So I recently fixed a fan that had stopped, and found that my BIOS was suddenly strange and had to be reset. I decided to update it (from v2.1 to v2.7) and re-optimize it from scratch. To make a long story short, I looked up a bunch of stuff that I previously didn't think had an effect on overclocking, found the new feature in my BIOS to change the CPU multiplier to a lower setting (yay more FSB potential!), and discovered the nuances that differed between my extremely similar pairs of 1GB RAM sticks, which may or may not have something to do with this unusual brick wall I've hit, in terms of OCing.
Before I was able to change the multiplier for my CPU, I've been able to overclock to just over 3GHz stably, my bus speed usually being set to 1337
. The CPU multiplier here was 9. I'd have my RAM timings set to auto which CPU-Z reported at 5-5-5-12, if I remember correctly. However, the RAM is rated to run at 4-4-4-12 at 400MHz, esp if it is below that. I'd never been able to meet the FSB potential to make use of the 400Mhz, so the RAM, which I set to be "linked" in the BIOS with the CPU bus, scaled with the overclocking, but not at 1:1.
After I updated and dug a little deeper into BIOS options, disabling all sorts of spectrum spreads, looking at the nuances of my RAM, and being able to set my multiplier to 8, I've hit a really nice little island of stability, but I can't seem to push it anywhere.
But first, my specs:
Intel Core 2 Duo E6600 2.4GHz "Conroe"
MSI P6N SLi (vanilla) BIOS v2.7
(4x1GB) Patriot PC2-6400 (400MHz)
-2 of these sticks have a part number 6400LL Series, the other 2 are PDC22G6400LLK
-I will post the slight differences with a comparative CPU-z screenshot. Differences may or may not have an impact on this issue...
CPU Fan is a Zalman 9500cu and I'm using Arctic Silver compound.
Running Vista Ultimate x64
Those should be all the pecs that matter here but let me know if I'm missing something. Now let me explain the odd RAM thing.
Here is one of the stick types:
And the other:
Unless I'm reading it wrong, they both have the same potential specs at 400MHz, depending on the voltage. At 2.1v, the slot 1 stick has all the same timings that the slot 3 stick as at 2.2v at 400MHz. Thus, I set my voltage in the BIOS to 2.1 which is currently working at these timings that I set manually in the BIOS:
I also unlinked the CPU bus and memory speeds, set each individually in order to achieve a 1:1, set the CPU multiplier to 8, bumping the CPU bus to 1500 and setting the RAM to 750 (that's how the BIOS portrays it), and bumped the voltage to the CPU to +0.0125 as you can see here:The problem is that as soon as I increase the CPU bus and the RAM speed any higher in such a way that keeps the 1:1 ratio consistent, my computer freezes at some point while booting up. The more I increase it, even while keeping the ratio consistent, the sooner it freezes, even during POST. When I reduce the overclock, I get farther in the boot up process but it still freezes with time, seemingly proportional to the amount of overclock.
For example, when I set the CPU bus to 1528 and the RAM speed to 764 (1528/4=382 ; 382x2=764), I froze early in POST. When I set the CPU bus to 1508 and the RAM to 754, I froze at the point where windows loaded up. When I set it to 1504 and 752, I froze at the Vista logo. Yet, when I set it at 1500 and 750, it's very stable, with no errors reported by Prime95 after almost 11 hours of blend testing.
What is the deal with this? Can anyone see something I'm missing? I was able to overclock the clock speed of the CPU a bit higher in the past, up to 3015 with the x9 multiplier. This is the brick wall I was referring to, seemingly perfectly stable here, and freezing with just a touch beyond now. What am I missing?
Thanks in advance.
(Edit: Changed topic title for clarification.)