Need Help Overclocking P35-DS3L

D

dhongyt

n00b
Joined
May 20, 2007
Messages
24
I'm having trouble getting a decent overclock on my e6600 using the gigabyte p35-ds3l motherboard.

I have followed this guide (http://www.hardforum.com/showthread.php?t=1169366). I know this is for the p965 but the bios configuration is very similar to mine.

I also tried following this guide (http://www.legionhardware.com/document.php?id=677&p=2).

My problem is that I can not even reach a stable 3.0 GHz.

I set the FSB:DRAM ratio to 1:1 using the System Memory Multiplier (SPD) to 2.00 while using a CPU Host Frequency of 333 MHz (FSB), which effectively makes my memory frequency 667 MHz (I use PC2-6400 RAM).

I tried setting the CPU Voltage Control as high as 1.475V as well as setting FSB OverVoltage Control to +0.1V.

When I try to run prime95 it crashes after 10 minutes.

Am I missing something here? Why can't I get a decent overclock?
 
At the settings that crashes prime after 10 min, boot into a memtest86+ CD and let it run at lease a couple of full passes. If it does not fail the issue is with the cpu and its settings, if it does fail it is a memory issue and probally raising the MCH voltage and putting a fan on the MCH/northbridge will help. But first determine what is holding you back, cpu or MCH (Memory Controller Hub) as I agree you are underclocking your memory and the dram itself is not the issue.
 
I'm running mem test 86+ now, so far it seems like it running fine. Do you think I need to increase the voltage on my memory higher than 1.9v?
 
What is the rated voltage for the memory? You shouldn't need a higher fsb voltage to run at 333.

Also make sure eist is disabled
 
Seems like I got it to 2.70 GHz stable, where the Vcore runs at 1.45 and my FSB OverVoltage Control at +0.1V and MCH OverVoltage Control at +0.2V. But when I try to get it to 3.0 GHz with no changes it can't handle it. What should I try increasing?

What is the default voltage of the MCH and the FSB?
 
It could be your ram holding you back. You might have to manually set the timings to the rated timings to run at 800fsb. Also what is your rated memory voltage? Stock on the board is 1.8 so you may have to up that.

I would run the ram with a divider so that it is running slower than its rated speed to see if that is causing the problem.
 
5 5 5 12 is a bit of an odd timing for CL5 Try 5 5 5 15 or 5 5 5 18.

It would help if you gave us the details or a link to your exact memory. Default voltage on the MCH is 1.25V and Intels max voltage per the data sheet is 1.85V if memory serves.

Try 7 x 400 and see if it will boot, we are attempting to jump over a "dead" space where system timings are apparently causing an issue. It it will boot, see if it will do a full pass of memtest. (this helps verify its is fairly stable and reduces the chance of windows freezing and becoming corrupted due to a fail attempt to load). If it will do a full pass of memtest, see if you can get into windows. Sometimes leaving all voltages on auto and letting the board decide during several startup/shutdown cycles as it attempts to boot and automaticlly set values works and then gives a baseline to work from. It also sometimes sets alarmingly high voltages so it everything on auto voltage works, take note of the voltages in the bios and check your temps in windows. Then you can see what you can turn down and still get it running. This is just experimenting to help find the root cause of your issue as that board and cpu should do 3.2 easy. It is possible you got a dud chip, it happens but well worth the time to try a few things before throwing in the towel.
 
I'm trying 7 x 400 like you say and using auto voltage and keeping the ram at 5-5-5-12 @1.9v since that is what its rated at.

I'm running speed fan right now and looking at the voltages.

Vcore1 @ 1.39v
Vcore2 @ 2.02v <-- This looks kind of scary, what is that?
+3.3V @ 3.34v
+5V @ 4.78v
+12V @ 5.50v
-12V @ -16.97v
-5V @ -8.53v
+5V @ 5.03v
Vbat @ 3.17v
 
I wonder if those are really the voltages you are getting. Seem a little bit out of whack to me. You might want to check your BIOS to see if it agrees with the voltages posted in Speedfan. If it does, you might be running into powersupply issues.
 
Looking at the bios

Vcore @ 1.428V
DDR18V @ 1.968V
+3.3V @ 3.360V
+12V @ 11.921V

Do these seem normal or high?

Last night I ran prime95 for about 2 hours and it seems stable at 7x400 with voltages on auto. Shall we try to reach 3.0 GHz now? How should I proceed?

The voltages also seem to flux, it sometimes drops a little bit and come back up.
 
quick reply since I see you are on, going to look at the rest of posts and think a bit and will edit this one.


The Vcore2 voltage is your memory voltage, somewhere in speed fan I think you can change the descripton. A lot of this 3rd party monitoring software (not supplied by the board manuf) makes a good guess at what is what for volts and temps but sometimes get it wrong, no biggie. This particual "mis-label" is a known deal and has been verified as true and accurate. So actually you are a tad high on memory voltage, not enough to be concerned about at all, and actually helps me tend to eliminate the memory as the problem, but it has me wondering what is going on with your board. It should do what you want. LOL I hate these kind of posts, no easy one thing to say "thats it !" it sometimes takes a lot of methodical testing to find out what is keeping you from getting your OC.

do us a favor tell us your speedfan temps while running orthos at your best/highest speed stable settings. Only need to run it for 5 min or so

the coretemp values are what we are looking for. should be 4 of them.


You voltages all look good, forget the - (negative values) some supplies no longer even supply one of the standby voltages, I forget which one but speedfan never seems to get them correct on gigabyte boards. Other are well within spec and should not be an issue. Vore is a bit high for my tastes but I understand, you are trying to get it to OC better, however sometimes, rarely, the cpu does not react to high Vcore the way most cpus do and it can actually hurt the OC. I would reduce it a step at a time until your current OC becomes unstable and then bump it back up a notch or two trying to find the voltage that will suppport your existing OC + a bit more voltage and then start testing again in raising the FSB in small increments.
 
Thanks for helping me out though Bill! I'm kind of new at this and having a difficult time, I think I might have burned my last board.
 
Temp1: 46C
Temp2: 48C
Core0: 56C
Core1: 56C
Core: 56C

These are the temp when running prime95.

But when I stop running prime95

Temp1: 46C
Temp2: 29C
Core0: 35C
Core1: 35C
Core: 56C
 
What does stable mean to you? How many hours should I run prime95 until I believe the system is stable?
 
Okay so now I'm starting to manually set my voltages, I kept the voltages of the FSB and the MCH at normal voltages while the Vcore is running at 1.375v

It seems to me like I may have Vdroop because Speed Fan is saying my Vcore1 is running at 1.31v

Maybe that is my problem?
 
Vdroop is a Intel design specification, it is noraml to have .06V of droop, nothing to worry about at all, its not excessive and can be compensated for by raising Vcore another notch. You will/should see even more of a drop under heavy load.

I still think it is the MCH. FSB voltage should be last resort.
 
So I should just raise the MCH voltages while leaving the FSB voltages alone?

7x400 @ 1.375v was not stable for me, should I increase the Vcore more or try for the MCH?
 
I'm beginning to think that maybe my CPU just isn't good for overclocking :(

I'm wondering if I should try it with C1E and EIST enabled, but it only makes it harder to overclock. My friend seem to have is working fine with those options enable.

Any other suggestions?
 
You must log in or register to reply here.
Back
Top