It is recommended that you keep cstates enabled as they run the sata controller not to sure on this tho, I wouldnt use HW to record your voltages pos in my opinion use cpuz and coretemp to work out what you need while running prime95 blend. If you are happy with the oc you have atm just blend prime for 24 hrs then maybe 100 runs with intelburn test.
1.6v PLL (1.5 seemed to cause errors, when i raised it to 1.6 those errors seemed to not occur anymore)
1.00/.875 VTT/IMC respectively (1.05/.925 is stock, i just drop 3 .05 notches on both together since i think they are related somehow, when i dropped 3 more on both to .985/.860 i had issues so brought it back to 1/.875).
Reducing PLL, VTT, and IMC voltage reduced my temps, maybe altogether like 2-6*C? I'm not sure exactly... but I don't think it was significant at all. It was just me being a perfectionist more than anything, i could have ran 10*c hotter just fine.
There are limited reports that reduced PLL voltage has increased stability for people.... but for all i know those guys had something else going on, you know.
you should definitely do custom blend instead of that one fft length. There are many, many other ffts that are very hard on ivy bridge. you really need to run prime95 at ~23+ hours to be stable (custom blend, make sure no whea errors).
Dont use ultra high llc. Unless you got a digilta multimeter to tell you best, use the 2nd highest LLC setting. What going to happen with ultra high, is your temps will likely be higher, and while the voltage offset you need might be lower, your true voltage will be way higher. Most likely, your software won't tell you that your voltage is higher (your software is not even close, as it's an average, so when your board is going up in voltage it's common it might just say your voltage is going down). Just use the 2nd highest LLC setting and use the lowest voltage setting you can from there.
If you got a multimeter though, by all means, see the lowets voltage you can be stable on each llc setting, compare, use the best llc setting. Though 99% of the time, it's the 2nd highest setting...
Belial's Comprehensive Guide to LEDs!
Belial's Comprehensive Guide to 7950s!
Belial's Heatsink & TIM Bench + Review: Hyper 212, H50, NH-D14, PK1/2/3
Belial's Comprehensive Guide to Fixing Bent & Broken CPU Pins
Belial's Comprehensive Guide to Quieting, Lubing, and Fixing All types of Fans, including Twin Frozr II
Woah, i dont have the courage to modify any voltages other than the vcore and pll. i dont even know if lowering the pll voltage will affect my 2600k even at the least. because i really want to use this rig for well at least 3 years, coz it was only given, cant afford it im not stressing right now, cpuz hw realtemp and aida64 were running, and some other programs. im only using hw because i think its the only one who can record the minimum vcore out of those 4. because i am not always looking at cpuz aida64 and realtemp as a im working on our thesis
Using a lower CPU PLL Voltage has the possibility of enabling you to lower your core voltage without really affecting the stability that much.
Then leave Internal PLL Overvoltage enabled.
Are those programs doing anything though?
I don't know if what you're seeing is normal, but so far it seems like nothing bad is happening. I would say that you are right about the C-states. This is probably C3 or C6.
yeah might be those c-states that makes a sudden drop in voltage once in a while, one time it dropped like 0.255. LOL
You must be getting more accurate readings than most people. The C3 state is a very low-power state and C6 is where there's no voltage being used at all. So, C6 is basically where the CPU is sleeping (not to be confused with Sleep Mode).
HW recorded another minimum voltage of 0.352v. I don't know if I should believe that. Sometimes AI Suite II is warning me of a voltage drop < 1v on the 12V. Is AI Suite II that reliable in terms of voltage readings? Because I think, there is no problem with my psu here. Or am I day-dreaming?
Not with that PSU. You have a very good unit there.
Anyway, software in general is not trustworthy to read PSU voltages.