. What gives? Should I give it even more juice? All this for just 166... doesn't make sense.

[NEOAethyr]

I need more chipset/north bridge and mem voltage for 2 sticks as well.

Not totally sure why because I can run it in single chan, at least I could, without probs at lower votlages.

 

 

An example:

266, 3-3-2-3 9-13, 0-0-2-2-0-6-3, etc etc @:

2.8v single chan, any of my 2 sticks.

3.0v dual chan.

 

It's the same with north bridge voltages, 1.6v single, 1.8v dual.

 

 

 

Something I have found issues with:

 

If the slots are dirty, even a pci slot, as long as somehtign is in it.

Then you'll have some major problems.

 

Agp slot, memory slots, all of them are this way.

I know this because pop was spilled all over my rig.

So I had to fit with it for quite some time, now it seems like it's not an issue but it's still posible I'll run into such a problem again where I'll have to take out the alco and nail polish remover and do some cleaning...

 

 

 

One play to start is a lower multiplier, cpu wise.

Higher multi's stress the memory more.

To get probs out of the way, start at a lower multi, sometimes it helps to start at the lowest posible, 5X.

 

 

Power connectors.

This is'nt a huge issue I think system wise.

However it can cause major probs hd wise and such.

A nice tight connection is good ^^.

 

 

Cooling.

How long has it been since you redid the cpu's gel?

This can cause some major probs even at what you wold think is the lowest of speeds.

To much heat and everything goes to crap.

 

 

Power supply. (WARNING!, DON'T DO, you better understand what you're doing)

Dissconenct the power cord from it completely, hit the power button on the mother board.

It'll try starting up but won't.

Maybe even switching on a psu protection which will need fixing afterwards...

Now take out the psu, clean it out good, just blow the dust out I mean.

It should be fine to touch say the bottem of the caps, but do not!(This I mean bigtime)

I mean that, touching the caps would be totally stupid and perhaps deadly.

What I mean is that by discharging it before hard should make it safe enough to take off the cover and blow out the dust.

Don't doit though, you either do or don't understand.

If you don't understand don't doit ^^.

 

 

Other misc cleaning..., like fans.

 

 

I know this all sounds like it's all wrong.

Why shoudl you have to redo this or that?

The problem should'nt be in those places at all right?

Well it does not hurt .

If you don't understand the psu thing though blow it out wihtout opening it, use a can of air or somehting, in short bursts.

 

Short bursts ensure that you don't leave that resedue(probably bad spelling sorry)...

 

 

A good place to start though is the memory.

Get some alcohol, clean off the contacts real good on the sticks themselves.

Maybe even nail polish remover liek I do because I don't have 99 or 98% alco.

After the nail polish remover clean again with the alco to make sure you don't leave a coating from the nail stuff...

Blah blah lol.

 

 

Try diffrent slots.

Swap the sticks around, maybe you got them backwards compared to what you had them before.

Even if they are the same brand and everything, this can help ALOT ^^.

 

Bios and cmos...

Clear 2-3 times in a row, booting each time, save and exit, then clear again.

And Or re-flash the bios.

Sometimes flashing the bios again can suddenly get you that stability, I'm not totally sure of the real issue with this though.

I do knwo tha the cmos can be stubbern, and may need clearing even when it seems allright and working(but unstable).

As well as it sometimes not cleraring all the way and still being unstable.

 

Like when you setup say 10x 250, clear and notice the multiplier must still be set(those odd clocks inbtween multible clears ).

Or the cpu's voltage still at say 1.45v when the actuall board sets up a mobile at 1.55v or something for stock on those(1.55 or 1.6v I dn, I think it's 1.55v).

 

 

Anyways start with a bios clear, doit 2 times.

On the 2nd time, do'nt do the save and exit thing.

Instead go and directly flash the bios.

From here try that 200mhz memory speed, but at say 5x or something cpu multi.

If this does not doit, clean the meory and such and try again.

Be careful though.

 

Mess around to much while you're unstable and you're gonna screw over the bios.

Perhaps this is what you did in the 1st place with your friends memory .

Partial bios corruption, or a really messed p cmos that does'nt want to clear on a single clear(never does though).

 

Good luck dude and report back.

 

 

Edit:

Stop witht he memory voltage increase though.

You're gonna screw somehtign up doing that.

2.8v should be fine for a starting point unless you got memory that specificly is known to be a hog.

[CaysE]

NEOAethyr, take a look at this thread about cleaning motherboards, it should help you out: http://www.badcaps.net/forum/showthread.php?t=1463

 

As for the issue at hand, have you tried, say, your friend's memory sticks in your machine? Be sure to do a CMOS clear before swapping any hardware.

