Okay, so I'm trying to clock my Q6600 to 3.6 with stability. I have managed to get to 3.4GHz with no trouble. However, at 390FSB, the system is stable with day-to-day usage, but Handbrake BSODs it with an 0x00000124 error, Prime95 fails within a few seconds on either cores 1 or 2 (0 and 4 haven't failed yet). After reading This Bit-Tech Article, Bindi says the GTL reference voltage should be 0.67xthe CPU Vcore. With my Vcore at 1.38V, 0.67 of that is .924 Volts. Now, On this board, the only relevant voltage that is that low is the CPU Reference voltage, currently set to .840V (default is 0.760V). My query is, the voltage range only goes up to 1.010V, so is .920V going to damage the chip or board? I've also had to bump the MCH voltage up to 1.240V, and set the reference to .840, same as the CPU (default is also 0.760V) Anyone? More info available on request.
Increasing any voltage will "damage" the CPU to some degree. The biggest killers are PLL and VTT though. Try 0.88 or 0.90 to see if it makes a difference on the CPU - edge it up slowly.
Hmmm, CPU Reference @ .875 and .905 make no difference. Seems it's consistently core 3 failing on P95 now though. CPU Termination and CPU PLL are both still at defults - 1.2 and 1.5 respectively. I've also got Load Line Calibration disabled. After the Anandtech article explaining exactly how much the Vcore can go out of spec with this enabled, I'm not too keen on enabling. MCH Settings: MCH Core: 1.280V (default 1.100V) MCH Reference: 0.800V (default 0.760V) MCH/DRAM Reference: 0.900V (default 0.900V)
I see from your sig that youve got an Artic Cooling Freezer 7 Pro, its a good hsf and all, but isnt it a bit weak for more than 3.4ghz? what are your temps under prime 95/orthos/super pi mod or whatever? more than 60 degrees?
I have the exact same problem, 3.4 is NP but 3.6 is no can do with my SLACR cpu. 50c full load after one hour.. but might be my chipset
Temps for 3.4 are 59C on the hottest core on a Prime95 load. I don't have the Cojones to see what Intel Burn Test would do to it. I am looking to get a new cooler - possibly the Akasa Nero, as 59's still on the warm side of 'alright' for me. I clocked it back to 3.4 while I'm at work. I'll have to pop to AUTdirect on Friday to see about a Nero. Mgith have another play then if I have time
no, go to 4.2ghz If it is only one of the cores failing (in this case #3) a test then I suspect there could have been damage to the processor. However I am not sure on that.
I doubt damage. It just sounds like a limited chip, or incorrect voltage settings to push it any further. I doubt it's running into a strap, the other 3 cores are stable. Have you checked your droop under load? If it's increasing as you gradually increase the voltages you might find the droop increases, just dropping enough to not be enough for the weaker core. Temps sound pretty acceptable tbh, I wouldn't worry too much about them just yet. All mobo-based temps need to be taken with a pinch of salt tbh. Course, you could well have just found the limits of your chip.
Try 1.3-1.35V VTT and 1.7V PLL It sounds like you're cooling limited or just simply CPU limited by Core 3, which isn't uncommon in a dual-die CPU.
If one core is much hotter than the others, it could be indicative of an uneven IHS, time to lap that CPU.
I'll have a look tomorrow hopefully I'll give that a go at some point, cheers. You may be right though - with a lower CPU Reference, it's consistently core 2 that fails. It may just be a less tolerant die, as 2 and 3 are the same temperature, so I would have thought are the same die. They're also the ones that drop first at this speed. No, the top two are 59C and 58C at load, the other two are 56 or 57, so there's not much in it. I don't fancy lapping it TBH, I'd rather just get a more capable cooler. I must admit though, for £12 the Freezer 7 Pro isn't doing bad!
dont clock higher, lower the multi, and up the fsb. might need to tweak voltages slightly. but really, need more info on ur rig and what exactly is failing when you try to reach higher clocks. @ badders: temps seem ok, dont think thats what is crashin you at all. have you checked the sticky on this forum for oc'in quads? a good read. as long as vtt and pll voltages are in line (that sticky will link you to a good article about that), i dont see you having a problem with the gtl settings. try hitting the settings bindi recommended, and if it works, test stable and then work your way down to instability, then back up to stability.
