Please just spell it out for me, reccommended settings.

I've spent a few hours searching and reading and still can't get past all the confusion. Most of the FAQ's i've found are years old.


Please, just give me some basic advice on what CONSERVATIVE settings you know I could safely use for my system. I'd be happy to get 3mhz.

P4 2.4(b) Northwood 533 fsb
Asus P4PE
RAM is PC2100

my interal clock is 2405, system clock is 133.64, system bus is 534.54 and multiplier is 18.

I know this is easy for you guys, but confusicating to an old man.

Thanks!

Scott

Well the first thing you will have to do is go into the Bios (Hit delete I believe on bootup) You won't be able to change the multiplier but you should be able to change the system bus to like 136 without problems. That's 3 mhz higher. It should put you about 54 mhz higher. Your system should be able to handle that without a problem. When you overclock you never know how high you can go. Just take little steps at a time and you should be able to find out the highest stable speed you can go.

lock your pci/agp bus to 66/33 and youre set

thanks for the advice, but I guess you didn't spell it out for me clear enough because I incorrectly stated my goals?

My clock is at 133 and you are telling me to change it to 136? 136 x 18 is 2,448?

What do you mean 54hz higher, from 2.4 to 2.45?

I must have completely stupidly phrased my question. I would hope to get my 2.4MHz running at 3.0 MHz. Or whatever you experts think I could safely do with my setup.

So, the only option/parameter I need to work with is the FSB that is currently at 133, just move that up? I don't have to change the vcore?

And Derivative, I will set my agp as you've suggested.

THANKS.

I do not know what you can safely put it at and no one else does. All I'm saying is that you need to go slowly with little steps. If you go too high at first you might damage something. I would raise the system bus at a stable point and when it starts to get errors or stops booting properly then you should raise the Voltage core to stablilize it. I just can't tell you to raise it to so and so and expect it to work. Every machine is different and the fun part is trying to get it to the highest you can go and have it run stable.

k, thanks, I apprecaite it, I just raise it a step at a time then. I guess I'll wait 24 hours between steps to see if there is any problem. And I"ll watch the temp too.

I went from 133 to 134 with no problem. Let's see what happens next?

You don't have to go that slow. If you have a benchmark program like Sandra or the 3dmark series then you can test it to see if it is stable then go higher immediately. You can get alot of benchmarks at this site Guru 3d

salling said:

k, thanks, I apprecaite it, I just raise it a step at a time then. I guess I'll wait 24 hours between steps to see if there is any problem. And I"ll watch the temp too.

I went from 133 to 134 with no problem. Let's see what happens next?

Click to expand...

it'll take you forever if you only go 1 FSB higher then wait 24 hrs to check for stability

try 5 steps at a time till you hit instability
then lower the FSB 1 step

yeah same as adjusting jets in a carburator go up 5 or 10 and if you have a problem go down its easier to go down 1 or 2 the go up 1 or 2 till ya find the sweet spot

cool,

Thanks @ 2.6 now.

Hi,

If your "safely" means not spoiling your CPU:
If you wil not raise the Vcore, you can not spoil an Intel CPU even if you start feeding them by absurdly high FSB. It just will not function at all will or give up the work after few seconds of action. But it does not hurt him.

Even if you will raise the Vcore in a reasonable range (10% over the specs is a acceptable for Northwood generation CPUs - yours one is this genereation) you can not spoil the CPU. If the resulting core speed wiil be too high for it, your system will hang or lock up, but nothing happens to your CPU. Raising the Vcore more than 15% over the specs gets it into conditions which can't be marked "safe", but anyhow as I have find out while playing with hardware, it is not just risky, but counterproductive too, because even one can reach stable operation of core for a few seconds or minutes, the core then overheats and the system hangs. And the overclocked system which isn't stable does not worth all that effort. Perfect stability is more important than an extra 1 or 2% of speed, in my opinion. I consider raising the Northwood core CPUs Vcore to 1.575 - 1.65 V is the happy mean, except for cases when you are using superduper cryo cooling. My celeron 2.1 gives me the better results for overclocking at 1.575V (140 MHz with stock heatsink).

If you mean "stable" when you say "safe", it's very easy to state whether your system is stable at a given frequency. There are programs which does nothing else just burden the CPU with a dozen of computing tasks and it gets it in the fully laoaded state. If you have some hardware monitoring tool you can notice that CPU heats up. And if it hangs, you have to go down with your FSB or with Vcore, until you reach tje state, that such testing software can run for several hours without a crash. Then, your system is rock stable

One of such stability testing programs is "The CPU stability test 6.0" by Jouni Vuorio.

My recommendations of how to play with your system is as follows:

1. Raise the CPU Vcore to 1.6 V. You don't have to afraid, it's harmless, because it is just 5% over the Northwood's default Vcore (1.525V)

2. Choose the slowest FSB:RAM divider. It means, that your memory will work slow, but this is the state, when the crash of the system can be caused only by CPU. For example, if you have an DDR333 or DDR400 memory, which is capable running stable up to 166MHz (200MHz respectively), then choose 1:1 divider, which will drive your RAM at the same frequency as the CPU (assuming you will start OCing from 133MHz). In case you have DDR266, choose a divider somewhat under 1:1, for example 4:3. Moreover, choose the slowest timings for memory. These are the memory timing settings:
- CAS Latency (or DRAM Cycle length)
- RAS to CAS delay
- RAS precharge (or Row precharge)
- Tras/Trc (or Row active time or Minumum RAS pulse width)
Choose the highest possible numbers for these settings in your BIOS.

3. Set the CPU FSB to say 150 MHz and boot the system. The run the CPU stability test. If it does not fail (it shouldn't), riase the FSB to say 155 MHz and so on with 5 MHz steps, until the stability test starts hanging in first 30 minutes. Then go back by 1 MHz and when it will run stable for over 12 hours, you have reached the limits of your CPU. I hope the wild overclockers excuse me for not mentioning advices like "go in the shop and buy one 150W peltier or cryocooling unit ..." it would be a whole different story.

But in general, it is better to run the P4 systems at limits of the system memory, not at the limits of the CPU. So, how you can find out the limits of your memory? The way is similar, just the stress is moved from CPU to memory.

For now, I think it is enough to have play with it for at least a few days. Maybe somewhat later I will write aomething about finding out the limits of your memory and what to do after you know both the limits of the CPU and the memory.

I am sorry for this broken story, but I am in a hurry right now and have to leave this "twadding"