Got this weird OC prob. Need some advice.

[Linux23]

Got a E6420 a few weeks back, and a DS3 Rev3.3 G.Skill 4-4-4-12 PC800 RAM, and the build went really well. I let the components break in for 48hrs. Then I upped the FSB a little (300MHz), and the system continually kept rebooting itself over and over until I cleared the CMOS. Found out there is a F11b beta BIOS for 6420 support.

Installed the BIOS fine. OC'd moderately to 333MHz FSB, everything is stable and runs fine. Now I bump it to 350 and it finally boots past BIOS, but as soon as Windows loads, it instantly reboot.

This is just meager 2.66GHz. What in gods name could be holding me back. I should at least be able to get a respectable 3Ghz. Any ideas, advice on how I can get a higher OC. My CPU temps under 333MHz was just under 35c. I know I should be able to push this CPU to at most 400x8.

Any help, tips, advice, please let me know. I'm at a loss here.

 

[aigomorla]

ahahahaha... break in for 48hours... Thats cute... i usually torcher the living crap out of it for the first 24 hours on stock to see if the chip is good or not.

But anyhow..

sounds to me you have a memory divider problem. Your memory is being pushed faster then it can process, hence your bluescreening and windows is auto restarting.

Try to lower the divider to a more friendlier lvl. Maybe down 1 option.

Or try to bump up the memory voltage a little bit. Your chip should do 3.0 at near stock voltage.



Also what PSU do you have? a crappy one will hinder your overclock.

 

[Linux23]

I have the memory at 2.1 volts (which it's rated at). I have the Tuniq Tower 120, which is a beast, but it keeps things cool. Memory divider is 2.0, so at 350 it's only running 700 out of its rated 800Mhz.

Should I bump the voltage on the memory up some more?

 

[GuitarDaddy]

You should skip the 350-399 range, as it's much easier on the 1333 strap(for internal timings) which kicks in at 400fsb. You can drop the CPU multi to 7x, 7x400 will give you 2.8ghz which you should be able to do without raising the vcore much, just to make sure your setup can handle 400fsb. Then start working on how much vcore it will take to do 8x400.

I seem to have one of the crappier E6400's and mine takes 1.375 to pass orthos at 7x400. about 1.45 to boot at 8x400 and about 1.475 to get completely orthos and TAT stable at 8x400. Part of my problem is my board (P5B-E) has a horrible vdroop/vdrop combination, ie... I have to set 1.475 in bios to get 1.38idle/1.35load according to Asus probe 2 in windows.

With my E6400, I really have to up the volts pretty quickly as I go above 3.2ghz, 3.5ghz is as high as I've been able to get it stable but the temps get kinda out of hand at that point even with excellent aftermarket cooler.

 

[Linux23]

Okay, i'm home now and will play around with it some more. I'll give your suggestions a try GuitarDaddy. Someone mentioned the north bridge getting overheated as the cause of me not being able to overclock any higher. Do you have Rev3.3 of the DS3 board, and have you heard about any issue with the MCH overheating?

 

[aigomorla]

Originally posted by: Linux23
Okay, i'm home now and will play around with it some more. I'll give your suggestions a try GuitarDaddy. Someone mentioned the north bridge getting overheated as the cause of me not being able to overclock any higher. Do you have Rev3.3 of the DS3 board, and have you heard about any issue with the MCH overheating?

this sounds like a good idea.

However, my other option was to increase your NB voltage. Since the ram control is located their, a slight voltage increase on your NB could solve problems as well.

And dont put your CPU at anything higher then 1.4v on air. The load temps on it will kill you. Unless you have a really loud high spinning fan, 1.4V on a air is not recomended.

Remember this guy said he has a very bad Vdroop. The DS3's arent as bad as our 680i im guessing.

 

[Linux23]

It works fine at 400x7. For some odd reason, the highest I can go is 365x8. Any other suggestions?

