TR Forums

Hello everyone. I'm trying to OC my E8400 on a Gigabyte ep43-ud3L (as shown in signature) just to see what I can get out of it. I've been running it at 3.6GHz and stock voltage for some time (idles down to 2.4GHz) and it is perfectly stable. I started fiddling around and noticed that I could only get the FSB up to 419MHz. The problem is that it can do 10 runs of Intel Burn Test at 3.771GHz and STOCK VOLTAGE (1.25V) no problem. I have a 120mm Xigmatek tower cooler and core temps are under 56C under load at that setting. However, it won't even boot at 420MHz (=3.78GHz) even with 1.35V on the Vcore. I've seen people do 4.4GHz on E8400's using 1.38V so it seems odd that I'm hitting a hard limit here.

The lowest ram multiplier I can set on my board is 2x which equates to 840MHz on my 800MHz DDR2 dimms. I've had the memory over 900MHz before, running Prime 95 blend overnight just fine, so I feel like the RAM isn't the problem. So, I'm thinking that the mobo is not okay with running the FSB any higher than 419MHz at this point. Any ideas to help me get over the wall?

I have a different mobo (Asus x38), but I'm also running an e8400 at 3.6 Ghz with the 400 MHZ FSB. You may have hit the upper limit that you can reach... but... do you have the ability to tweak the northbridge voltage? The FSB interacts with system peripherals more than the CPU other than being used as one of the numbers used to generate the clock signal. Additional voltage to non-CPU components might help, but I wouldn't expect miracles either.

I'm running a different mobo (Biostar P43D3) and CPU (C2D e6320), but have the same issue: I can't push the FSB above 419-420mhz. It doesn't matter what voltage I use. Maybe it has something to do with the chipset?

I'm pretty experienced with LGA775 and Gigabyte boards in particular, having held on to a socket T system for so long.
NB voltage is part of the equation for hitting 400+ fsb sometimes. There are also FSB holes present, so while 420fsb might not work, a setting 1-3mhz in either direction might.

Have you tried dropping the multi to find your max FSB? What are your vdroop settings?
IBT is a harsh mistress, dude.

@thegst: I see no options for NB multiplier. The MCH Core voltage seems to be for the Northbridge, but I don't know what a reasonable voltage bump would be. I'm also not sure I want to push that too much for a "just for fun" OC. I honestly don't see anything for vdroop either. Is it named something different?

As stated in my OP, I've dropped the RAM multiplier as far as it will go to eliminate RAM timings from the equation. (Looking at it again, I've run my RAM at 960MHz overnight on Prime95 before). When I first started the OC, I plugged in 460MHz FSB = 4.14GHz which didn't work. Admitting that it was a bit agressive for stock voltage, I successively dropped the FSB to 440, 430, and then to 415 which worked. Then I tried 420Mhz and nothing. So I bumped the Vcore up to 1.35V and tried 420Mhz again....nothing. Dropping down to 419MHz was stable even when backing down to 1.25V on the Vcore again (I didnt run Prime95 though). I expected to see instabilities on Intel Burn test before the thing wouldn't boot.

I've also tried setting all voltages to Auto and locking the PCI frequency to 100MHz which didn't help either.

PS. I love the owners manual: MCH Core Voltage = "the default is Auto" ....thank you, very helpful.

The problem is your Gigabyte ep43-ud3L, it has a FSB wall around 420Mhz. You will not get it to run anything past that, it has to do witht he P43 chipset. If you had a P45 you might get a better overclock but with the P43 you'll have to be happy with what you have. If you google around long enough you should be able to find some info on this.

I did the same research a few years back. I have the exact same Gigabyte MB as you but with a Q9400 in it. I would have liked to run the CPU faster than 3.2Ghz but it's better than nothing.

tfp wrote:

The problem is your Gigabyte ep43-ud3L, it has a FSB wall around 420Mhz. You will not get it to run anything past that, it has to do witht he P43 chipset. If you had a P45 you might get a better overclock but with the P43 you'll have to be happy with what you have. If you google around long enough you should be able to find some info on this.

I did the same research a few years back. I have the exact same Gigabyte MB as you but with a Q9400 in it. I would have liked to run the CPU faster than 3.2Ghz but it's better than nothing.

This. Coz normally bumping NB voltage should do the trick, also for RAM overclocking..

c'mon, the p43 'wall' isn't set at 420mhz or some nonsense like that.

it can be anywhere from 417-440 or slightly more, and there are ways to push it depending on BIOS options - that's why I asked if max FSB was ever determined.
You very well might be hitting the wall on that board, but it is not a specific thing on all P43 boards.

Page 38: http://download.gigabyte.us/FileList/Ma ... -ud3l(us3l)_e.pdf

"CPU Clock Ratio"

Do you not have this setting? If so, is it adjustable downward? That's what I meant by finding out your max FSB.
If you reduce this and run a higher FSB with a lower multi, you can find the high limits of your FSB and see if it's the chip holding you back or the board or some other factor.

There are other things I've seen work on P43 boards like setting the PCI-E frequency to 104mhz. The '104' tweak. Make sure to have all the weird power save states disabled (C1E)
All, if you're going to present enthusiast advice as canon, at least be complete. I realize it's still unlikely to push much past 420mhz, but we should be giving our best effort instead of just saying 'CAN'T BE DONE NO ONE HAS EVER DONE IT LA LA'

Edit: Also, don't push it past 1.4 or so, but try a CPU Termination of 1.3v. More commonly known as VTT, but labeled funny here. Looks like no actual NB volt tweaks, so this might be the best you can do on this board. Good luck.

I've seen a few other ep43 users getting the same limits. I also have a hard time believing that the P43 chipset is hardwired to a 420MHz limit as well, but I have no knowledge or evidence to prove or disprove that claim since I'm one user with one mobo.
I hadn't thought of lowering the cpu multiplier since I figured the CPU should be able to get quite a bit higher than 3.6GHz. I will try that along with the "104 trick" tomorrow and let everyone know how it goes. After reading some about VTT (voltage threshold termination) it sounds like that setting is preventing me from feeding the CPU any more voltage than what the CPU Termination Voltage (VTT) is set at? This is turning out to be quite interesting

Yeah I remember trying out the PCI-e trick of setting the bus to 104 or 105. It was good for a few more mhz on the FSB but not a huge amount.

I don't believe the P43 is really hard locked the issue is the PCI-e and FSB seem to be based on the same clock or at least related. Once you push around 415-420 you have to move up the PCI-e to get any higher FSB. The issue is PCI-e can't be run to much higher than 100-105 or you're graphics card will be unstable.

Either way good luck. I have the EP43-UD3L as well if you can figure out something to get higher clocks I'll happily try them on my Q9400 as well.

I have an EP35-DS4 and an EP45-UD3L here, both run 400 perfectly, 410 with a little tinker, and won't go past 420 in any way, shape or form. Sorry that I don't have anything to add beyond the +1.

Well, I had to do a combination of the "104 Trick" and setting the "CPU Termination" to 1.3V, but I was able to break the 420MHz wall. In fact, I got 435MHz (=3.915GHz) with a Vcore = 1.36875V. 440MHz wouldn't boot and I didn't really feel like nailing down the exact limit since I wasn't going to be breaking 4GHz anyway. Stress test temps maxed at 60C. I think I was still limited by the motherboard, but I didn't try lowering the CPU multi to find the max FSB. Dissapointing, but what can ya do. It was a fun little experiment, but I would never run at those speeds on a daily basis. I'll be sticking to stock voltages and 3.6GHz until I sell for my IVB upgrade. A special thanks to "thegst" for all the help.