E6600 Overclock

Mullet

New member
On the overclock I had some success yesterday:

Ran the pc using an option within the bios which clocks the system by 20%, this took my cpu up to 2.9Ghz. I had a browse around on the web to find out what others had been doing and tweaked a few voltages and managed to boot @ 3.6Ghz (unlinked the memory and FSB) then manually inputted 1066Mhz for my RAM and 1600Mhz for the FSB. The FSB is quad pumped so you have to divide that figire by 4 to get the true FSB....400Mhz. As I say, the PC booted and got me all the way to the black windows screen (you know the one with the windows 'flag' on it. The image of this screen was a little faded...then the system restarted before I was able to get into the desktop...weird. Tried upping the cpu to 1.5v still no joy....the Northbridge was running at 1.45v (stock is 1.3v) and is not very well cooled so didn't want to pump more juice through that...well not until I get some better cooling on that!! What d'ya think would cause that?? Maybe I should try running a 5:4 memory divider?

After trying to get into desktop @ 3.6Ghz I gave up (need to read around a bit more) I dropped the FSB from 1600Mhz (400) down to 1550Mhz (387.5Mhz) which gave me a CPU speed of 3.48Ghz. I got back into windows and run a benchmark 3DMark06....was obviously expecting good things....

But...was shocked to find my score @3.48 was lower than the score I achieved with the cpu @2.9Ghz....why d'ya think that happened?

Help guys!!
 
OK well for a start never use the automatic overclocking features ever.

Set it urself in the BIOS

1.5v on the CPU isn't enough for 3.6GHz in my experience.
 
It depends on the CPU. My e6600 needs 1.65v for 3.6Ghz which is far too much, even tho temps are still ok

I leave it on 3.4GHz 24/7
 
My E6600 needed a lot of volts as well. To be honest mate, you probably wont notice much difference between 3.3 - 3.6 for normal use anyway, however, it will increase your E-
cussing.gif
cussing.gif
cussing.gif
cussing.gif
cussing.gif
greatly :D

Edit ^^ that wasnt a swearword!
 
am I being stupid or is throttling an issue on processors. If it is still a feature could the temps be too high causing the processor to throttle back?

Dave
 
Nah, temp are barely above 30 degrees. And when I was benching they didn't break 38 thanks to my new water loop.
 
name='Mullet' said:
Kemp....in your experience is 3.6Ghz a do-able 24/7 overclock? Or am I setting my sights too high?
Most deffo do-able. But the majority require too many volts for a sustained 24/7 oc.

name='Kempez' said:
It depends on the CPU. My e6600 needs 1.65v for 3.6Ghz which is far too much, even tho temps are still ok

I leave it on 3.4GHz 24/7
Holy crap! I thought mine was high @ 1.5
 
Woah! Seems like I am not the only one with a moody chip!

I can go up to 3/3.1ghz stable on stock volts, 1.45v for 3.3 stable, 1.5v for 3.4 stable... Then it is no longer orthos stable but will f@h 24/7 simultaniously gaming @ 3.5 needed 1.575v... All that was on water.

With the pelt temps are over 20*C cooler so I gave it more juice!

Max OC so far is 3.78ghz with 1.6/1.65v - only super pi stable!

f@h/game stable @ 3.6ghz 1.6v...

Anyway I am changing my setup for far superior rads and fans so by mid March I should have got all my stuff and my RMA ram back and a sweet setup :)

Bottom line is you decide what is 'safe'
 
Stealth-Li now running @ a shade over 3.6Ghz. I reckon my 1004 beta BIOS is helping me along!!

Vcore in BIOS 1.5v in windows 1.47v

NB - 1.5v

CPU VTT - 1.35v

RAM - 2.35v

SB - 1.55v

Not Orthos stable though....ran Orthos for 20 minutes or so and the test failed. At these settings when running Orthos the cpu temps max was in the mid 60s and the GTXs again stayed @ or below 50. These settings are OK for gaming but really would like to get the system Orthos stable at these settings....
 
my E6600 takes 1.525V (set in BIOS) to get orthos/prime stable at 3.6GHz

my E6700 is a bit better and will do 3.6GHz stable with 1.4375V

nice setup Mullet, we have fairly similar components ;) :beerchug:

have you experimented with memory timings and 1T yet?
 
name='Pyr0' said:
my E6600 takes 1.525V (set in BIOS) to get orthos/prime stable at 3.6GHz

my E6700 is a bit better and will do 3.6GHz stable with 1.4375V

nice setup Mullet, we have fairly similar components ;) :beerchug:

have you experimented with memory timings and 1T yet?

Thanks alot bud.

No, aint messed with mem timings yet....what you advise I lower first??
 
erm...

depends really on how high you want to take them, whether you want to use 1T or 2T, and the voltage you are prepared to put through the sticks

i've tried to keep my mem voltage at around 2.1-2.25v for 24/7 use

it took me 2-3 days just to find some stable subtimings, and i'd still like to find time to try to take the mem slightly higher

at the moment i'm finding DDR850 4-4-4-8 1T to be faster than DDR1000 4-4-4-8 2T and 850 1T only takes around 2.1V - although i used a multimeter to check and this board overvolts the mem slightly by 0.06-0.07V

my mem timings are set in the bios atm as:

(stability tested with hci memtest 4000%)

TCL 4

TRCD 4

TRP 4

TRAS 8

CMD 1 clock

TRRD 3

TRC 5

TWR 4

TWTR 7

TREF 1

TRD 3

TRFC 35

Async. 2.00ns

voltages:

CPU 1.525V

MEM 2.100V

1.2HT 1.30V

N.B. 1.50V

S.B. 1.55V

CPU VTT 1.40V

if anyone else has any tips, please post :beerchug:
 
Unlinking is fine -- to find the limits of individual components....but...to keep the channels of communication running to as nice a ratio as possible I would opt for a divider, I tend to use 2:3 or 4:5 depending on what I am trying to do. If I opt for the higher of these then my max OC at reasonable RAM speeds is 3.00ghz (ish)....with all stock voltages. Then OC I have noted earlier in this thread was done at 2:3 (if I remember rightly).

Hope this helps Tox.
 
name='Toxcity' said:
Thanks matey! :)

So divider is best.. or is said to be best!?

Yup....you can always calculate your own divider and input the speeds yourself....so you'll do this in unlinked but...it's run on the divider you have worked out (FSB / Memory).

Think of the Linking and divider thing as a process which ensures the FSB and memory talk to each other at a speed they can all 'hear'...any move away from this may reduce performance as the 'talk' gets all muddled...

Excuse my analogy ;)
 
name='Toxcity' said:
Thanks matey! :)

So divider is best.. or is said to be best!?

*supposedly* linked and using "sync mode" is fastest as sync mode is supposed to reduce internal chipset latency

i didn't notice any diff in speed when running unlinked 1:1 and linked, w/sync tho *shrug*

personally i just test with a quick movie encode, super pi or 3dmark to see which is faster

i've just increased my 1T mem clocks slightly to ddr2-867 (faster than 800 1:1 sync or no sync)

the only mem timing i changed was TRD to 5, voltage is still 2.125v

1.2HT = 1.40

NB = 1.45

SB = 1.50

VTT = 1.45

fsb=400 (1600), mem ratio is 12:13
 
Back
Top