 

[Linux23]

Here are my stable 365x8 settings
PCI Express Frequency (102Mhz)
CIA 2 Disabled
System Memory Multiplier 2.00

System Voltage Control
DDR2 Overvoltage Control +0.3V
PCI-E OverVoltage Control+0.1V
FSB Overvoltage Control +0.1V
CPU Voltage Control 1.37500V
Memory manually set to 4-4-4-12 (specs on the box)
Refresh to ACT Delay is at 0 (Not sure if this matters)

As soon as I bump it to 270, it reboot 3 times and sets everything to stock.

I can't figure out what the problem is with it going higher. I even tried bumping the voltage on the CPU to 1.40 at 370x8 and no good.

 

[GuitarDaddy]

Great advice aigomorla

You may need to stick a fan on the NB if you have to up the volts much.
I have an aftermarket heatsink on the NB (Thermalright HR-05) with a 1600rpm 120mm fan blowing through it. But I only have to bump the NB volts one notch to get up to 450 fsb.

And I don't know much about the DS3, I have the Asus P5B-E P965 chipset board which is notorious for the vdroop. And I also have possibly the best aftermarket HSF the Thermalright Ultra 120 extreme. And I've got five rigs around the house and this one is my overclocking toy, and am a pretty hardcore overclocker and will push voltages and temps farther than most I know as I'm not afraid to kill a lowend CPU. Although I have been overvolting chips like crazy for years and have yet to kill one, and have come to the conclusion that CPU's are tougher to kill than most people would beleive.

If you plan on keeping your rig for several years and/or don't have the resources to easily replace it I would definately recommend keeping your volts as low as possible. I do beleive 1.4 is probably pretty safe with good cooling. But consider the source

 

[GuitarDaddy]

Originally posted by: Linux23
Here are my stable 365x8 settings
PCI Express Frequency (102Mhz)
CIA 2 Disabled
System Memory Multiplier 2.00

System Voltage Control
DDR2 Overvoltage Control +0.3V
PCI-E OverVoltage Control+0.1V
FSB Overvoltage Control +0.1V
CPU Voltage Control 1.37500V
Memory manually set to 4-4-4-12 (specs on the box)
Refresh to ACT Delay is at 0 (Not sure if this matters)

As soon as I bump it to 270, it reboot 3 times and sets everything to stock.

I can't figure out what the problem is with it going higher. I even tried bumping the voltage on the CPU to 1.40 at 370x8 and no good.

Overclocking between 370 and 399 is very tough because your on the much tighter 1066 strap for internal timings. Try going straight to 400x8 or 401x8(some boards don't switch straps until 401), but that may require more vcore than your comfortable with depending on your chip. The earlier conroes seemed to go much higher on lower volts

 

[Linux23]

Tried 401x8 and 402x8 at both 1.4 and 1.45 voltage and no good.

 

[GuitarDaddy]

Bummer

I wish I knew more about the DS3, it may have the same vdroop issue as my P5B-E.
Is there a large difference in vcore from what you set in bios compared to your uguru readings? And what vcore works for 7x400? You can also try setting the vcore on auto, but I will warn you that it may give it lots of vcore. My board on auto vcore and 8x400 gives more volts in windows than if I set it 1.55 in bios

*edit*
I will add that in my opinion(again consider the source) that voltage degrades CPU's over time and that short term usage of high voltage for testing really doesn't pose a risk of hurting the CPU unless you get crazy and give it 1.75v or something and most boards won't go that high. I would never run a CPU 24/7 overvolted by more than 15%(hardcore, most say 10%) but I have frequently overvolted many CPU's 20% or more for short periods of time during testing.

 

[Linux23]

Yes. I set it at 1.4 in BIOS and it shows 1.39 without load and under load it goes to 1.360v and fluctuates.

