Case: CoolerMaster Stacker (original version)
PSU: Tagan 580W EasyCon
Graphics card: XFX 7800GTX (to be replaced with EVGA 8800 GTS KO 640MB)
HDDs: 2 x WD Raptor 74GB [RAID 0 - WinXP & Programs]; 1 x WD Raptor 150GB [Game Installations]; 1 x WD 300GB [Data]
Sound card: SB X-Fi Music
DVD RW: Plextor PX-716A
Speakers: Logitech Z5450
Mouse: Logitech G5
Keyboard: Logitech G15
Monitor: Samsung SyncMaster 244T
Installation of the new components went without problems - probably a first for me
I followed Wusy's Guide when overclocking. I chose to leave the 'No-Execute Memory Protect' function active in BIOS as it is a function I like to have enabled. I also chose to set the memory speed manually at a 1:1 ratio with the FSB, with 'aggressive' timings of 4-4-4-12, knowing I could always relax these if I experienced any problems. All other recommendations in the guide were followed to the letter.
Initial voltages were set as follows:
vDIMM: 2.2v (just on the warranty limit for my memory)
I first attempted to use ClockGen to increase the FSB in Windows (to save some time). However, any attempts to use this application resulted in severe video corruption, so I had to do it the hard way (reboot and change FSB in BIOS).
During the OC process, I did experience some problems with SpeedFan 4.32 with the vCore at 1.4V (SMBus read errors resulting in a number of temp and fan readings not showing). However. this was the only problem I experienced throughout the whole process.
To reduce the time required to OC, I only stress tested for 2-3 minutes with Prime95 at each FSB increase.
During the complete process, my idle temps (using SpeedFan 4.32) did not increase at all (24C). The load temps tended to increase by around 1C for every 0.1GHz increase in CPU speed (initial load temp at 3.0GHz was 43C). Core temp at final OC of 3.4GHz was 47C.
At around 3.45 GHz, I had my first failure in Prime95. I then dropped back to 3.4GHz, and began more thorough stress testing. The system was stable for around 30mins, so I then started to reduce the vCore.
Passed 2-3 min test at 1.3875v
Passed 2-3 min test at 1.3750v
Passed 2-3 min test at 1.3625v
Failed 2-3 min test at 1.3500v
At 1.3625v, I had another failure in Prime95 after around 5mins
At 1.3750v, I managed to complete the 2 hour stress test.
A reduction in the vDIMM voltage to 2.1v resulted in failure, so I left it at 2.2v.
I have no intention of trying to increase this OC. I was happy to be able to drop my vCore below the original 1.4v due to the problems I experienced with SpeedFan reading the SMBus. Also, this system flies compared to my previous system (Opteron 170 OCd to 2.65GHz), so I am very happy with the OC.
Well, a long post, but maybe there's some useful information for those new to overclocking who are unsure what to expect during their first attempt.
Edit: question regarding final temps moved to separate post
When you hit the wall at 3.45, did you try further increasing Vmch? The following link is an excellent coolong mod (shown in my sig) for the Asus P5W DH Deluxe, which addresses a problem with some boards topping out around 380Mhz FSB.
With part one of this mod, my board is stable at 414Mhz maximum. Also, I took the mod a step further by adding a 40mm fan to the northbridge heat sink to further reduce the temperatures. The fan is a Cooler Master Blue Ice SLC-S41-U1, and is on the shelf at Comp USA for $10.00. Just discard the heat sink that's boxed with the fan, and use the supplied screws. It's a perfect fit, and the three wire plug will allow the motherboard to monitor the RPM.
The quickest and easiest temporary solution (until you can remove the board) is to remove the decotative but rediculous heat disipation inhibitive shiny covers, and install the 40mm fan. This really drops the northbridge temps big time, and allows for higher and more stable FSB frequencies when Vmch is increased. My P5W DH Deluxe requires 1.5 @ 378, 1.55 @ 389, 1.6 @ 400, and 1.65 @ 411.