All right guys, a little OC'ing advice needed here. Built the PC in my sig a month or two ago, and while it OC's pretty well (~3GHZ from stock of 1.8GHz is nothing to sneeze at), it seems like I'm hitting some kind of artificial wall. I have read numerous P5B-family OCing threads at places like XtremeSystems and TechRepository forums, but have had no success. I'm hoping someone here sees something obvious that I'm missing, short of soldering the board to get NB voltage adjustment.
The voltage options in my P5B-E (rev. 1.01G) are rather limited. I can alter the CPU vCore, vDIMM (up to 2.1v max), and FSB Termination Voltage only. These boards are known for their vDroop, but a simple pencil mod has rectified that.
I can OC the CPU (running memory 1:1 with relatively loose timings of 5-5-5-15, which keeps the memory underclocked, so it should not be the problem) at stock voltage all the way up to ~330MHz FSB. Adding vCore beyond this point only gets me a slight OC improvement (340MHz FSB, not 24/7 stable), at the cost of huge temperature penalty.
Speaking of temps, this board is really hard to believe when it comes to CPU temp reporting. Flashing different BIOSes has altered the reported CPU temps, as has turning on/off "PECI" in the BIOS. Sometimes I can get temps that are below ambient, which I know is wrong. Using CoreTemp/SpeedFan/MBM gets me the individual core readings, which are more believable. I idle in the low-to-mid 20's C (at a room temp of ~21C). At stock voltages, load temps rise to around 50C. As I type this, I'm running at 340MHz FSB/1.55v w/Orthos small FFTs running in the background, and my cores are running at 64-65C (my motherboard CPU reading shows 75C). That's literally a 40C temp delta from idle, which I'm NOT happy about, and is about the absolute max I'm willing to try.
I have vDIMM set at 2.1v, FSB Termination voltage maxed at 1.45v (I don't even know for sure if this has any affect at the relatively low FSB speeds I'm using), and vCore has been tried anywhere from stock (1.325v) to a max of around 1.55v, which is about as far as I want to go on air.
The passive chipset heatsinks get VERY warm, even at stock speeds. The NB heatsink is will burn my fingers when OCing. Adding a 120mm fan (see below) has helped.
Thinking it may have been my old case not coping with hot new components very well, I modded it to add a 120mm fan on the side panel (blowing in over the video card and chipset heatsinks), as well as a 120mm blowhole fan exhausting thru the top. This has reduced the "motherboard" temp my P5B-E reports from 42C to around 30C. I also lapped both my Tuniq Tower and my E4300 heatspreader (they were both surprisingly concave), and while that's helped w/idle temps slightly, it hasn't really affected load temps or OC stability.
So, after all that, any ideas? Am I just missing the NB voltage adjustments? Or does this sound more like a CPU that is just maxed out? This is my first go-round with Intel OCing, after Athlon XP/64 OCing in the past, so am I just screwing something up? I guess I was hoping for something around 3.4GHz with this chip, with its 9x multiplier, or even as high as 3.6GHz, which would be 400FSB, running memory in sync. Maybe I'm just dreaming...
__________________E4300 @ 3GHz (lapped) || Tuniq Tower 120 (lapped) || Gigabyte EP45-UD3L || eVGA 8800GTS 640MB @ 648/1512/999 || 6GB DDR2-800 @ 5-5-5-15|| 74GB Raptor || 80GB Caviar || 250GB WD || Corsair HX520W || Samsung 225BW 22" LCD (1680x1050) || X-Fi XtremeGamer || Win7 RC x64 || Antec SX1040BII Server Tower || Altec-Lansing ATP3 Speakers
No comprehension to fail; I vacuum the wind for my sail
Can't be the rest; Let others waste my time?
Owning success is the bottom line