I recently upgraded my somewhat aging system with a new CPU and memory. I am not an avid overclocker by any means, but I build my systems and then tune a stable overclock. At that point, I generally don't change the settings for 2-3 years.
My old system was :
Corsair 550 watt PSU
E2200 pentium dual core
4x1GB DDR2 800 mhz
2 aging WD HDs
8800GT video card
Noctua cpu cooler
I changed the cpu to a Q9550 and upgraded to 8GB of 1100 mhz DDR2 ram.
I initially took a mild shotgun approach to the overclock. I initially ran into a wall at 3000 mhz. After adjustments, I had a stable 3600 mhz oc (DDR2 at 1030) with increases in the cpu voltage (1.35), ddr2 voltage 1.9v, fsb voltage +.1v. The system would get to the upper 60sC at load and idle in the low 40s.
At this point, I decreased the oc to 3500 mhz as I didn't want such a hot system. Next, I started reducing the voltages by .025 increments. I found that I could get my voltage down to 1.15v(stock=1.225). I also decreased my DDR2 to stock (1.8v) and my FSB to +.05v. These changes resulted in a much cooler system. My load temps are high 40s, low 50s. My idle temps are in the mid 30s.
Anyhow, searching the forums, I found information on undervolting with underclocking or stock clocking, but not much on undervolt overclocks. My end results seems much better in terms of heat and energy efficiency. Has anybody else had similar experiences?
I have some CPUID and coretemp screenshots, but couldn't figure out how to attach pictures on this forum.
ALL motherboards and ALL CPU's and ALL RAM react different with every overclock setup.
Overclocking is ultimately trying to get the highest clock with the least amount of voltage and thus heat.
I usually start overclocking by keeping the voltage @ stock and upping the FSB in small increments then briefly testing. (all while keeping a 1:1 ratio with the RAM). Once i become unstable with upping the FSB i usually back it down one small notch and then start upping the voltage AND FSB in small amounts while keeping an eye on temps and voltage.