First off, this is not about the DS3L rebooting at POST with an unstable overclock. I actually like that feature.

Specs:
P35-DS3L (obviously)
E2160 M0 stepping
Ballistix ddr2 800 2 x 2gb ram (slot 0 and 2)
pc p&c silencer 750w

FSB 400
Multi 8x
SPD 2 (1:1 ratio, 800mhz)
Auto timings (5,5,5,18,30,2t)
tread 11 (standard)
ddr volt: 2.2
fsb volt: +0.1
mch: stock
cpu: 1.4

The problem I have is the machine rebooting without displaying an error when running Prime95. Those numbers are for the last run I made. I ran Prime95 (v25.6?) small FFT for 7 hours. Sometime around then the machine rebooted. I checked Prime's results log (no errors), application event log (nothing), system event log (nothing). Auto updates were turned to notify of updates (not automatically restarting from update from Microsoft).

CoreTemp, HWMonitor and SpeedFan (not calibrated) all reported max load temps at 55-57. The Temp1 in Speedfan was listed at 42-43.

I have upped the voltages for cpu, fsb and mch and no help. The machine either doesn't boot, BSODs before entering windows or reboots sometime during Prime95 testing. In fact, when MCH was upped by +0.1 or +0.2 the machine wouldn't post.

I'm pretty certain it's not CPU heat, or RAM. I know there has been some people saying that Crucial has gone downhill, but the ram should be stable at stock (with rated voltage). I have not run MemTest86 yet, though.

I'm wondering if this might be a heat issue on the NB, or some other issue. Is there a program to monitor the NB temp?

System was stable for 8+ hours at fsb 333, 9x multi. If that's what I'm going to get, so be it.

Thanks.

Can't help much, but I'll bump your post.

Ask on the p35-ds3l overclocking thread http://www.tomshardware.com/forum/ [...] erclocking
your problem might get more views there.

Maybe you've hit the fabled e2160 FSB wall. Many people have reported that it is impossible for their e2160 to exceed FSBs of as low as 370, while others can go much higher. You mention you got 333x9, did you try 356x9 to get you to 3.2 ghz? or did you go straight to 8x400?

I have a similar configuration - e2160, 8x400=3.2ghz@1.35Vcore, Ballistix ddr2 800 2 x 1gb, SPD 2 (1:1 ratio) timings 4-3-4-11@2.1V, tRD (static tread value on the ds3l)=7. mch and fsb voltage at stock. It's stable, so fsb wall is all I can think of.

NB temp shows up on HWmonitor and speedfan. I really doubt it's your problem though, your cpu load temps are really good so I'm guessing your case airflow etc is just as good. In any case NB temps couldn't possibly stop you from booting.

you need to put +0.3V on both FSB and MCH.plus you dont need more then 2.1V for 800mhz with CAS 4.too much voltage on the Ram will make it unstable.1.4 vcore seems reasonable.

i was helping a guy out with the same mobo and cpu he couldnt post at anything higher then 275x9.333x9 the bios just reset itself.

make sure you have the latest beta driver for that board.F7 bios i have seen lots of problems for overclocking.

Interesting. Last night I was able to run at 8x400 reliably. +0.1 FSB, 1.4vcore, 2.1 or 2.2 ram, stock mch. When I checked this morning, the machine had a BSOD (running Prime95 overnight). However, my wireless router was also messed up so I don't know if I got some sort of power fluctuation from the earthquake this morning.

I'll update the BIOS. I can't change memory timings at all with stability, so they're all running at auto for the moment. I'm going to go searching, but can anyone point me to a memory timings guide (for the full set of timings, not just the primary 4).

Also, last night I had some temp discrepancies. CoreTemp, RealTemp, SpeedFan all showed CPU temp topping out at about 57. HWMonitor was up at 72-73. Very odd, but I'll go with 3 vs 1 on the temps.

Last thing that seems weird, is that I can't change MCH voltage at all. Any adjustment (+0.1, 0.2 or 0.3) all cause a failure to POST. Good thing is that everything seems to be working at stock MCH, aside from the random restart.

