Stability issues w/ DS3 and 6400 OCed setup

[Cuhulainn]

My system (see below) runs fine at stock. I want to get the most bang for my buck and have begun overclocking. I'm going for 380fsb or ~3ghz clock. I am having trouble figuring out what is causing the stability tests (namely Orthos), to fail.

Some details:
Bios F6: all thermal options disabled, all unnecessary items disabled.
gmch: +.1v
vdimm: +.3v (to get to the 2.1v recommended)
vcore: 1.425v
Temps: ~30C idle, ~45C under load

Initially had memory at rated speeds of 4-5-4-15, loosened to 5-5-5-15 (no difference)

I can pass 2 runs of memtest at this speed with no errors. Orthos small FFT test runs no problems, but large FFTs or the blended tests both fail within 2 minutes. I thought I could rule out memory if it passed memtest. It seems that it would be the memory, as the cpu only stress tests run without error. Is this necessarily true?

This is my first attempt at overclocking and I feel like I've done my homework on this, but I'm pretty much stumped on why this is not stable. I have pretty much upped all appropriate voltages to the high end of all the oc's I have seen.

Any other suggestions???



EDIT: only to update title

 

[Cuhulainn]

There are three of these kinds of threads on here now, and no answers. I have reverted to stock settings again, and I am ready to give up with overclocking this system if no one has even a suggestion for me.

 

[Technonut]

Any other suggestions???

Instead of going straight to a high FSB, start out @ default clock and VCore. Slowly bump the FSB up and see how high you can go @ default VCore. (Orthos stable) When Orthos craps-out, bump the VCore up a notch, and start raising the FSB again. Rinse.. repeat... Keep the RAM speed and timings relaxed during this phase. You can play with those when you find your processors OC ceiling.

Stop when your processor maxes-out, or reaches whatever is comfortable for you as far as VCore and temps go.

EDIT: I forgot to mention to go ahead raise the chipset voltage from the start since you are going for a high FSB...

 

[davidos]

Try these setting Madellga gave me... I have the same memory as you and it booted fine at 2.1V 400MHz...

In BIOS second screen disable the thermal features (C1E, EIST, etc)

Overclocking bios settings (MIT screen) PRESS CTRL-F1 to show all options:
CPU Rate 8x
CPU Host ENABLED
CPU HOST FREQ 400
PCI-E 100 (or AUTO)
CIA2 DISABLED
System Memory Multiplier 2x (that's the 1:1 option)
DRAM TIMING MANUAL
CAS 4
RAS TO CAS 5
RAS PRE 4
TRAS 15
ACT ACT DEL AUTO
RANK WRITE AUTO
WRITE TO PREC AUTO
REFRESH TO ACT 0
READ to PRE AUTO
MEMORY PERF NORMAL (try also FAST or TURBO as you go above FSB400)
DIMM OVERV +0.3V
PCI-E OVERV NORMAL
GMCh OVERV +0.1V (or 0.15V)
FSB OVERV +0.1V
CPU OVERV 1.375V (try up to 1.40V also)

 

[Gary Key]

Run the memory at 1:1, go to 401FSB (365~399 the MCH is heavily overclocked requiring better memory and higher MCH voltages), MCH to +.15V, change to PCI-E to 103. Also, Orthos Beta has some issues with Conroe, run 3D06 three times, no failures, good to go for just about everything. However, that said, no issue getting 4 plus hours of Orthos on this board at 7x505.

 

[cmdrdredd]

As Gary stated it's better to jump right to 400Mhz FSB. No E6400 I've seen will give trouble at 3.2Ghz (400Mhz FSb with 8x multiplyer). Some chips need a little more voltage than others, but every one I've seen can do 400x8 easily.

 

[Technonut]

Originally posted by: Gary Key
Run the memory at 1:1, go to 401FSB (365~399 the MCH is heavily overclocked requiring better memory and higher MCH voltages), MCH to +.15V, change to PCI-E to 103. Also, Orthos Beta has some issues with Conroe, run 3D06 three times, no failures, good to go for just about everything. However, that said, no issue getting 4 plus hours of Orthos on this board at 7x505.

What issues does Orthos have with Conroe? I managed 24+ hours Orthos stable @ 3.6GHz (so far) with my X6800.... I was not aware that 3 runs of 3DMark06 is all that is required for stability testing nowadays...

I will still go with my usual hours of Memtest86+, dual P95 (Orthos), 3DMark loops, and multiple AutoGK encoding to deem my OC stable. It takes time running up the OC ladder, but I am certain that my OC is stable when completed....

 

[Cuhulainn]

Originally posted by: Gary Key
Run the memory at 1:1, go to 401FSB (365~399 the MCH is heavily overclocked requiring better memory and higher MCH voltages), MCH to +.15V, change to PCI-E to 103. Also, Orthos Beta has some issues with Conroe, run 3D06 three times, no failures, good to go for just about everything. However, that said, no issue getting 4 plus hours of Orthos on this board at 7x505.

In what is fast becoming a trend, I am unable to replicate the success of the good folks at Anandtech.

My system won't get past POST with those settings (401, much less 505), I just get auto restarts until it's back to stock.

The DS3, BTW doesn't have an option for mch +.15v. Just .1 or .2... neither do the trick.

So, what's the beef? Is there some way to isolate the problem? Did I just get a bad board? Is this a good enough reason for an RMA?? Please help!

 

[Cuhulainn]

frustrated bump

 

[darkhorror]

How far were you able to go with VCore of 1.325?

Also what do you have the FSB overvoltage set to?

 

[Cuhulainn]

I started out this process looking at the success other people had had with this setup, and thought I should be able to obtain the same results. Everybody said start at 400 with stock volts and go from there. The lesson that I learned is that even if "everybody" else is getting the same great results, it in no way implies that you will get those same results.

The system would not post at 400 no matter what bios setting I tweaked. But I was stubborn and refused to even try to start at say 300 and go from there. I got frustrated and gave up, went back to stock settings until I saw the AT article about the F7 bios and the improvements to overclocking that they had. I decided it was worth another try, but maybe a new approach.

Well I did, and here is a screenshot of the results.

Not a huge overclock, to be sure, but my main priority was stability for a system that I wanted to run 24/7 100% stable without sucking down too much power, so I'm considering it a success. BTW the vCore in the bios is @ 1.350v, and the only other voltage I changed was for the memory +0.3v to get it to it's recommended 2.1v. I'm comfortable with the voltage and temps, and I think I'm done o/cing...

for now. (as long as Orthos hasn't errored by the time I wake up in the morning.)

Thanks to all for your advice. Now to get back to the folding team.

:beer: