How to increase the speed of the Intel Q9550 to 3.6 GHz and above?
How to increase the speed of the Intel Q9550 to 3.6 GHz and above?
The interest in low-multiplier Core 2 stems from its limitations with FSB speeds. The presence of multiple dual-core chips on a single bus restricts the achievable multipliers, making it necessary to opt for higher options like X5460 or X5470, which support frequencies such as 3.8GHz or 4GHz at the maximum stock FSB of 400.
The dual-core processors can operate at extremely high FSB rates, but only by using DDR2-800 memory. This choice also affects cooling options, as the Q9550 chip consumes around 125W at 3.6GHz, and a 424MHz FSB might exceed safe operating limits for both RAM and cooling solutions.
Voltage constraints are another factor; the maximum safe voltage for 45nm Core 2 is 1.45v, while 65nm models require 1.5625v. With sufficient power, many users can achieve over 500MHz FSB and 4GHz speeds on the 65nm E6750, though this demands more heat than the current quad provides. The available cooling capacity remains a limiting issue.
Manufacturers cautioned that Intel Core 2 chipsets for this platform do not support multipliers below 2x for RAM to FSB compatibility. Using DDR2-1066 would have enabled higher speeds, up to 533FSB, but would have required more power than the current setup allows.
I wouldn't aim high for much of an original character on that board.
It wasn't meant for it—just a 4-pin CPU power connection, no VRM cooling, etc.
If you're curious about how far the Q9550 can go for experimentation, try locating an older P45 board.
The Gigabyte EP45-UD3P (and UD3R) stood out for handling high FSB speeds.
I really pushed it overclocking. I significantly lowered the voltage. At 1.325V and 3.65 GHz, I’m planning to try 3.8. My cooler is the LC-Power LC-CC-120 with a TDP of 180W, which gives me plenty of room. I also got new case fans and even tweaked my case a bit. I might add more cutouts for additional fans later. The maximum CPU temperature should stay around 66°C. I used this formula:
Core1 + Core2 + Core3 + Core4/4
Core1 refers to the temperatures, just an average.
Just a bit more FPS. The main goal was to test how far I could push this nearly 17-year-old chip. This PC is really old and I don’t need it much anymore—it’s already built from cheap parts. I also beat the Q9550, temperatures weren’t too high, though I could have done better. My next target is the Xeon E5450. Microcodes are ready, the cooler is good, I’ll get cheap mini heatsinks, try 4 GHz in air, maybe it sounds crazy, but I’m betting on it. I know most gamblers quit before winning, but I’m sticking to this and taking a chance with old hardware.
I believe a daily voltage of 1.5V for the Q9550 to achieve 4.1-4.2 GHz is acceptable as long as it stays below 75°C. When voltages rise, safe temperatures drop because heat accelerates electromigration, potentially reducing chip lifespan and requiring higher voltage for stability.
For 65nm Core chips used over time with adequate cooling, 1.55V is the upper limit. If your chip is a newer 45nm model, you should reduce this by about 0.1V, aiming for a maximum of 1.4-1.45V. Smaller nodes need less voltage since they use lower default levels and have thinner insulation.
Older 130nm chips can run safely at 1.6-1.7V, while newer FinFETs may struggle above 1.3V; however, FX SOI 32nm remains safe between 1.6-1.65V.
TL;DR – You’ll likely need to overvolt the CPU and ensure strong cooling, possibly settling around 1.5V for 4.1-4.2GHz.
I believe my chip isn't up to the task. 3.8 GHz and above are out of reach. I've tested everything possible, checking all combinations. I plan to try a Xeon E5450, keeping the side panel off for better results. The lapping won't really make much difference here. Appreciate the clarification too.