OC 4Ghz at very low temperature

Soldato
Joined
12 Jul 2009
Posts
4,878
I don't agree bulldog, like I have been saying all along, after setting CPU clock skew to 100ps, I can now use considerably less Vtt, and I also managed to knock Vcore down a couple of increments as well.

I have tested these voltages before, and my system would BSOD after 2 passes of LinX or a short run of Prime, but with 100ps CPU clock skew, it is LinX/Prime stable, I've had it like this for a couple of days now, and have used my PC for gaming, watching movies, listening to music etc, and I haven't experienced one BSOD yet.

To say we should leave these settings well alone makes no sence to me whatsoever..

Also, do you worship rjkoneill or something, you are always expecting him to come and answer questions, no offence mate, but there are plenty of knowledgable people on these forums, Rich isn't the only guy who knows his stuff!!!

LOL, its my OC bulldog is actually using.
 

rjk

rjk

Caporegime
Joined
8 Aug 2007
Posts
25,381
clock skews are for fine tuning

get your original clock stable first then do the clock skews after if you really need to fine tune it
 
Soldato
Joined
12 Jul 2009
Posts
4,878
So does changing CPU clock skew add any heat to the cpu? and does IOH clock skew help with anything.

Your right though my LinX failed after 20 mins, with 1.275 Vcore, i then changed CPU clock skew and it did 20 pass, result. Any other voltages i should tweak other than, ram, voce, qpi??

I'm not sure, any extra heat thats being produced is being masked by the lower vcore and/or QPI voltage.


CPU PLL seems to like being 1.3v instead of 1.8. so its worth tweaking that as well, do not go above 1.88v as I have heard thats a bit iffy. Thats the core clock voltage if you didnt know.

One of the guys who works for me has set his skew to 300ps, but on a Xeon. Its not really relevant to us, but might be worth looking into.
 
Soldato
Joined
6 Dec 2008
Posts
2,692
Location
Burghead, Elgin
I haven't witnessed any extra heat from adjusting CPU Clock Skew.

I thought by running Vcore and Vtt slightly lower than I had been previously would have made them go down, but my temps are still the same as before, so maybe it does add extra heat, but I didn't see it because I also lowered Vcore and Vtt.

@Rich, I am glad you didn't think I was offending you in anyway mate, as it wasn't the case at all, like I said, I know you know your stuff ;)
 
Soldato
OP
Joined
22 Mar 2009
Posts
5,712
I did my homework:

cpu pll: 1.3v
qpi/vtt: 1.2v
cpu vcore: 1.26v
dram: 1.5v
pcie: 1.5v
qpi pll: 1.1v
IOH core: 1.1v
ICH I/O: 1.5v
ICH Core: 1.1v
cpu clock drive: 900mv
pci express clock drive: 700mv
cpu clock skew: 400ps
IOH clock skew: 200ps
191 x 20 (3.82gb) with intel turbo boost enabled
Memory timing: 9-9-9-28-2t
qpi link: x36 (6.87ghz)
Uncore Frequency: x16 (3056mhz)
Memory multiplier 8.0 (1528mhz)
Performance Enhance: standard
PCI Express Frequency: 101mhz
Base clock: enabled
BCLK Frequency: 191
C.I.A.2: disabled
Rest of it leave as stock

Here is the result of my 1 hour: (idle and max temperature much improvement with mx3 thermal compound paste)

2hd06t1.jpg
 
Associate
Joined
15 Oct 2007
Posts
2,183
Location
Bedfordshire, UK.
Im not sure about CPU Clock Skew, something is too good to be true!!!, with it enabled im able to get a stable overclock (well an hour of LinX) with a lower Vcore than my cpu's VID!! My VID is 1.26250, and at the moment im at 1.25625 vcore, yet my temps arent really any lower. It just doesnt add up
 
Last edited:
Associate
Joined
15 Oct 2007
Posts
2,183
Location
Bedfordshire, UK.
yeah from the bios?? Have you actual had a stress test fail yet when youve had 100ps?? Just strange how without any ps, my system wasnt stable at 1.275, but with 100ps my sytem is stable at 1.25625, thats like 3 increments lower
 
Last edited:
Soldato
Joined
6 Dec 2008
Posts
2,692
Location
Burghead, Elgin
I've had several runs of LinX fail me when I tried using CPU Clock Skew to begin with, but raising Vcore up 1 increment sorted it out..

I wouldn't trust the BIOS for getting your CPU's VID, as Gaidin said it is unreliable, sometimes I enter my BIOS and it tells me that my VID is 1.1935v, and other times it says it is 1.244v or something like that, and it gives different readings when turbo is enabled also.

There has even been times that I have been in the BIOS and I have watched the value change right in front of me!

If your system is stable, then don't worry about whether you are running vcore less than what your VID is in the BIOS, just be happy that you have a stable system!!! :)
 
Last edited:
Soldato
Joined
12 Jul 2009
Posts
4,878
Just means the timing is better. We are all still fiddling about with it. My bios says my VID is 1.256v but cpuz says 0.975v. Also another vcore reading in bios says 1.23v and ET6 says 1.275v so who knows what the VID is.
 
Soldato
Joined
12 Jul 2009
Posts
4,878
Having a low VID doesnt guarantee a good overclock, you can have a great VID and a crappy IMC which means you cant get a decent clock at low vcore or QPI and vice versa. messing about with the timing between the two can help.
 

rjk

rjk

Caporegime
Joined
8 Aug 2007
Posts
25,381
example
my chip has a vid of 1.1v
which is fantastic - but needs a heafty increase in vcore to go from 4ghz to 4.3ghz

however, it is one of the few 920s i have seen that can run 2000mhz+ memory as well as a large cpu clock

its swings and roundabouts
 
Soldato
Joined
6 Dec 2008
Posts
2,692
Location
Burghead, Elgin
Set your BIOS to optimized defaults, save and exit, enter Windows, and startup CPUZ, the voltage that it says is your chips VID!!!

bulldog said:
Well, mine is stable all that I care ! But, I cannot see any problem with that. Done that 3.82ghz (191x20) with turbo enabled and done linx and all passed.

Can't see any problem with what???

We know you are stable, you post several threads each day with your Prime95 and LinX results... now go and use your PC for something else other than stress testing!

I personally don't think that it can be doing your system much good stressing it out like this ALL OF THE TIME!
 
Back
Top Bottom