Problems with E8400 > 3.6GHz

dechy

Distinguished
Mar 7, 2008
227
0
18,690
I'll start by saying that I've read countless threads about overclocking this relatively standard setup (cpu+heatsink+mb). So many people use the P5K-E, E8400 and CNPS9700LED heatsink to excellent results.

Now, I'll also state that I'm extremely happy with the CPU running stock (MASSIVE upgrade from my socket 939 4000+!), even happier with a 600MHz overclock needing all of 30 seconds worth of BIOS settings (following the C2D sticky'ed guide)

All I'm looking for is the possibility to get higher in a safe way, here's what I did in the BIOS:
Locked Multiplier to 9.0x
Locked FSB to 400MHz
Locked DDR to DDR2-801MHz, so 1:1 (at stock FSB ot 333, DDR2-800MHz appears, but at 400MHz, it's 801. Didn't think any of it)
Locked PCIE to 100MHz
Locked Memory Timings to 4-4-4-12
Disabled Vanderpool & C1E
Disabled all Spread Spectrum settings

That gave me a 3.6GHz fully stable running Prime95 for 8+ hours with no errors and temperatures stabilized at 46C
then I looked at CPU-Z & HWMonitor... Vcore was at 1.225V... I thought that was a bit high to my liking. I hadn't touched the VCore in the BIOS, so I'm guessing the MB auto-compensated somehow for more FSB and thought it necessary to overvolt it on it's own. I'm guessing the BIOS setting was over 1.3V considering that damned Vdroop.

I went back, forced a low 1.14 if I remember, and XP would bluescreen. I ramped it up all the way to 1.17 and XP booted but Prime95 would give an error in less than 2 minutes. Ramped it up some more with more Prime95 and found that 1.2V was the correct setting for a 8+ hour Prime95 stress test. This gave my a CPU-Z & HWMonitor VCore of 1.14. I don't understand the whole thing about Vdroop, I just know it's there, and it's for $$$ purposes using cheaper MB parts and that a lot of people try and kill the Vdroop with that pencil mod; I didn't go there at all.

1) Do my numbers seem alright up to now? 3.6GHz, 1.20V BIOS, 1.14V Windows, 43C after 8+ hours of Prime95

The next step, trying to get more, but I'm not even sure that will be possible because of my RAM, it's PC-6400, so "native" DDR2-800, obviously the 400MHz FSB is a joke for it, going higher is where I can get more points of failures and makes things a headache.

So, I went back in BIOS, tried 422MHz FSB, left it at 1:1 for RAM, loosened timing to 5-5-5-15 (seems like a timing a lot of people use for Oc'ing) and didn't touch the VCore. That didn't even post. Turned down the FSB to 415MHz, still no go. Went to 1.21VCore, still no go.

I'm guessing I'm trying to many different things at the same time, but I'm not sure which one I should take out of the picture. Should I relax timing to something much higher like 8-8-8-20 and focus on CPU, then turn down timings? Should I focus on FSB only going 1MHz increments at a time until it crashes, then raise VCore & even losen timing if that doesn't work?

I'm @ work right now, thought I would get confirmation/tips while I'm here and put those to good use when I'm back home. Not sure I wanna waste another full night of trying to OC a system another ~200MHz or so when it's rock solid at 600MHz running at lowest DDR timings.

Any suggestions greatly appreaciated, thank you!
 

Evilonigiri

Splendid
Jun 8, 2007
4,381
0
22,780
I'd focus on the ram first. Drop the multiplier to x6 and lower the vcore (not necessary but you'll see) as low as you see fit but stable (or set it at auto). Then concentrate on the ram and mobo. Taking 422MHz as an example, do 1:1 with auto timings and use different voltages. Increasing the FSB and NB voltages by .1V or so.

Tweaking those settings around, run Prime95 or Orthos blend test for at least 12hours, preferably 24hours. This is where the x6 multiplier and lower vcore comes in place. With such a low speed (422 x 6 = 2.532Ghz) and vcore, you can be sure that the temps won't go very high, so it's safe to leave the pc on for long periods of time. If you happen to sleep near the pc, you could even turn down the fans.

Finally, when you're sure the mobo and ram are stable, simply crank the multiplier back up to x9 and do what you did before. This is a method I just came up with lately, hopefully it works for you.
 

jay_l_a

Distinguished
Feb 15, 2008
576
0
18,990
It's worth considering:
Those who have gone much over 3.6GHz aren't seeing massive performance gains. You are increasing the potential for DEGRADATION on the CPU.

Your vCore seems pretty good.
I think mine is around 1.2875 in BIOS. I haven't tried dropping lower than that but I intend to.
 

RADIO_ACTIVE

Distinguished
Jan 17, 2008
897
0
18,990


I have a E8400 and I am running it at 3.6 right now your numbers (1.18v in Bios and I run around 39C idle and 42C load) seem good. You could push the proccessor beyond 4Ghz but I would suggest leaving it at 3.6Ghz being as you will not see much a difference between 4 and 3.6. Plus it would be added stress on your CPU.
 

beurling

Distinguished
Jan 2, 2008
316
0
18,780
E8400 at 1.4V Bios. Beyond 3.6Ghz, you need to push Voltages much higher than 1.2V. Your CPU might last 10 years instead of 20... but who wouldn't upgrade before that time anyway?....That's what I thought......

I sometimes encode videos for my PS3, and at 3.6ghz, a 1 hour TV show took 12min.. at 4Ghz, it took 5min. The difference IS noticeable in my case.
 

beurling

Distinguished
Jan 2, 2008
316
0
18,780
true, but I rarely encode, and games I play don't benefit from quads... next compy will be quad for sure... especially since everything will take advantage of it in a few years.
 

RADIO_ACTIVE

Distinguished
Jan 17, 2008
897
0
18,990
Well the rate Intel is pumping out proccessors just think what they will have in a couple of years lol, and if your a hardware junkie like me I sure I will buying what they are selling lol :bounce: