Pentium 4 Overclocking

[bobross419]

Reading through posts I've seen mention that the old Pentium 4 chips run pretty hot. I have a Pentium 4 520 2.8Ghz Prescott and am working on my first overclock of a computer. There is plenty of information around about max temps for the C2Ds and such, but haven't really found any solid heat ceilings for the Pentium 4.

My current max temp is around 55C. This was during the 2nd test inside of Prime95 (1st and 3rd tests had lower max temps).

Is 55C too hot for this chip?

I currently don't have a fan on the front of my case and will be installing one soon, but a good max temp to shoot for would be great.

Thanks in advance for the help.

Bob

 

[alfa147x]

Originally posted by: bobross419
Reading through posts I've seen mention that the old Pentium 4 chips run pretty hot. I have a Pentium 4 520 2.8Ghz Prescott and am working on my first overclock of a computer. There is plenty of information around about max temps for the C2Ds and such, but haven't really found any solid heat ceilings for the Pentium 4.

My current max temp is around 55C. This was during the 2nd test inside of Prime95 (1st and 3rd tests had lower max temps).

Is 55C too hot for this chip?

I currently don't have a fan on the front of my case and will be installing one soon, but a good max temp to shoot for would be great.

Thanks in advance for the help.

Bob

wasnt their a poll and I think people said they were fine with CPU's running 70ºc max

 

[iBPJohn]

ya I'd think 70c would be a good temp at load to shoot for.

How high have you gotten so far?

 

[bobross419]

Right now I'm at 3ghz (2998 or some such number) and around 55C max temps.

 

[DSF]

What are you using for the temp measurements?

 

[bobross419]

I believe it is called HWMonitor and Speedfan. I'm not at that computer, so I can't say for sure which I was pulling the numbers from, but I remember them both showing the same.

 

[Drsignguy]

At close to 3Ghz and temps @ 55c, I would say that isn't to bad at all. I know It's a 200mhz bump but all considering that the P4 prescott's run a bit warm anyway ( I have a 3.6 ). If you have your load temps @ 30c or below TJmax, then all is good.

 

[bobross419]

Originally posted by: Drsignguy
If you have your load temps @ 30c or below TJmax, then all is good.

I don't understand what you are saying here. Sorry for being a noob, but could someone elaborate on what Drsignguy is saying?

 

[Drsignguy]

It means that whatever that chips highest rated temps is, stay below by 30c and you will be fine. An example, my Q6600 is rated at max temp of 105c ( using coretemp ) If I stayed below to 75c, I would be ok. Personally, I like to stay below 65c.

 

[bobross419]

So if I install the coretemp software it will tell me what the max temp is for my chip? I tried looking for that info on Intel's site, but couldn't really find anything

 

[Drsignguy]

CoreTemp. I think this will work with the P4 chips.:thumbsup:

 

[bobross419]

Excellent, thanks very much.


One of the benchmarking programs that I am using is 3dMark05 and I noticed that on the CPU tests my framerate is often dropping down to 1 or 0 FPS.

While at work today I decided to try and find some extra information about overclocking this particular processor and came across this article: Overclocking the Pentium 4 520 2.8GHz to 3.57GHz


On Page 6: 3DMark05 of the article I found the results of a 3dMark05 test. I was pleased when looking at the GPU results because mine are higher than any listed there, but looking at the first set of CPU results I was sorely disappointed.

In the article the Pentium 520 at stock speeds got a CPU score of 3668, and when the Pentium 520 was overclocked to 3.57 a CPU score of 4494 was attained. In contrast, my score for the 3dMark05 test is a dismal 1669 at stock falling to 1631 when overclocked to 3.0 Ghz (2.996).

Is this particular test really only dependent upon the CPU as the test's name implies, or is there something else with the system that might be having serious issues?

I'm using DDR400 Corsair Value RAM with a 1:1 divider (214 is the current speed) and my Sisoft Memory Bandwidth results are about 1000 points lower than the Pentium 520 at Stock speeds in the article. Could this be the reason why my CPU test is so low?

Thanks in advance for any assistance that can be provided,
Bob

 

[VirtualLarry]

Originally posted by: Drsignguy
At close to 3Ghz and temps @ 55c, I would say that isn't to bad at all. I know It's a 200mhz bump but all considering that the P4 prescott's run a bit warm anyway ( I have a 3.6 ). If you have your load temps @ 30c or below TJmax, then all is good.