I guess I would be more relaxed if my box was throwing some sort of error, rather than just rebooting. If I knew I was unstable that would be good, instead of this unknown issue.

I did make sure the NB heatsink and CPU heatsink were seated tightly. I was going to put an extra 80mm fan I have on the NB (somehow), but there just isn't room in the current setup. I might rotate the CPU HSF to point up rather than towards the back. That'll put the NB heatsink and GPU inline with the pull from the CPU HSF, and might help cool both of them.

dude!if your pc crash/freeze/restart/BSOD means there is something wrong with you overclock.what you can set doesnt mean it can run on those speed.after 24hours of small ffts and 24hours of blend test no error then you can say its 100% stable.but thats for your FINAL OC.normal oc judt do 8 hours of small ffts and 12 hours of blend test will do.

it would help if you post some screen capture of cpu-z on on all its tab.

I realize that if it is rebooting that it's not stable. What I'm trying to find is what that particular issue might indicate.

What I find odd (and that may just be a mistake on my part) is that the reboot won't happen for hours, and temps are stable.

Are your Ram modules in the 1st and 3rd or 2nd and 4th slots?
You said slot 0 and 3.
I know this is very trival if its just mistaken nomenclature.

1st and 3rd. I'm a software developer so I tend to think in 0 based list terms.

Also, I'm going to edit my original post to reflect that I meant slot 0 and 2. Thanks for pointing that out. Stupid mistake to make.

@OP: In order to up the MCH voltage you need to add a cooler to the NB. You won't need to up the NB voltage. The Max OC for the E21x0 CPUs are about 3.25Ghz without really high voltages (past 1.48).

Well, updating to the F8b BIOS allowed me to reach stability with 400x8 OC. Now on to the RAM.

the older bios on any Gigabyte mobo seem to have problem in overclocking.even the new Beta bios will improve things.

a thing to consider when buying your new mobo huh?

Yeah, it's one of those things I wish I knew before buying...all of a month ago. That's what I get for listening to everyone saying that it overclocks like a champ (actual words).

I am happy with it, just wish it were a little more stable. Then again, I am running it past spec. I wish I had gotten an X38 board, though.

Interesting note, for those that are interested, here is a screen shot with my current setup, including memory timings. Those are the biggest gotcha with this system, because only setting the first 4 seems to not work that well. I went with "Extreme" for the auto memory settings, and this is what I have. I'll find out in the morning whether it's stable, but so far, so good.

nice set of setting!go and put it on 450x8 see what it can do!

Hmm...450x8 would allow me to run memory at 1:1.
Edit: My math skills are apparently failing me.

Not sure about heat though. Currently maxes out about 58. Also, I have the voltage set to 1.4 in the BIOS, translating into about 1.36 in CPU-Z.

Should I be worried that HWMonitor reads my temps much higher than the 3 programs running above (SpeedFan -not configured-, RealTemp -also not configured- and CoreTemp)?

if coretemp say 58 then you gota HUGE margin.try 450 or at least 425.loose the memory timing to 5-5-5-15 just incase.vcore to 1.45v.try it out.it wont hurt anything.

Actually, at this point I'm allowing the BIOS to control timings. I have yet to be able to set them reliably (possibly because you can't just set the main 4, but have to set all of them).

Tonight I think I'll see what the FSB can reach. Although it may wait, since I haven't had time to use the box since building it.

Damn, just talked to my wife, and BSOD after about 12 hours Prime95 blend.

So, back to looser memory timings. Perhaps push the FSB, but I'll try and push memory speed more.

Keep in mind that you need to turn off all auto updates for Windows and Virus protection before you start an overnight Prime run. Some people leave there virus protection on stock settings which means during a 12+ hour Prime run, your virus protection could be starting a scan. This can also cause a BSOD or instability. Personally I think it becomes a memory based crash due to the virus protection or whatever not being able to allocate enough memory.

When you run a blend test you can adjust the amount of RAM being tested to accommodate for the running processes using some of your available RAM.

I turn everything off and unplug from the internet. I've reduced my BSODs during Prime by over 50% when running overnight.