Pentium 4 Northwood Overclock

Associate
Joined
22 Feb 2006
Posts
747
Location
Belfast
hi all.

i have a p4 northwood 3.2ghz and 2gb of crucial ballistex pc4000 400mhz ram.
my system is running at 3.6ghz with no extra cpu voltage. i have the ram voltage at 2.7v and my motherboard is an Abit IC7-G. my system is watercooled and at present my idle temp is at 44 degrees.

cpuz.JPG


I was wondering.. What is the max overclock my cpu will take. and what kind of extra voltage is needed for it. some assistance would be much appreciated.

Thanx
 
Only you can tell as every chip is different. Voltage is less critical than heat. If your watercooling is keeping the chip nice and cool, don't worry too much about a little extra voltage. Try giving it a 0.1v boost and see how far that gets you. RAM can easily take 2.8v and most can do 2.9v without active cooling.
 
i had my system running at 3.81ghz and the cpu voltage at around 1.65. i done a torture in place test with prime95. on full load my cpu temp was 65 degrees
. after 20 minutes or so it failed. the pc froze. does that mean i need more voltage ?
 
crazeconnolly said:
i had my system running at 3.81ghz and the cpu voltage at around 1.65. i done a torture in place test with prime95. on full load my cpu temp was 65 degrees
. after 20 minutes or so it failed. the pc froze. does that mean i need more voltage ?

Depends. How far overclocked are you running your RAM? That may be causing the drop-out as well. How stable are the voltages when you overclock?
 
i think all northwoods cap out around 3.8....

you see lots of people with 3.5-3.8ghz overclocks but not many over (unless they are using phase change cooling.)

it is strange that it will do 3.6 on stock volts but wont do 3.8 at 1.65, myabe you have just hit the cpu limit
 
To rule out the possibility of the RAM holding the CPU overclock back, go into the BIOS and select the divider so that the RAM speed is running at a lower MHz than the CPU FSB. Once you've done this, carry on to find the maximum stable CPU overclock.

Also, make sure the RAM timings are not tight (eg. 2-5-2-2). Reduce them. You can always tighten them up when you discover the max CPU overclock.

The problem I find with Northwoods is that they can go from 3GHz to 3.6GHz on default volts (or very little voltage increase). But to go an extra 100MHz or so higher, you need to ramp the voltage up quite considerably just for a few extra MHz.

Mine can do around 3.6GHz on default volts. Thats 600MHz with no voltage increase. 3.7GHz is attainable but with 1.65v.

Like all CPU overclocking temps are important and you really want to be keeping those load temps at around 60oC or under.
 
Voltage kills more than temps, and if you need a considerable amount of vcore to attain a small 100MHz etc.. then you've hit the core limit and personally i won't do it as its not worth it.
 
Thats basically what I was getting at. If the CPU needs more than 1.6 - 1.65v just to get an extra few Mhz then it isn't worth it. The extra voltage and heat that could kill the CPU isn't worth another 100Mhz that you won't even notice.
 
So do u reckon 3.7ghz would be ok on 1.65v Guru ?

Ill have a try at that. also....

What should i put my ram at. the timings etc. and the volts???

c
 
After some trial and error. I have managed to create a stable overclock. The overclock speed is shown below. I was wondering is this a decent enough overclock speed. Idle temp is 48 and Full load temp is 63.
Ram is at 2.8V (5:4 ratio)

cpuz.JPG


Or.... can i overclock more with this specific chip ?
 
MikeTimbers said:
Only you can tell as every chip is different. Voltage is less critical than heat. If your watercooling is keeping the chip nice and cool, don't worry too much about a little extra voltage. .

Gotta disagree, voltage is much more critical than heat, in the long run. Remember Northwood 1.6As etc.. dying even when cooled to minus temps? Heat is critical for stability in the now, overheating dosen't usually kill a chip but vcore kills over time, electromigration etc is sped up with more voltage and even the best cooled chips can't run away from that. Too much voltage and your chip can be dead in days/weeks/months instead of the years it was supposed to live. The smaller the process, the more intolerant to overvoltage and the higher the voltage the lower the cpus maximum thermal threshold is, another factor in overvolting.

Personally i would'nt go to 1.65v on a Northwood on air cooling and even better. Most of my PIVs at and over 2.8 did at least 3.4 to 3.8Ghz on stock voltage. Also when you say stable overclock, do you mean prime and 3D stable? Those few hundred mhz with all that vcore are not worth in in my opinon, but thats just me.
 
Actually looking back, I don't think my [email protected] was higher than 1.6v. I remember doing the droop mod on my Asus P4C800E to cure the voltage fluctuation. So yeah, I'd agree with Justintime and say 1.6v is probably the max you should put through it.

When you took that screenshot, what voltage did you have selected? If it was 1.65v, it looks like your motherboard is undervolting. So 1.65v looks like its actually delivering around 1.61v. Monitor the voltage with another program to check.

If you went from 3.2GHz to 3.6GHz on default volts and then had to bump it up to 1.65v to attain 3.7GHz, I'd personally leave it 3.6GHz. The extra heat produced and higher chances of killing the chip isn't worth an extra 100Mhz that you won't even notice.

With regards to RAM there are 2 things you can do and should try.

1. Leave it on the 5:4 divider. Tighten the timings to the lowest they'll go. At that CPU FSB of 232MHz, the RAM should be running under 200MHz. You should be able to tighten them up to 2-2-2-5/6.

2. Deafault timings of that RAM at 250MHz is 3-4-4-8. Return the divider back to 1:1 and select those timings. Should boot up fine as your running the RAM well below spec. Now you'll have to reduce those timings to improve performance. With Intel systems, the most import 2 timings are the middle ones (RAS to CAS and RAS Precharge). Reducing these will give you a much better performance boost than reducing the CAS Latency (1st number).

So try 3-3-3-8 and test stabilty. Pass? Try 3-2-2-8 and test again. If that passes then your doing well and won't be able to reduce those 2 important timings any lower. Then try 2.5-2-2-8 and test. I think you get the picture now.

With both cases, run some benchmarks with the 5:4 divider and 1:1 divider options and see which one returns the best results.

That RAM is rated by Crucial to run at a default RAM voltage of 2.8v, so just leave it at that.
 
Back
Top Bottom