Ran into a interesting issue with BSOD with Windows 7 64-bit on a overclock that was similar to another build. I picked up a Athlon 64 x2 4450B 2.3Ghz for $5 to upgrade my one Sempron 140 2.7Ghz single-core to a better dual-core processor system. All was good initially. And I stress tested the CPU at 2.3Ghz and then performed the 200 to 220Mhz RAM overclock which forces the locked multiplier Athlon 64 x2 4450B to jump to 2.53Ghz just like I did on another build with the same CPU and similar motherboard. Ran stress testing and all was good as well as ran Memtest86 on it and it passed 3 full tests and figured all was good to go. The RAM in this system is also good quality higher end Corsair DDR2 800Mhz RAM a matched pair of 2GB sticks for 4GB.
Symptoms I had was for starters my Windows desktop CPU gadget would report 2.3Ghz yet the system was really running at 2.53Ghz as confirmed with benchmark. All other systems that use this gadget when overclocked the CPU gadget reports the correct overclock, so this was the first indication that something was odd with this system in its 10% overclocked state. The Windows 7 gadgets would destabilize and freeze and eventually an error message would pop up stating that the gadgets failed. Windows would then kill off the gadgets and then restart the service for them. When using this system for its intended purpose which is to convert recorded video gameplay files to size them down for youtube sharing of gameplay among friends in multiplayer games we play to look at them as instant replays on who was where and how it all played out etc, the system got to about 23% complete of the conversion process of the raw video file that was recorded with FRAPS which is being sized down greatly to a smaller AVI file and it BSOD on me and rebooted Windows. I thought maybe this was a fluke and so I tried this again and once again BSOD almost at the same point in the conversion. So I am then thinking is there something up with the original raw video file that the system is choking on in its conversion process. The RAM and Hard Drive are healthy as tested both.
I dug into event log and it pointed to a DLL. I searched the BSOD error message with the DLL reference and found an interesting article on RAM timings the cause of this exact error and sure enough the other people with the problem with same DLL also had Corsair RAM and also DDR2 800Mhz just like what I had. In the info shared on others helping the person with the problem it was discussed that RAM timings were off.
So I ran Memtest86 once again for 6 full tests which took about 7 hours at 2.53Ghz with its 10% overclock and no errors detected. Memtest86 said that all was well, but all wasnt.
I set the system back to 200Mhz vs 220Mhz FSB to remove the 10% overclock and set the RAM and CPU back to their normal speed and ran memtest86 and still passed. I then tried the video processing at 2.3Ghz and DDR2 RAM at 800Mhz and now no more crashing of the gadgets and no more crashing during the video conversion.
Well I guess I will just run this video game, game footage video converter system at its intended 2.3Ghz and it will just take slightly longer to convert videos to small AVI's to share with friends. On my other system with the same CPU it ran 2.53Ghz with no problems. I even go it up to 2.7Ghz, but at 2.7Ghz it was roasting, so i had to back down to 2.53Ghz to keep the operating temperature within spec. I tried initially for 3.1Ghz and closest i could get was 2.9Ghz in which at 2.9Ghz the video card was freeze framing and surging and theCPU was climbing to 80C fast on stock air cooled heatsink.
At 2.53Ghz I was running around 60C back at 2.3Ghz its running in the 50C range when running full tilt and 37C idle.
Figured I'd share this because what i found is that MEMTEST86 isn't always correct about health of RAM. It completely didnt detect the issue I had and this allowed for me to crash Windows 7 hard during a full tilt CPU processing conversion. It could be CPU caused vs RAM caused however especially since the memory controller is built into the CPU. And maybe the other Athlon 64 x2 4450B is a stronger CPU that can take the 10% overclock without a problem and this other Athlon 64 x2 4450B is weaker ( more sloppy ) and prone to problems when running it harder than AMD intended for it, since they did lock the multiplier for a reason.
