Hello OCF,
I'm afraid I may have destroyed my Q6600 (G0). I know the risks associated with adjusting clocks, and am prepared to deal with the results. I try to learn on my own as much as I can; though I did read quite a bit beforehand -- it apparently was not enough.
I have had my setup a few weeks. I ran it stock for a while, until I was prepared to attempt an overclock. I reached about 2940mhz with the processor, with the stock 9x multiplier at a FSB of about 325mhz.
After observing load temperatures (one instance of Prime95 "blend", as well as one "small" for each of the four cores), I decided once I reached 71C (Intel's limit to the G0 stepping) that I was unsatisfied though it was stable so far.
"I did not use the thermal paste correctly," was my conclusion. Which is true; I applied AS5 on the heatsink as well as the CPU. It was also spread unevenly.
Since I was going to reapply the TIM, I decided I may as well try my hands at lapping the surface of the die heatspreader. The Rosewill heatsink's surface was already smooth as could be. I probably did not do a perfect job (which lapping requires, however is besides the point), since after rebooting, the core's temperatures were about the same.
Then, I figured since I just reapplied the AS5 as TIM, I may as well burn it in. I only knew of the concept at the conception of the idea, but learned recently it requires lowering the voltage. (Some may argue this is feckless; however, I wanted to see for myself. I also learned it may not actually have anything to do with the TIM, although it does change characteristics under alternating heat and cold conditions -- Correct me if I'm wrong about that, for the AS5 paste.)
For those unfamiliar with what I am talking about, I'll skim the surface of the burn process: Lower the core voltage after a fresh installation, in small increments, using stress tests to confirm stability. Once instability occurs, raise the voltage to the previous stable level. Run the stress test(s) for an extended period of time, and repeat until desired.
If you can see where I'm getting, this (supposedly) allows you to end up with lower voltages for stock FSB speeds, which translates to a higher overclock potential, or generally less power usage. It seemed like good logic to me; and I was ready to give it a shot.
Only to be shot down. With the only software (that I know of) to lower voltages unavailable -- Clockgen -- I resorted to my own flawed logic.
If I lower the multiplier, I could use less voltage. If I raise the FSB to match stock speeds, maybe it could essentially be the same as lowering the voltage - while achieving stock clock speeds - without the option to do so in the BIOS (or through software).
Well, I can see where that went wrong. I should have lowered the multiplier, checked if I was correct in thinking that the CPU used less volts at a lower clock speed, and then proceeded to up the FSB to a level teeter-tottering on stable (again by stress-testing), then checked the voltage at that point to see if I could essentially pull off under-poweing the CPU without a software/BIOS option.
The motherboard I'm using has a feature in which it detects failed overclock attempts on the fourth failure to boot. However, I am receiving a grim outlook..
I installed as well, an optional bracket in an expansion slot, which uses a system of four red or green LEDs to communicate 16 possible processes on the motherboard.
The system won't POST. The message I get from the LED signal is, quoted from my mainboard manual: "System power on. The D-LED will hang here if the processor is damaged or not installed properly."
Gulp.
I assume the rash sudden increase in FSB (1066 stock at 9x, as opposed to 1600 at x6) toasted something. If I were to get a new processor, I'd like it to be one using 45nm technology, in which case I'd have to flash the BIOS, which I couldn't do without a processor to begin with (I have no friends with the same motherboard to swap ROM chips [if that's even possible on my motherboard, I haven't checked], and I don't even trust myself to do that).
Is it possible to save my mainboard, or CPU?
I should also mention this is my first build. Whoops..
I'm afraid I may have destroyed my Q6600 (G0). I know the risks associated with adjusting clocks, and am prepared to deal with the results. I try to learn on my own as much as I can; though I did read quite a bit beforehand -- it apparently was not enough.
I have had my setup a few weeks. I ran it stock for a while, until I was prepared to attempt an overclock. I reached about 2940mhz with the processor, with the stock 9x multiplier at a FSB of about 325mhz.
After observing load temperatures (one instance of Prime95 "blend", as well as one "small" for each of the four cores), I decided once I reached 71C (Intel's limit to the G0 stepping) that I was unsatisfied though it was stable so far.
"I did not use the thermal paste correctly," was my conclusion. Which is true; I applied AS5 on the heatsink as well as the CPU. It was also spread unevenly.
Since I was going to reapply the TIM, I decided I may as well try my hands at lapping the surface of the die heatspreader. The Rosewill heatsink's surface was already smooth as could be. I probably did not do a perfect job (which lapping requires, however is besides the point), since after rebooting, the core's temperatures were about the same.
Then, I figured since I just reapplied the AS5 as TIM, I may as well burn it in. I only knew of the concept at the conception of the idea, but learned recently it requires lowering the voltage. (Some may argue this is feckless; however, I wanted to see for myself. I also learned it may not actually have anything to do with the TIM, although it does change characteristics under alternating heat and cold conditions -- Correct me if I'm wrong about that, for the AS5 paste.)
For those unfamiliar with what I am talking about, I'll skim the surface of the burn process: Lower the core voltage after a fresh installation, in small increments, using stress tests to confirm stability. Once instability occurs, raise the voltage to the previous stable level. Run the stress test(s) for an extended period of time, and repeat until desired.
If you can see where I'm getting, this (supposedly) allows you to end up with lower voltages for stock FSB speeds, which translates to a higher overclock potential, or generally less power usage. It seemed like good logic to me; and I was ready to give it a shot.
Only to be shot down. With the only software (that I know of) to lower voltages unavailable -- Clockgen -- I resorted to my own flawed logic.
If I lower the multiplier, I could use less voltage. If I raise the FSB to match stock speeds, maybe it could essentially be the same as lowering the voltage - while achieving stock clock speeds - without the option to do so in the BIOS (or through software).
Well, I can see where that went wrong. I should have lowered the multiplier, checked if I was correct in thinking that the CPU used less volts at a lower clock speed, and then proceeded to up the FSB to a level teeter-tottering on stable (again by stress-testing), then checked the voltage at that point to see if I could essentially pull off under-poweing the CPU without a software/BIOS option.
The motherboard I'm using has a feature in which it detects failed overclock attempts on the fourth failure to boot. However, I am receiving a grim outlook..
I installed as well, an optional bracket in an expansion slot, which uses a system of four red or green LEDs to communicate 16 possible processes on the motherboard.
The system won't POST. The message I get from the LED signal is, quoted from my mainboard manual: "System power on. The D-LED will hang here if the processor is damaged or not installed properly."
Gulp.
I assume the rash sudden increase in FSB (1066 stock at 9x, as opposed to 1600 at x6) toasted something. If I were to get a new processor, I'd like it to be one using 45nm technology, in which case I'd have to flash the BIOS, which I couldn't do without a processor to begin with (I have no friends with the same motherboard to swap ROM chips [if that's even possible on my motherboard, I haven't checked], and I don't even trust myself to do that).
Is it possible to save my mainboard, or CPU?
I should also mention this is my first build. Whoops..