P4 1.8A @ 2.3Ghz - Why can't I go higher?

[hcranman]

I just upgraded my system a couple of days ago and as soon as everything was set up and stable, I went ahead and overclocked her. After reading on here for a while and given my chip's operating temperatures and such, I can't figure out why I can't squeeze a better overclock out of chip. Here are my components:

Mobo: Asus P4B533
Chip: P4 1.8A (Retail HSF w/ Radio Shack Thermal Compound) - Pack Date 12/26/01 (Kinda Old)
Ram: 512MB Samsung PC2700 DDR (in third slot, if that matters)

Right now the chip is running at 2.3Ghz (128Mhz FSB) at a reported ~1.69v idle, 1.664vcore under load (BIOS set to 1.6v). Ram settings are default except for 3:4 multiplier, giving 340Mhz DDR. The interesting thing, and what's got me confused, are the very low CPU temps. PC Probe reports 43C (109F), which from all I've read is pretty darn low. I haven't run an extensive Prime95 torture test yet (several hours is max so far), but the system appears to be totally stable (lot's of "testing" in Warcraft III ).

If I run the chip at 2.3Ghz @ 1.575v (BIOS), it's unstable -- Prime95 crashes with a couple minutes. Also, if the fan is running at low RPM's, 1.6v (BIOS) is also unstable with Prime95 crashing quickly. 2.344GHz @ 1.6v (BIOS) - same deal. However, I can get the chip to boot into windows at 2.4Ghz at 1.625v (haven't tried 1.6v yet, I don't think), but the voltage is too high for my tastes (~1.74 idle), and it crashes in Prime95 almost immediately. I'm not sure if the RAM was set at 3:4 or 1:1 for the 2.4GHz testing, but I'm pretty sure it was 1:1. I haven't tried any higher on the CPU, but it may be possible.

I'm thinking it's one of several things:

1) The RAM is at fault -- my old sys died, so I had to go to a local vendor for the components, and I'm not positive about the specs on the RAM. If anyone can tell me what to look for on the chips, that'd help me identify exactly what I've got so I can decide if this is an important variable.

2) Since the chip is probably fairly old (I didn't get a chance to take a look at the stepping or manufacture date before I installed the chip), even though it's nice and cool, it just can't hit a higher speed.

3) However, since when the fan slowed down 2.3Ghz became unstable, maybe the issue is heat. How much of a difference would some good thermal compound and a new, heavy-duty HSF make?

What are your thoughts, guys?

 

[Mikki]

Hiya hcranman, your chip could definitely be one of the suspects, I actually think you've done good to get it that high at that voltage with an older chip. You don't have much voltage to play with (1.75v actual max), you're probably near the end. I personally think your cpu temp is a little high, my 1.6a runs 32c idle @ 155fsb with default voltage, and goes up a max of 2c idle when I bring the voltage up to your level. I know it's a bit different with a 1.8, but I don't think it's that much. But maybe that's close to what a 1.8a would be at 1.69v.

You can call your system stable in my opinion (I would still do an all-day Prime95 run at some point) with what you've done so far. If the ddr pcb has Samsung on it as well as the chips, you've got good ddr. Try putting it in a different slot and try some less-aggresive timings.

Heat might be affecting it, but I just think you're at or close to your limit. You could try some AS3, that MIGHT bring your temp down. Also, you could try a different hsf, I think the stock hsf's are great (you see my temps), but in your case it would definitely help.

HTH and good luck!

 

[cnhoff]

My 1.8A has the following pack date: 05/01/02

Can't get it stable at 2.4 without upping the vcore to 1.725V

2.3 even was very unstable at 1.6V.

I guess some people just are not that lucky...

 

[Mikki]

cnhoff, what were your temps with those voltages?

 

[Barrei]

Classic case of not enough juice to cpu try upping to 1.7v and if you can't go higher then , it's your particular cpu has hit it's max output.

 

[hcranman]

Well guys, looks like I've probably just got a bum chip. :/ Prime95 crashed at some point during the night while I was asleep even though the CPU temps were stable. Damnit. The really frustrating thing is that the system runs great at this speed, but I can't run it here without feeling like I'm going to crash or something. Time to back it down and see where I'm stable for sure.

At this point it does seem very likely that it's voltage and/or cooling. Unfortunately, I'm not willing to risk running the chip at a higher voltage because I really don't want to damage it. Anyone have any recommendations for good HSF's to replace the stock Intel with?

 

[Jgtdragon]

Go into Bios and set it to 1.65v. I heard Asus reads the bios voltage higher than it really is. Abit boards reads it lower. I set my in bios as 1.625 and with MBM5 it reads 1.56v. I have an Abit board BTW.

 

[Duvie]

Jgtdragon,

Are you sure it is a reading issue???? Where is your evidence to that effect?

I think some study of asus boards versus epox boards in my own experience is that asus is actually delivering more juice at a given bios setting...my 1.8a took 1.55v bios (1.58v actual) to run 2.25ghz on asus mobo....It needed 1.65v bios (1.58v actual) to run in epox...HMMMMM....

I think if you look at earlier mobo (i845ds) that suffered from reboot issues that needed the wire mod trick employed, asus mobos often did not need them do to the overvolting out of the gate. I am not talking about using the vcore booster on some of the boards either....

Interested in seeing where you base that conclusion on....

I think it is not a reading issue but the different manufacturers applying different voltages...

 

[hcranman]

Interesting thought, Jgtdragon. However, I must say I am inclined to believe the reporting of the motherboard. Regardless, I've decided to go ahead and give it a bit more juice. I'm running her at 2.3Ghz with a reported ~1.72v idle, 1.68v under full load. We'll see if I can't get her stable at that voltage.

Am I going past the "healthy" voltage limits of the chip? What's the most one should go past? From what I've read on the boards, some believe it's ~1.71, others ~1.75.

 

[jhites]

Am I going past the "healthy" voltage limits of the chip? What's the most one should go past? From what I've read on the boards, some believe it's ~1.71, others ~1.75.

I have read many posts in the forums where ppl are having to run their chips at default after a couple months of 1.7vcore.

 

[hcranman]

Originally posted by: jhites
I have read many posts in the forums where ppl are having to run their chips at default after a couple months of 1.7vcore.

Hmmm, kinda makes me want to just sell the chip and get a new one. That may very well be what I end up doing.