Overclocking help/cpuz incorrect voltage

buphmin

Distinguished
Feb 6, 2011
13
0
18,510
for reference my rig:
asus sabertooth x58
i7 950 w/ megahalems
gskill ddr3 1600 c9
asus 460 gtx
corsair TX950w psu
haf 932


ive been overclocking this week to find exactly the settings i want. i want my cpu to be around 3.5-3.7 ghz and my ram to be around 1700 mhz with the lowest voltage and i have been doing trial and error to find the best stable settings. today a rather strange thing came up. cpuz does not display correct voltages, but only sometimes it doesnt display the right volts. This happened after i lowered my cpu multiplier to 21x and base clock to about 170 at 1.20v. Cpuz (which has agreed with bios for a long time) said it was at 1.24 volts. i then changed it to 1.23 volts in bios and checked again and cpuz said 1.27 volts. though the temperatures were low for the voltage cpuz stated. i then set everything back to stock settings, cpuz said volts was higher than normal .98v idle 1.26 load. i then went to the lowest voltage ive found previously at stock clock 1.0625 and cpuz agreed with this and with every value until bios voltage was raised past 1.15v in which case cpuz voltage disagrees with bios settings. Now ive tried different versions of cpuz and redownloading them but they all do the same thing. Basically id like to know how concerned i should be about this and how to fix it if possible. the highest overclock ive done was 3.86 ghz at 1.23 volts with max temps in prime about 64c (29c or so ambient) which from what i can tell is generally considered safe as far as overclocking goes.
thanks in advance for reading my monologue.
 

buphmin

Distinguished
Feb 6, 2011
13
0
18,510
well i have some new information based off some tests i did. I went back to stock clock and lowered my voltage to a previously tested minimum of 1.0625v and ran prime to check stability and temperatures. cpuz showed higher voltage then set in bios, but had the same delta T from ambient as when i tested previously. I then tested with enabled Load line calibration and disabled. when disabled it showed a lower voltage of 1.024 but the temps were unaltered. enabled showed 1.096v but still the same temperature in prime. stability unaffected in each case. i then tested voltages lower than 1.0625 which proved unstable like my experiments before this incident. i then looked at the in bios hardware monitor and compared to cpuz with LLC enabled and they were the same. One interesting point i also noticed is the voltages from the psu rails. They were far off from specs. +3.3v rail read 2.85v, +5v read 4.30v and +12v read 14.28v. now i know atx specs are supposed to be plus or minus 5% voltage difference. but thats WAY beyond 5%. by the way things look my power supply isnt doing what it is supposed to anymore. and as a side note ive starting hearing this high pitched noise from my psu. ive read around with people saying things ranging from very bad to harmless annoyance.
 

buphmin

Distinguished
Feb 6, 2011
13
0
18,510
i do have turbo on. My min voltage of 1.0625v in bios was tested with turbo on. I used this setting for awhile when i was playing a lot of games to keep temps low and it would drop a little below 1.0625 at idle but would raise to 1.06 +- .002v during prime. last night at the same setting of 1.0625v with turbo bios hardware monitor read 1.72v and cpuz agreed with 1.72v. this morning with the same settings cpuz and bios monitor says 1.056v, some odd inconsistencies. i think either my voltage sensors are damaged, or my power supply is having problems. considering the noise and the voltage readings from the power supply in, im leaning towards power supply.
 

Cygnus x-1

Distinguished
Jul 28, 2011
901
0
19,060
It could just be a matter of different amount of load on the cpu. Small variants in vcore is normal depending on what the computer is doing. When heavily loaded the turbo will kick in and add voltage to match the added speed increase.
 

buphmin

Distinguished
Feb 6, 2011
13
0
18,510
right, but prime is pretty consistent in its load, and cpuz vcore readings have been agreeing with bios input voltage during prime everyday until suddenly it stopped yesterday. If there had been variations from the start it would not concern me since that would be normal. The fact that there hasnt been variations for a long time then suddenly there are, and all of them inconsistent variations is a bit concerning.
 
If CPUZ is showing voltage variations on the second decimal point to the right of the zero (0.0x, or the hundredth mark) I wouldn't be worried about it. A difference of 0.03 or 0.04 is probably not a huge deal, unless you're pushing your CPU/MOBO to their very limits of voltage safety.

Make sure it's stable, and your CPU stays well within thermal limits (temperatures). Keep in mind that power states set in Windows or in the BIOS can alter voltages on the fly. Many like to disable any Intel power management features in the BIOS to help stabilize voltages. Some motherboards also offer settings which facilitate a more stable voltage.
 

buphmin

Distinguished
Feb 6, 2011
13
0
18,510
@Bipedal Disaster before yesterday, i set 1.0625v in bios and cpuz read 1.062 or 1.063. or id overclock and set like 1.2000v and cpuz would say 1.200. then after i tried adjusting multipliers it started saying strange things. right now i have 23x mult with turbo for 24x at 133 base for 3.2 ghz clock. at this clock i have it set to 1.0625 volts. cpuz and the in bios hardware monitor says 1.056v.

@jerreece makes sense but why is it going all crazy on the voltage readings now? ive had this computer for about a year and have been toying with lower voltage for a few months and there has never been more than about .oo2v variation. After i felt i knew enough about i7 and x58 chipset i started overclocking early this week and there has been no voltage variation outside of .002v. like ive said i havent done any really crazy overclocking, 3.86ghz at 1.22v with about 64c temps stable in prime for 3.5 hours before i stopped. but just yesterday the voltage readings are all over the place.
 

buphmin

Distinguished
Feb 6, 2011
13
0
18,510
i assume you mean the hardware monitor within the bios, which states the current temperatures, fan speeds, voltages etc. looks something like this http://www.xbitlabs.com/images/mainboards/asus-p4p800/bios-3.jpg in here ive gotten results for the rails all over the place. last night they displayed in red +3.3v rail read 2.85v, +5v read 4.30v and +12v read 14.28v. today using the same exact settings 3.3v rail was 2.90v, +5v was 5.02v and +12v was 13.40v with the +3.3 and +12v in red +5v in blue.
 

buphmin

Distinguished
Feb 6, 2011
13
0
18,510
okay, im fairly certain ive concluded this. In short, my voltage sensors are now broken or at least the one for the cpu. now the evidence which led me to this conclusion. i put my voltage at 1.0625v the min stock clock w/ turbo that i gained before voltage reading oddities. cpuz and bios hw monitor said same thing of 1.081v and temperatures during prime were the same from ambient that ive tested before malfunction. later i rebooted did not change the settings and voltage now read 1.056v in bios and cpuz. temps from ambient were the same during prime. i then tried all stock settings with OS controlled voltage. read the temps which were the same from ambient that ive tested before malfunction with cpuz reading 1.28v where it used to be 1.24v. rebooted, manually entered 1.24v in bios reran prime. same temps at same ambient, cpuz states 1.28v again. i then reboot with same stock settings w/ turbo and put to 1.21v. bios and cpuz both claim 1.264v temps were appropriately less. reboot again and change nothing, bios and cpuz now read 1.216 volts and temps remained the same. then i overclocked to 3.7ghz with 1.21v and have the same temps as 3.2ghz w/ 1.24v. since voltage significantly changes temps i can reasonably conclude that the actual voltage going through the cpu isnt changing despite what the sensors say. Ill wait for what you guys have to say, but broken sensors seem to be the issue here. id like to know what all of you think about this conclusion. and if im right it sure does seem that overclocking CAN in fact damage your hardware, even at relatively safe overclocks lol.