I don't believe that there is such a thing for TJMax on a P4. At least P4s do not contain a DTS like C2D chips do.

 

[Drsignguy]

Originally posted by: VirtualLarry

Originally posted by: Drsignguy
At close to 3Ghz and temps @ 55c, I would say that isn't to bad at all. I know It's a 200mhz bump but all considering that the P4 prescott's run a bit warm anyway ( I have a 3.6 ). If you have your load temps @ 30c or below TJmax, then all is good.

I don't believe that there is such a thing for TJMax on a P4. At least P4s do not contain a DTS like C2D chips do.

I am going to take your word for it, but I am sure there is a max heat limit on the P4. Still, I would follow the guide lines of the more advanced chips where temps are concerned. Would you?

 

[bobross419]

I went ahead and installed CoreTemp, but when trying to run it I receive this message:

"This is not an Intel 'Core' Architecture based processor. This program will not continue."

Then I get a message that Coretemp has encountered a problem and needs to close lol.


And I'm using both the SpeedFan and the CPUID Hardware Monitor (HWMonitor) and both are showing the same temps.

 

[bobross419]

For the Prime95 testing should I be using the Small FFT, Large FFT, or Blended? I'm going with the Small FFT for now because I've got memtest to stress the memory...

 

[THUGSROOK]

blend

my Prescott hits 66*C @ 3.8ghz 1.525v.

Gallatin (Northwood) @ 3.8ghz never reaches 50*C by comparison.

 

[DSF]

I was going to mention that CoreTemp is not compatible with P4s, but it looks like you found that out the hard way. I haven't read a consensus on a program that accurately measures P4 temps, so I would be carefully watching what you're doing, and backing off at any sign of instability or slowdowns. I assume you're stress testing the overclock?

 

[VirtualLarry]

Now that I think about it, P4s don't have DTS inside like C2Ds, but they do still thermal throttle, using TM1 (thermal monitor 1). You can detect this with RMClock. So as long as you aren't throttling, you should be alright.

 

[bobross419]

I bumped up my OC last night to 3.2 Ghz and was unable to boot. I got some BSOD message about Page faults or some such. I should have written it down . I backed off to 3.1 Ghz and all seems well. My mobo doesn't allow for vcore adjustment, so this might be the best I'm going to get.


I ran Prime95 for 30 minutes or so to stress test. I live in Florida and the ambient in my room is pretty warm as is, so I'm waiting for a day where the fiancee and I go out for a few hours to run a longer test.

 

[DSF]

Originally posted by: bobross419
I bumped up my OC last night to 3.2 Ghz and was unable to boot. I got some BSOD message about Page faults or some such. I should have written it down . I backed off to 3.1 Ghz and all seems well. My mobo doesn't allow for vcore adjustment, so this might be the best I'm going to get.


I ran Prime95 for 30 minutes or so to stress test. I live in Florida and the ambient in my room is pretty warm as is, so I'm waiting for a day where the fiancee and I go out for a few hours to run a longer test.

30 minutes doesn't tell you much, and even a couple hours isn't a sign of complete stability. I've had overclocks fail prime at the 10 or 12 hour mark after being completely stable up to that point.

 

[bobross419]

Few hours = 8 or more

I appreciate you pointing that out for me though - I know I wasn't very clear on that in the original post.

 

[myocardia]

Originally posted by: bobross419
Is this particular test really only dependent upon the CPU as the test's name implies, or is there something else with the system that might be having serious issues?

Yes, it only uses the CPU for rendering, which is why it's so slow. It's 100% CPU dependent in its speed.

I'm using DDR400 Corsair Value RAM with a 1:1 divider (214 is the current speed) and my Sisoft Memory Bandwidth results are about 1000 points lower than the Pentium 520 at Stock speeds in the article. Could this be the reason why my CPU test is so low?

Yes. Raise your RAM timings to 3-4-4-8 (assuming it's stock timings are 3-3-3-8), and as crazy as it sounds, your performance will go up. If you've got the older 2½-3-3-7 Corsair Value, raise the timings to 3-4-3-8.[/quote]

 

[bobross419]

Alright, I'll try to tweak my memory timings a bit. I think they are rated at 2.5-3-3-8 (Maybe 7).

 

[jskill2003]

I tried doing this exactly and my CPU is fried. Now i get to buy something better haha