Intel Overclocking Club

[amuffin]

This is as far as I'll take my 3960X on air, http://www.meyithi.com/graphics/oc.png, 10pm NW England so I'd want a good 10c leeway just in-case we ever get a decent summer in the UK

Specs in sig

I'm jelly

 

[mrstab]

Nah you can keep it here. I mean...It kinda is the purpose of this thread.

Ram timings can be an issue. you also want to make sure you disable:

All "C" states
Spread Spectrum
Speed Step

gah! so many differences in opinion. before i set up this rig last month, i read this long-winded rant on another forum (which i can't find) about Spread Spectrum being preferable for Sandy Bridge CPUs. could disabling it result in lower temps/greater stability/both?
all C states are off, i had Speedstep off at first for testing but it didn't seem to affect anything when i enabled it. don't you have speedstep enabled, as per the core speed in your screenshot? i'd agree it's best left off for testing at least.

i'll go back down a few volts on the vcore, try disabling Spread Spectrum first, then if that doesn't work reset the RAM timings & i'll let you know how it goes. cheers!

 

[DarkOutlaw]

gah! so many differences in opinion. before i set up this rig last month, i read this long-winded rant on another forum (which i can't find) about Spread Spectrum being preferable for Sandy Bridge CPUs. could disabling it result in lower temps/greater stability/both?
all C states are off, i had Speedstep off at first for testing but it didn't seem to affect anything when i enabled it. don't you have speedstep enabled, as per the core speed in your screenshot? i'd agree it's best left off for testing at least.

i'll go back down a few volts on the vcore, try disabling Spread Spectrum first, then if that doesn't work reset the RAM timings & i'll let you know how it goes. cheers!

Spread Spectrum lowers EMP fields. Its kinda like the tin hat vs aliens or satellites theory imo. Its mostly used in work environments where you have hundreds of computers. Speedstep, I do not know the exact numbers, but it lets your CPU go from 1.6 to 2.5 to 3.3 to max. with it disabled it goes from 1.6 (idle) to max (4.6 for me). This way there is no throttleing, because your CPU will constantly try and use a lower multiplier. Both will affect stablility, I dont think temps will really change though.

 

[mrstab]

cool, i think i get you with the disadvantage of speedstep. still seems like a pretty divisive subject, but i'll turn it off and only test it with it on again if i manage stability.
i did read recently that Spread spectrum was better suited to environments with loads of systems, which made me question the other source i mentioned, but by that point i had been using it for a while and hadn't noticed this temp/voltage issue.

what's the deal with the short/long power duration setting? can't remember the exact name, but i wasn't sure so after some experimentation i found stability (4.5ghz) at ~200w. the TDP of my cooler is 150w, although i added another fan which reduced temps about 10 degrees.

i'll actually reboot after this post and change stuff instead of procrastinating.

 

[DJDeCiBeL]

cool, i think i get you with the disadvantage of speedstep. still seems like a pretty divisive subject

And I'm one of those that swear by SpeedStep and think that you should never really disable it (unless it's causing instability). It hasn't ever caused a problem for me (absolutely NO instability issues caused by SpeedStep). Maybe I'm lucky or just maybe, it's not THAT big of a deal.

It USED to be a problem with previous generations, but from Sandy forward, I don't think it really is.

 

[mrstab]

And I'm one of those that swear by SpeedStep and think that you should never really disable it

i'm inclined to agree more with this view, less power consumption when you don't need it and it seems to have been fine with me thus far.

ok, so i disabled speedstep and sp. spectrum to no avail, but then i did a coupla other things: i disabled this setting called "Package C States", must've overlooked it as i went straight to disable the more obvious C3 & C5 a while back. then, on a hunch because my bus speed is 102MHz just to get a better-looking rating in Windows lol, i thought hmm, maybe i should turn up my RAM voltage a little bit. so i turned it up to 1.365v (from rated 1.35), booted and passed 10 runs of IBT. i'll give it another go though as i was probably just lucky and i have a pessimistic outlook lol.

btw, Coretemp is reporting my VID as going up to 1.42v even though voltage is at fixed and CPU-Z reads it as 1.352. wtf?

edit: passed again, but just noticed max temp on one core hit 90 on the first attempt (lower on the 2nd). i wonder if it's worth trying to lower the vcore if it was the RAM voltage that fixed it

 

[DarkOutlaw]

If you have an Asrock board try setting LLC to 3, If you have another board set the LLC to whatever 50% is.

