I don't touch the switch on the PSU unless I've done something badly wrong and need to pull the CMOS battery to make completely sure the BIOS is cleared. So power switch on the case. I don't know for sure. All I know is when I set LinX to 'all' RAM, I get behaviour that I don't get at all if I just leave Windows with a little bit of breathing space in the memory. Righto. I'd leave most stuff at auto, then, maybe setting QPI to 1.26-1.28v, Vcore to... 1.25v? See what that gives you. I tried EasyTune out of vague interest in what it was going to do, but as soon as it reported CPU temps as 16*C, I knew I wasn't going to trust it. ... I have another question... did you mount the TRUE120 as it was in the box? 'Cause when I checked the base of mine, it was so convex it was ridiculous.
I did, though it seemed flat and shiny. I didn't want to ruin its seemingly mirror finish. I do have to admit though I was extremely stupid when I mounted it, as I forgot to take the fan off initially and I had trouble with the spring loaded screws so I took it off a few times, ruining the thermal application. I don't know whether I'm willing to take everything apart and reapply it all over again considering the trouble I had the first time around. I still have to consider the break-in period, and the fact that my temperatures aren't that high... I hope. Max 62°C (Core #0) during full LinX load at stock settings.
Yeah, those temps are about what I'm getting at stock speeds. Either you got lucky with a decently flat one, or I'm running fans a lot less powerful than you. I wasn't seeing 81*C @ 3.2GHz, though... even with 1.35v on it. 81*C @3.8GHz with 1.35v, yeah...
Well, I'm back at the settings I mentioned in my initial post, except I'm at 1.2125 Vcore. I managed 100 tests in LinX without errors (4096MB), and multiple error free passes in Memtest and Windows Memory Diagnostics Tool (extended tests). I'm about to stress with Prime95 for 18 hours, which I hope I can manage. I think as long as I avoid any hard resets I won't receive any more supposed "boot failures".
be careful with all this stressing as you seem to be putting your machine though hell and it is possible to damage your CPU and system with prolonged stressing. prime95 for 18hrs seems silly as are you really gunna use any app for that long that will need 100% CPU constantly. i think not. even if your machine was on 24hrs a day running games, encoding folding etc you would never have 100% CPU usage.
One of the other reasons why I like LinX... it's a bit more representative of the fluctuations the system will experience during normal use - but really, unless you're gonna be folding 24/7, I'd not do that much stress testing either. To each their own, though. And if that's your thing, folding is gonna be a better stress test than Prime anyway.
No way... That is BS. Computers are made for constant full usage. If not, then there's something SERIOUSLY wrong with the system.
While most systems can take full load without issue - and should - if one of the power regs is a little weaker or a capacitor starting to go... constant full load of that nature will make it go pop quick than 'normal' usage. However, I'd rather know if something is gonna blow on the board within 30 days, rather than think everything is fine and dandy before getting a nasty shock when it does blow. I agree that full load shouldn't really damage anything. Providing that all components are good. If they aren't... well in that case it's just a matter of time before something goes pop anyway.
hmm well computers are intended for constant usage but...... under factory conditions. so it is not BS at all. The system should and will work at 100% for prolonged periods IF its not overclocked. overclocking your system puts extra strain on the vast majority of your components. it should be ok but heavy constant stressing for very long periods is not good for a machine that is over volted and running far higher than manufacturers specifications.
I'd agree if it wasn't for the fact that Jewels OC is within Intel's rated voltages for Vcore, QPI etc. So it's not over manufacturers specifications. Intel's recommended max QPI is 1.35v IIRC, same with Vcore. 1.65v on the RAM... although that's been revealed that there is a safety margin provided that QPI and DRAM are within 0.5v... Swings and roundabouts, really.
true i guess it is within the safety margins of specifications but point still stands if you go higher than a mild overclock which does require you to go over the spec on Vcore and maybe PLL or QPI depending on ya board. guess its just opinion really cant say id want to stress my machine for that long. ive stressed my current rig i7920 C0 stepping @ 3.95Ghz for 2hrs with prime95 on small FFTs and blend. used customPC benchmarks on loop for 2hrs. rigs perfectly stable for all my encoding, gaming use etc
I don't plan to perform any more stress tests, and my machine seems stable at this mild overclock. My voltages are at the lower end of all specifications, except DRAM which is at 1.64v. I stress tested for 15 hours with Prime95 using small FFTs, with Core #0 maxing out at 66°C. LinX pushed it to 68°C. I've had a few crashes due to some anti-virus related instabilities, though I've managed to avoided the dreaded boot failures by powering off instead of doing hard resets. Overall I'm glad I've finally managed to built my first computer and overclock it, and it's been an enlightening experience. My previous PC was from PCSpecialist, and at one point the CPU (stock E6300) was idling at 70°C!
Stock coolers are always poor. Which antivirus is that? Just so I know to avoid it like the plague...
