What would make an OC unstable?

[crydee]

I have a ASUS P5B-Deluxe with a E6400 running at 3.4GHZ (425x8 1.4v with vdroop) with some Crucial 10th anniversary running at 5-5-5-15 (2.3v) ddr2-850 (I had it running at ddr2-1000 fyi before). It was 24 hours orthos stable and like 16 hours blend stable when I just built it and OC'd it.

Now during an intense gaming session was going 2-3 hours the game crashed then my computer would just randomly reboot. After the reboot I would just get the game crashes on load over and over so being frustrated I lowered my OC to 400 fsb and same results. I ran orthos cpu stress test fine for a few hours so figured that put my cpu through more then the gaming I was doing.

Then I tried orthos blend and it wouldn't go past 4 seconds heh even with slight more voltage to the ram and the fsb lowered to 400 so it was running at ddr2-800. It also seems like a software problem kind of due ot the game not even loading after the forced reboot.

Any ideas?

 

[myocardia]

It sounds to me like either your video card overheated, or your power supply is needing to be replaced. I'd definitely say that it's one of those two. BTW, you don't need anywhere near that kind of vdimm, for 425 Mhz, with your RAM, at those timings. You should lower it.

 

[Arkaign]

Myo is probably onto something, his posts are generally spot-on.

Psu is culprit #1, but at that voltage, your ram may be slightly smoked.

 

[crydee]

I have a Corsair HX-510W and Asus PC Probe shows my voltages within .01 of what they should be. I will lower my Vdimm. I upped my fsb back to 425 and ran orthos cpu for awhile and it was fine and now I'm on 15 minutes of blend no problem after several attempts of not making it past 4 seconds earlier.

I'm fairly confident my video card didn't over heat as that isn't even overclocked right now and it's an x1950pro with the accerlo x2 cooler from power cooler and atitool has never shown that to get above 47ish during long gaming.

Now I'm confused because blend is working just fine now =(

crucial 10th anniversary is designed to operate at 2.2v why would it be smoked btw at 2.25?

 

[Arkaign]

Originally posted by: crydee
I have a Corsair HX-510W and Asus PC Probe shows my voltages within .01 of what they should be. I will lower my Vdimm. I upped my fsb back to 425 and ran orthos cpu for awhile and it was fine and now I'm on 15 minutes of blend no problem after several attempts of not making it past 4 seconds earlier.

I'm fairly confident my video card didn't over heat as that isn't even overclocked right now and it's an x1950pro with the accerlo x2 cooler from power cooler and atitool has never shown that to get above 47ish during long gaming.

Now I'm confused because blend is working just fine now =(

crucial 10th anniversary is designed to operate at 2.2v why would it be smoked btw at 2.25?

Maybe because you listed 2.3v in the first post? Besides, it's enthusiast O/C ram designed to get crazy high FSB, and it's stock voltage supports that. It's default voltage is already extremely high for memory.

 

[crydee]

It was running at 2.25 all this time with 2.3 only because when I had the problem and blend wasn't working I tried to up it. P5B Deluxe goes from 2.15 to 2.25 in vdimm voltage and windows won't even load at 2.15.

 

[jkresh]

sometimes a processor run close to its maximum over clock overtime will start to deteriorate and the maximum overclock will slowly drop (when this happens you can either raise voltage and keep going (which cuts the life even more) or dial back the clock a little). If everything seems stable then let it run orthos for a while and it may have just been a flook. If you dont have anything cleaning the power before it gets to your pc its possible that something was going on with the power to your house (minor brownout?) that caused the problems and it resolved itself.

 

[n7]

RAM i bet.

There have been lots of reported deaths with the Ballistix...my guess is yours is slowly dying.

That being said, my previously Orthos solid 3.4 GHz isn't anymore either.
After a couple months, it degraded to only being able to do around 3.36 GHz for hours on end. I still don't know if it is due to CPU or mobo or RAM though.

 

[crydee]

what would you recommend for stable power to the pc? I really hope it isn't the ram that crucial is supposed to be good stuff but everything points to it. I don't supposed oc'd ram is covered by any warranty is it?

 

[jackwhitter]

since orthos blend failed and not orthos cpu, i would say your ram is the culprit. ram can die, and running at voltages that high will likely kill it faster (i know ballistix is supposed to be rated for that high, but that does not mean it can run it indefinately). either way, crucial has a lifetime warranty. it is possible your power supply is dropping voltage levels too. run super pi and super pi mod 1.4 and see what they show.

 

[Duvie]

Originally posted by: crydee
what would you recommend for stable power to the pc? I really hope it isn't the ram that crucial is supposed to be good stuff but everything points to it. I don't supposed oc'd ram is covered by any warranty is it?

If you ran it at the rated speeds and vdimm they said would work...heck yes it is covered...

Make sure your case cooling is still good....dust on fans inhibits airflow...raises case temps, etc...it only takes a few degrees to make something unstable....Trying blowing a fan on the ram and see if that helps...