SpeedStep is a great power saving feature...but on that note most all of us throw power saving out the window just by reading this stuff. It's more personal preference, me personally when I am using my computer want that power there without it needing to be called/calculated.

 

[DarkOutlaw]

And change your FBS back to 100, that will cause issues even if you shoot it up by 1. Intel is made to work @ 100 or some odd some such. Only AMD chips should need the bus changed.

 

[mrstab]

And change your FBS back to 100.

i am well and truly out of ideas so i'll try that.

unfortunately i can't find an option for LLC in the UEFI (P67 Pro3), hence the droop.

i've been spending the past coupla hours trying various things but no luck.

edit: changed the bus to 100mhz but still BSOD in IBT at 1.35v. maybe ive just hit the limit of my chip

 

[DarkOutlaw]

What are your voltage levels for 4.4 and 4.5?

 

[mrstab]

i put it back to 4.5 earlier actually, passes IBT repeatedly at 1.330v, could probably put it down a bit more though. that's the weird thing.

 

[4Ryan6]

@amuffin

The darker blue background on your stable OC spreadsheet would be easier to read with white lettering.

 

[DarkOutlaw]

i put it back to 4.5 earlier actually, passes IBT repeatedly at 1.330v, could probably put it down a bit more though. that's the weird thing.

Nah weird was it takes me 1.38v @ 4.5 but 1.328v @ 4.6

 

[deathengine]

Name: DeathEngine
CPU: i5 2500k
Motherboard: Asrock Z77
CPU Voltage: 1.36
CPU Bus Speed/Multiplier: 52
Clock Speed: 5200
RAM: 8 GB Corsair xms3 9-9-9-24
Cooling: Noctua D-14
OS: Win7 Home Premium

Not stable but low voltage

 

[mrstab]

Nah weird was it takes me 1.38v @ 4.5 but 1.328v @ 4.6

lol that is pretty weird, maybe you have an antimatter CPU.
thanks for the help, looks like i'll be stuck back at 4.5 unless i get some better cooling.

 

[DarkOutlaw]

4.5 is perfect for everyday use. Even 4ryan6 who is running 5.2ghz? pushes his down to 4.5 for 24/7 use. Or am I wrong on that?

 

[4Ryan6]

2nd Entry:

Name: 4Ryan6
CPU: i7 2700K @ 5300mhz
Motherboard: ASRock P67 Extreme 4
CPU Voltage: Bios set @ 1.650v
CPU Bus Speed/Multiplier: 100/53
RAM: 8g > 2 x 4g G-Skill 1333 @ 9,9,9,24
Cooling: Water Cooling http://www.tomshardware.com/forum/275185-29-exploring-ambient-water-cooling
OS: Win7 Professional 64 bit

Stability as requested:

 

[4Ryan6]

^ They left my door unlocked and I got out of my room! :lol:

 

[amuffin]

20 18 20 20

5.3ghz

1.65V.


WTF!

 

[4Ryan6]

Isn't it wonderful!

 

[4Ryan6]

20 18 20 20

5.3ghz

1.65V.


WTF!

That was the idle temp at the screenshot!

Load during Intel Burn was: 64~73~72~70

Water Temperature during the test was 9c, that's 14c below my ambient room temperature.

Water pump was on level 5, maximum flow rate.

 

[Max1s]

http://valid.canardpc.com/show_oc.php?id=2284794
uhmm..not stable :lol:
Cooling was bong evap+ice

Holy crap, you guys are all crazy, but thats why I love you.

Bus speed is over 105 mhz? How did that go?

 

[4Ryan6]

http://valid.canardpc.com/show_oc.php?id=2284794
uhmm..not stable :lol:
Cooling was bong evap+ice

You've got yourself a really good 2500K!

 

[Max1s]

You've got yourself a really good 2500K!

Well at almost 1.7v you'd assume he'd get pretty high. I still can't get over these voltages I'm seeing!

 

[4Ryan6]

Well at almost 1.7v you'd assume he'd get pretty high. I still can't get over these voltages I'm seeing!

He used a Bong Cooler and added Ice to control his water temperature, my cooling is linked above.

It takes extreme below ambient cooling to even think of taking voltage levels up there like that, especially shooting to stabilize at those type of clocks.

Though I stabilized at 5300mhz, I had no intentions of leaving it set like that, no one should even attempt something like those clocks without the cooling necessary to do so.

The sanity of it even with the cooling is debatable, Intels top end voltage test range was 1.520V, exceeding that is just flat risky and something done at your own expense.