 

[GuitarDaddy]

Looks like yours may be as crappy an overclocker as mine

Here is what it takes for mine to be orthos stable at 3200

3200


As you see my vcore under load shows a 1.39, at idle it show 1.42-1.44
And I have it set to 1.475 in bios

 

[Linux23]

Originally posted by: GuitarDaddy
Looks like yours may be as crappy an overclocker as mine

Here is what it takes for mine to be orthos stable at 3200

3200


As you see my vcore under load shows a 1.39, at idle it show 1.42-1.44
And I have it set to 1.475 in bios

I wish I could even post at 3.2GHz. Cooling on my system is great, so temps wouldn't be a problem. I have it at 2911MHz now, testing with Orthos. Seems to keep rebooting after about 10 minutes or so.

 

[Compgeek1979]

Heya guys, Linux32 how did it go with your setup? Reason i'm asking is i have the exact same hardware.

I went through alot of stuff over the last past weekend to OC my setup. I can't get over 380FSB stable. i tried all kinds of voltages for Cpu/ram/MB only things i didnt try are severe voltages, like over 1.45 for cpu or 2.3v for ram... Tried f11b-f10-f9 Bios, i tried different cpu multipliers, and 2 things i found in common with my experiments are

1. I seem to have a big VDroop when orthos screams instability. (running orthos and CPU-z at same time)
2. lowering the CPU multipliers and upping the FSB no matter what anything past 380 is unstable even at 400x6 @ 2.4ghz

I have same setup hardware wise - Ds3 mobo/ 6420 C2D / G-skill DDR800 4-4-4-12
Arctic Freezer pro 7 for cooling with a small fan on NB heatsink.


I originally thought it was RAM limiting me, but i'm not so sure.. People give great reviews for OCability on both the ram and Mobo, although the 6420 are relatively new cpu's on the market. Also heard alot of forum talk about 44g (week) was altogether a bad batch of CPU's(which is what my cpu is from batch wise).. I guess where i'm at now is i can settle for 3.04ghz@ 380x8 - 2.5 mem ratio = 950 memory speed or go for more voltages to ram/cpu - which bothers me since i am on air cooling and don't want to kill my hardware Latest thing i notice is i used to no be able to post @400FSB but now i can make it to windows with orthos crapping out within 1 minute and on some occasions rebooting before desktop.

I guess my question would be with my cpu voltage currently @ 1.45 in bios and 1.408 in CPU-Z and DDr2 voltage @ +.3 =2.1v Should i try voltages @ 1.5+ on cpu and 2.3-2.4V on ram? I've read alot about overvoltage i don't want to kill my stuff but if it's not as risky and some say or is bad news according to others.

 

[GuitarDaddy]

Have you tried the 1:1 memory multi and 400x8? That would run your memory at stock.

 

[Compgeek1979]

Yeah, when i try 400x8 i get a constant reboot, or 2 reboots before it defualts to base clock speed.

i did some small testing tonight with 401x6 and 401x7 @ both 2 and 2.5 actually ran somewhat stable, as i only test each setting about 20-30 minutes with orthos. I even dropped my cpu voltage to 1.375 and it seemed stable, although again only on short testing.

Saw someone suggesting 401fsb for 1333 strap. vrs 400 or below to 366.
I also tried other settings such as 420fsb 1:1 and it doesn't post at all. Or even ramping up my ram voltage to 2.3, ran cpu voltage all the way up to 1.5v these seem to have no effect on 400x8 at all. it does the same regardless of volts. it makes me thing something esle is going on or its just crappy cpu maybe

maybe i should try 2.4v on ram... i am thinking i may just settle for 3.04ghz soon as i am
limited on parts and budget. I had hopes of reaching 3.2-3.6 but it looks like a pipe dream


I am wondering what other people with 6420 week 44g are getting on average and if they had to jump to 1.4+ volts for stability when getting around 3ghz.

 

[qliveur]

Are your PCIe and PCI buses manually locked? If not, that could be the cause of the instability. If they are, just ignore me.

 

[Compgeek1979]

Yesir, i started locking PCIe bus at 100mhz due to the fact i had a corrupted wd raptor drive, which i might have done with overclocking ...