My gskills acted like that occasionally...the heatspreaders were placed on by a moron...and really a sloppy job...

 

[crydee]

Super Pi mod I get an error ... "not exact in round" my +12 was showing 12.09 my +5 was showing 5.02 and my +3.3 was showing 3.31.

 

[jackwhitter]

your voltages look good. does normal super Pi throw an error?

 

[crydee]

No my time for 32M places last time was 15minutes. Is it possible I have a virus that could be making everything act all funky I would hate for this to end up windows related. I could run superpi to about 1M if I go beyond that I get that error. My case is a p180 with dust filters and everything looks cool and clean inside powered it down and going to reset the ram but it feels only slightly warm to touch on heat spreaders.

 

[Duvie]

well why hasn't anyone stated this...

Download memtest86+....dos version...make a bootable CD and run it....it will find errors in the ram if they exist

 

[jackwhitter]

super pi Mod throws an error, but does normal super pi? i've had a couple times where super pi Mod throws errors on perfectly stable machines whereas normal super pi does not.

 

[crydee]

I didn't know there was a difference in super pi. I am running memtest right now what should I look for?

I got 5 errors after 30 minutes at 2 failing address and I'm just letting it run whatever it does on default bootup. Does this mean I need to lower the speed on the ram maybe set it to the default specs ddr2-667 3-3-3-vs ddr2-800 5-5-5-15? Or is this ram gone bad? I am assuming im SOL with warranty since I OC'd it? Right now the vdimm is at 2.15 when spec is at 2.2.

update: 13 errors after 47 minutes oh god =(

 

[legendma]

Originally posted by: crydee
I didn't know there was a difference in super pi. I am running memtest right now what should I look for?

I got 5 errors after 30 minutes at 2 failing address and I'm just letting it run whatever it does on default bootup. Does this mean I need to lower the speed on the ram maybe set it to the default specs ddr2-667 3-3-3-vs ddr2-800 5-5-5-15? Or is this ram gone bad? I am assuming im SOL with warranty since I OC'd it? Right now the vdimm is at 2.15 when spec is at 2.2.

update: 13 errors after 47 minutes oh god =(

Therefore you go. Ram definitely looks like the problem. Try putting it to the default speed and see in memtest86+ still errors out.

 

[jackwhitter]

i'd use some different ram that you know is good from another system if you have any. if you don't, i'd call crucial asap. mayhaps they have an advance rma you can use.

legend may be correct in lowering the timings to get the ram to run stable, at least for a time. you should be able to run overclocked with a memory divider and lose little performance until you can get the ram rma'ed.

 

[crydee]

I lowered it to a 333 fsb and the timings to 4-4-4-15 and left the vdimm on 2.15. I ran test #5 for 2 and a half hours while I was at class with no errors. So maybe my 10th anniversary is just dieing and can't OC for poo anymore? Or was the other test at ddr2-800 5-5-5-15 a big enough indicator that I should try to RMA this stuff?

 

[jackwhitter]

set the to exactly its rated timings (and voltage) and see if it causes errors. if it does not throw errors at the advertised/purchased speed, then you will not likely be able to rma it. from what i could tell from reviews, it should do 333 fsb @ 3-3-3-12 timings @ 2.2 volts. if it can not do this successfully, then RMA time. 4-4-4-15 is NOT what you paid for.

 

[crydee]

Ok, the P5B deluxe only offers 2.15 to 2.25 vdimm though should I set it to auto for the one and all test?

 

[jackwhitter]

set to 2.15 first, and see if it passes. if not, try 2.25 and see if it passes. you should not *need* to run above 2.2 to get it to pass, but boards are not exactly what they say (2.25 may actually run 2.21 or something). if neither pass, you should definately seek an RMA. if only 2.25 passes, you still might want to try. sounds like the ram is failing.

 

[Dolorous Dave]

When Orthos errors immediately for me, it's because my RAM timings are fracked. I'm guessing your RAM might either be fried or faulty, like the others suggest. Have you tried running Memtest86? Over on the hard forums there are a number of people having problems with the Crucial 10th Anniversary RAM.

 

[Arkaign]

Originally posted by: crydee
I lowered it to a 333 fsb and the timings to 4-4-4-15 and left the vdimm on 2.15. I ran test #5 for 2 and a half hours while I was at class with no errors. So maybe my 10th anniversary is just dieing and can't OC for poo anymore? Or was the other test at ddr2-800 5-5-5-15 a big enough indicator that I should try to RMA this stuff?

Yes, try to RMA. Don't mention your abusive voltage though. I'd go easy on your next set. Ocing is cool, but it's not worth killing hardware to push it so far. Find max stable o/c, then back off voltage/fsb about 5% off of that for a good safety measure. It's not going to make that much difference in performance, and the peace of mind + lack of downtime is def worth it. Cheers :thumbsup: