Sign in with
Sign up | Sign in
Your question

e6400 Overclocking Problem with DS3

Last response: in Overclocking
Share
December 4, 2006 12:34:32 AM

Hi, I have a current e6400 build:
- Gigabyte S3 mobo
- overclocked to 3.2Ghz
- Ram Voltage: 2.2v
- Orthos stable for 3 hours (and counting)
- TAT temp: 58 degrees C (Orthos Full Load)
- vCore: 1.328V
- MCH Voltage: normal (ie/ 1.45v)
- FSB Voltage: normal (ie/1.2v)
- pci-e voltage: locked at 100v

Although it seems like a solid overclock (please tell me if the temps are too high however), I am worried that I cannot set the memory multiplier to 2.00 for a 1:1 FSB:RAM operation. Currently, my Crucial 10th Anniversary ddr2-667 (rated 3-3-3-12 at 2.2v), is at 4-4-4-10 at 2.2v running at a ddr2-500mhz on a 4:5 ratio. When I try to set the memory multiplier from 2.5 to 2.00, my system fails to boot up. In fact, the system keeps rebooting until it doesn't boot up at all. When I restart my computer, the system has automatically changed everything to e6400's default 2.13Ghz settings.

I've read more than enough to drive a normal person insane, and it's driving me a little crazy. I know I should feel satisified but to see that my system will not run at a 1:1 ratio is driving me nuts.

Please help. Thanks.
December 4, 2006 2:42:12 AM

anyone?
December 7, 2006 12:18:25 PM

The problem might be your solid 400Mhz FSB overclock. Set it to ~407 and try again.
This is just a preliminary test so dont worry if it doesnt work, I've got more tricks under my sleeve.
Related resources
December 7, 2006 12:53:05 PM

And going to 5-5-5-12 doesnt help?
December 7, 2006 1:22:30 PM

Your temps are good at that OC, the thermal design is 61.4c:

http://processorfinder.intel.com/details.aspx?sSpec=SL9...

However, if you raise the voltage or go higher with it, you may want a better cooler.

Quote:
And going to 5-5-5-12 doesnt help?


No doubt you either need to relax the timings to run DDR2-667 at DDR2-800 speeds. Depending on the SPD info on your RAM, it may not be relaxing it enough - it may not have timing info for that speed.

I have the DQ6 (basically the same BIOS) and Corsair DDR2-800 C4 - it automatically relaxes the timings to 5-8-8-18 at the 450 FSB I'm running at. Seems happy, yet to experiment with voltages. I do know that when I tried tigthening the timeing, I has the EXACT same problems you describe.

Download and run CPUz to see what your timings are at the workable speed, then relax them in BIOS, test again, then try the 1:1 and see if it works. Good luck
December 7, 2006 1:33:04 PM

Someone correct me if I'm wrong, but the ratio set to 2.5 would actually have the memory running faster than the FSB, right?

At 2, which is a 1:1 ratio x 2 (Double Rate) you get FSB=400 X 2 = 800, but at 2.5 you would get FSB= 400 X 2.5 - 1000!

Seem strange that the it gets unstable when actually LOWERING the memory speed.
December 7, 2006 1:42:45 PM

Hmmm - a little more thought and this seems to make some sense. If the MB is reading the settings from the RAM, it's getting the 2.5 based on no overclocking...

FSB = 266, RAM at 667 needs the 2.5 setting.

Seems like there is a setting in BIOS to disable reading from the this from the memory - hit Ctrl + F1 to unlock the advanced options then go into the performance section and let us know what the settings are.
December 7, 2006 1:56:53 PM

One more question, which BIOS version are you using?
December 7, 2006 5:19:09 PM

Quote:
And going to 5-5-5-12 doesnt help?


5-5-5-12 will not work. I though those timings were as loose as you can get them. Guess not.
December 7, 2006 5:21:47 PM

Quote:
Hmmm - a little more thought and this seems to make some sense. If the MB is reading the settings from the RAM, it's getting the 2.5 based on no overclocking...

FSB = 266, RAM at 667 needs the 2.5 setting.

Seems like there is a setting in BIOS to disable reading from the this from the memory - hit Ctrl + F1 to unlock the advanced options then go into the performance section and let us know what the settings are.


On a 1:1 ratio, I can get as high as 380FSB with upped voltages and 5-5-5-12 timings.
I will try loosening the timings to the 5-8-8-18 someone was successful with.
December 7, 2006 5:22:24 PM

Quote:
One more question, which BIOS version are you using?


I am using the latest BIOS version. F6 for the S3, which is basically F5 for DS3.
December 7, 2006 5:59:00 PM

Is your northbridge heatsink too hot to touch? ... that passive cooler needs good airflow.
December 7, 2006 6:31:06 PM

Quote:
Is your northbridge heatsink too hot to touch? ... that passive cooler needs good airflow.


Currently, according to the EASYTUNE5, my system temperature (which I think is the temperature of the northbridge) is sitting at 27degrees. I have been overclocking with a very loud (but very effective) fan blasting air at everything in my case. It decreased my temperatures by around 15degrees C. Obviously, I'm only using this fan to eliminate the heat from my northbridge.

However, even though I have eliminated the heat from my chipset, I still cannot go beyond 380FSB on a 1:1 operation.

I have raised the voltages of MCH, vcore, FSB, and memory. Nothing seems to work.

I have even loosened the timings as far as 5-8-8-18.

System doesn't boot up past 380.
December 7, 2006 7:03:08 PM

I have been able to hit 400x8....on easytune5. It has been stable for more than an hour on Orthos, while my very large fan is hitting the inside of my case and lowering the temperatures by 10 degrees. However, I have tried to do the same thing on BIOS using the same settings I used with Easytune5 along with the large fan hitting the chipset, the cpu, and the ram and yet my system will continue to reboot 3 or 4 times with nothing on my monitor. System will start up and stay on as if it was in hibernation or something because the monitor would not come out of its power saving mode.
December 7, 2006 7:37:46 PM

Do you have QST enable in the BIOS?


FROM GIGABYTE S3 FAQ...

Q. When Intel QST is enabled, will the result of overclocking be affected?


A. Yes, it is an Intel QST limitation.
December 7, 2006 8:12:18 PM

I am pretty sure I disabled everything including the Thermal throttling control.

What I find most strange is I can run at 400x8 operation...However, I can only do that when my memory multiplier is at 2.5 for a 4FSB:5RAM divider.

Shouldn't I be able to run the memory at 1:1 if I'm able to get the same FSB at 4:5?

its puzzling.

Is there something more straining when the FSB is run at 1:1 than when it's being run at 4:5? Am i putting more strain on the CPU if I was running it at 1:1 than if I was to be running it at4:5?
December 7, 2006 8:47:02 PM

That measurement was taken at the top of the IHS, not the internal diode/core. That measurement translates to 85c in TAT and CoreTemp.

@theOP
If raising the FSB a few Mhz doesnt work, give it some voltage help to MCH and set your timings at 5-5-5-15.
December 7, 2006 9:05:19 PM

Quote:
That measurement was taken at the top of the IHS, not the internal diode/core. That measurement translates to 85c in TAT and CoreTemp.

@theOP
If raising the FSB a few Mhz doesnt work, give it some voltage help to MCH and set your timings at 5-5-5-15.


I've already tried loosening the timings. I am now suspecting that my cpu is a dud. My memory can work at 500mhz at 4-4-4-12 timings in a 4:5 operation. Therefore, at 1:1 operation, theoretically (heating issues aside) my cpu is capable of a 500x8 overclock.

Upping the voltages for the MCH does not help at 1:1.

Right now, I am currently running 400x8 = 3200mhz OC @ 4FSB:5RAM operation. This crucial memory is amazing in that it can do 500mhz at 4-4-4-12 timings. now its a matter of why i cannot run this at 1:1.
December 7, 2006 11:08:44 PM

Quote:
Therefore, at 1:1 operation, theoretically (heating issues aside) my cpu is capable of a 500x8 overclock.

Hm, not really.

Quote:
The problem might be your solid 400Mhz FSB overclock. Set it to ~407 and try again.
Did you try that?
December 8, 2006 12:35:28 AM

Yes, I've tried setting the FSB at 407x8.
December 8, 2006 3:02:27 AM

Lets try several things:
Vcore 1.4v
FSB ~409
5-5-5-15 and 2.2v
and a few notches to MCH, FSB termination, and ICH (SB).
It's just easier than going one by one... lol
December 8, 2006 3:21:50 AM

I don't think on the S3 board, I can change the voltage for the ICH. I can only change the MCH.
December 8, 2006 3:42:08 AM

Right, I just checked and you cant. Just try the rest of the setting altogether.
December 8, 2006 3:45:11 AM

OVERCLOCK SUCCESSFUL!!!!

I just changed my FSB from 383mhz FSB (1:1) to 425mhz FSB and it booted up on the first time!!!!!! Absolutely amazing. Now I have an overclocked e6400 @ 3.4ghz with stable temps and 5-5-5-15 timings.

And now for the questions......

Does anyone know why I could not boot up at 384-407 but when I changed it to 425, it booted without problems? Would really like to know so that when someone else on the forum comes across the same problem I can give back to them what everyone on this forum has been trying to help me with....

Also, does anyone have a guide in lowering these temps or do I have to just lower them one timing at a time? (ie/ 5-5-5-15 --> 4-5-5-15 --> 4-4-5-15, etc)?
December 8, 2006 4:03:45 AM

I believe the problem was the strap change on the DS3, weird because its so wide on this boards. FSB and MCH could have something to do too, its hard to tell without thoughtfully testing it tough.

Quote:
Also, does anyone have a guide in lowering these temps or do I have to just lower them one timing at a time? (ie/ 5-5-5-15 --> 4-5-5-15 --> 4-4-5-15, etc)?

Go 4-4-4-12 @ 2.2v, the memory has proven not to be the problem.
December 8, 2006 4:50:54 AM

Quote:
I believe the problem was the strap change on the DS3, weird because its so wide on this boards. FSB and MCH could have something to do too, its hard to tell without thoughtfully testing it tough.

What exactly is the strap change?

Mind explaining what the memory strap is all about? Tried googling it without success.
December 8, 2006 5:03:38 AM

All the way up to ~400Mhz, motherboards use the 1066 and 1333Mhz strap, but somewhere around there (widely varies for each motherboard) it switches to a theoretical 1600Mhz strap (not known to exist up to date), usually you would only experience a loss in memory performance (due to the slacked internal chipset latencies because of the strap change) detectable only by benchmarking, but the stock vMCH seemed to really have affected your memory in this case.
Im not very good at explaining things but I hope you got a rough idea of how memory straps change. :D 
!