Failing to boot with 4GB RAM

[AD-15]

hello,

I've just installed another 2GB of RAM, making that 4GB using 4 DIMMs. However, using my overclocked settings, I get a blue screen on startup. Using my non-overclocked settings, with my memory set to 1066 makes it to the windows loading screen, but hangs there. Using my non-OC settings, with my memory at 800, boots successfully into windows.

As it works eventually, I'm thinking it may be a voltage setting somewhere. Any ideas guys?

[Webby]

You may need to give a touch more voltage to the north bridge (may be called MCH over voltage control)

[Thorsson]

What's your CPU clocked to? Is your RAM running at 1:1?

[AD-15]

hello,

I'm using a multiplier of 9, with an FSB of 267Mhz.

[PeteSmith]

Have you checked the manufactuers recommended voltage for the memory - may be you need to change that. What timings are you using? I had a similar problem just recently - i solved it by relaxing the timings ever so slightly.

[AD-15]

hello,

I've tried upping the MCH voltage, but it did no good. I had increased it by about 0.2V by the time I gave up. I also tried using a 2T command rate rather than 1T. I'm running it now at 800Mhz, but I simply can't get it to boot at 1066MHz.

[AD-15]

hello,

I've just tried using my OC settings, but with a divider of 1:1.2. This resulted in a speed of 852MHz. This booted successfully.

Having tried a million different settings, I think I can say I have narrowed it down to being a problem with the divider. I'll check to see how high I can go.

EDIT: 1:125 also works.

[Thorsson]

Well if 1.2 works then 1.125 is gonna, or do you mean 1.25? But something isn't right here. If the memory is showing 852 @ 1.2 that means it would run 710 at 1:1 which is an FSB of 355 not the 267 you say you're running.

[AD-15]

hello,

Sorry, yes, I meant 1:1.25.

These are from my OC settings, with which I am running an FSB of 355.

[Thorsson]

So at 1.25 they are running at 887, and that's OK? But if you put them on any higher divider (well multiplier really) they fail?

Is the memory running on the settings specified for Voltage and timings?

Presuming so then you could try relaxing the timings slightly (I don't think 2T does much on DDR2) like 6-6-6-18, upping the voltage a notch and upping the MCH voltage a notch.

I can get my 4x1Gb Ballistix PC6400 to around 900 on the same board at stock timings and voltages, so I'm a little surprised that you're having so much problem.

[AD-15]

hello,

Thanks for your help so far guys.

Yes, 1:1.5, 1:1.6 and 1:2 do not work.

I've tried your suggestions Thorsson, but no luck. Infact, increasing MCH voltage immediately introduces instability.

[Thorsson]

It didn't allow me to get my RAM any higher either.

So does the CPU not like a higher FSB than 355? You could check whether it was the CPU or the mobo by lowering the multiplier to 6x, putting the RAM on 1:1 (or even lower) and seeing what you could get the FSB to. If it's the mobo limiting things you could possibly RMA it, as it's supposed to support 1066.

[AD-15]

hello,

Thanks for your help.

Fortunetly, I just managed to fix everything by simply swapping the old modules and the new modules around.

I think I had the old modules in DIMMS 1 and 3, now they are in 2 and 4.

Upon closer inspection, I can see that although spec wise they are identical, there seem to be a couple of differences between the two pairs. The first being that the heatsink on the newer pair is lighter than the older pair. Also, "v4.1" is printed on the new modules, and "v1.2" is printed on the old modules.

Well, I'm happy now, and I can enjoy ridiculously short loading times on everything with my new RAM.

[shaithis]

Be careful, you may have just moved the problem to the end of your ram space and therefore once you start heavily loading stuff into the address space that those modules service, you may see the instability return.

[AD-15]

hello,

How can I test the modules for this? Thanks!

[Webby]

Run Memtest a few times that will do the job

Edit: Link Here