[zstryder]

Well, I actually DID reseat the CPU and clean the dust etc. everything when the basics failed. I just tried cleaning the memory contacts - no difference. I let memtest run overnight though, looks like 3.1V did the trick and now it's booting into Windows. :eek2: What kind of long term effects are we looking at if I leave vdimm @ 3.1V? At least it is stable at 166 now. I'm afraid to see what would happen if I try to go up to 200...

 

Unfortunately my friend went out of town and so I can't get a hold of his memory for now.

[charlie22911]

make sure cpc (command per clock) is off in the bios.

[zstryder]

Okay, I got my system stable at 166, at 1:1, in dual channel mode. Seems to be doing fine in prime95 so far... vdimm is still 3.1V.

 

I've given up trying to get back to 200MHz. I'm not sure, but for whatever reason the RAM won't stabilize at 200. The CPU seems to be fine, but the RAM throws errors at 3.1V, and turning it up to 3.2V or 3.3V just causes memtest to freeze during testing.

 

Question though.. at 166, will it make much of a difference if I push the multiplier even higher? Like say, 14 or 15x? I'm thinking the higher FSB would make a much better result...

[loggan26]

Loosen the timings up and try the CPC off like was suggested?????

 

Also have you tried slots 2&3 as these throw out less errors than slots 1&3......

 

End of the day your one or both of your ram sticks might be knacked , so it might be rma time or replacement if you want to run higher fsb, off course this is if you have exhausted all other avenues in testing ram.

 

FSB is king at the end of the day but if your cpu can run a higher MHZ and your happy with the temps etc as more MHZ will need increased vcore etc then i dont see why not...= ie higher mulitplier....

 

At end of the day if your happy running 166fsb on a board that has the potential to run 250+fsb with the right ram then thats your choice man .....Me i would be looking at a way of sorting my ram problem......Thats just me though......

[zstryder]

Yeah, tried the CPC disabled (even though when it was running 200 before it seemed fine with it enabled) and tried 3-3-3-8 across the board.

 

Problem is, I'm running dual channel with 3 sticks of RAM, not two. So all 3 DIMM slots are taken. As my sig indicates, it's 2x512 for one channel and 1x1GB for the other. Which I know is probably hard to get the board to play nice with, but I figured it's worked before so why not now?

 

I can run 200 if I'm only using one stick.. two sticks or more and it doesn't want to boot. I memtested each stick individually, each one comes out clean... I'm just wondering what could have changed between now and then that would cause my system to not even run 200.

[loggan26]

Maybe a ram stick copped out who knows man, hey even a clean install might sort the prob, it might be software related who knows the only way to know is troubleshoot everything and end of the day if that is the answer your left with then that must be the answer....Yes elimination through trial and error is a big part of OCing,,,,,,,,,,,,,,crap but true

[NEOAethyr]

CaysE, Dude, this tech is freaking cool!

I did it last night and wow, cleaner then stock, now I can run lower timings, super fast bootup, etc.

Even 278mhz fsb memtest stable without messing the the drive strengths and slew rates.

Thanks very much dude for showing me that .

 

I did it to my dial up modem, my mobo, my cpu, and 2 sticks of memory(my mom's tcc5 and my adata).

I was gonna doit with my quadro as well because it was the card in my rig at that time of the pop spill, but since it has a rom on it I figuered I'd better just wait and see if my board still works after this.

 

I can doit later, boot off a pci card and flash the quadro if I have to.

If the rom acts up I can allways stick the quadro by a magnet to clear the rom, which I learned from modding the 6600gt's strap id's from it's rom...

 

Nice nice , hehe

There's actually a ton more diffrences but I'm not gona hijack the thread over it.

Very cool thing though, I'd consider it a black art, like breakin.

Could be dangerous if you rush it.

 

 

zstryder

Name is familiar...

 

Anyways try this.

1gig stick in slot furthest away from cpu.

Then place one of those 512m stick in the next slot closest to the 1gig stick.

 

Like this:

cpu

 

1

2 512m

3 1gig

 

 

If this fails, replace the 512m stick with the other 512m stick.

I would hope this works because it's likely the only way it's gonna really work out I think.

[SpursIndonesia]

How about cleaning your Memory stick golden grip ? It can help you know, cause i've had some memory problems that are solved by doing this.

 

Or perhaps the slots themself are quite dirty ? Try to clean them with a thin strong stick covered by a soft clean lint free fabric soaked with Compact Disc cleaner foam/solution, it can do wonder (in my case, i gain around 20 MHz stable OC over memory)