Currently my P4 Prescott is running at 3.6GHz from the default speed of 3.0GHz on an Abit AI7 with a small voltage boost of 0.05v.

Temps for the CPU on full load with Orthos are 67C, PWM 85C RAMs are 4x256MB Vitelic SAT5Bs DCed running on 3V on 240MHz 2.5-4-3-7 1:1.

PAT is disabled, and to top it off, I'm using a stock heatsink

Anyone can give advice on what I can I go further?

Haha, knew that you would say that

Anyway, are anyone here enabled PAT? When I enable PAT it stunts my overclocking by a lot. So far from my tests, dropping latencies do not make much of a difference at all. (I used to have an AMD )

Also, does running FSB and RAM speeds async kill performance like AMDs do? I haven't done much tests on that yet.

I'm still running stock.

I've just discovered a problem on my rig, recently I found out that it restarts by itself during SP2004's Large FFT test.

When the room is cold, the PC will not restart. By this clue here it's obviously heat related.

I've tried decreasing and increasing RAM and Vagp voltages, but it doesn't help the problem. However when I dropped the processor voltage by a bit the test lasted slightly longer.

Now, the big mishap is my processor runs 7 hours+ without errors with Orthos(using the Gromacs core test), and also runs Memtest86 4 hours+ without any errors, with the same voltages before I found out the problem. All tests were done in room temperature. Yet, the SP2004 Large FFT test fails within 3 minutes or more.

Anyone with advice on this one?

Previously it was 70C, but now after reducing a bit of voltage to the processor, temperatures are now 66C after a voltage reduction to the processor, for 4 hours of flawless Gromacs testing.

After doing a few tests, my computer now fails after 12 minutes in the Large FFT test, but gives me a fatal error instead of a restart.

Full load temperature is now at 65C.

Seems that your Northwood's hitting it's limitations eh? What's the temps on yours?

I'm gonna try dropping my RAM voltages (3V to 2.8v) and redo the Large FFT tests to see if there's any improvements.

I always use Priority 10 for SP2004.

So far, dropping the memory voltage doesn't help in prolonging the Large FFT test. Today I ran the test for an hour in a cold environment, no problems at all. Just found out, My CPU voltage has been dropped to stock, and it's able to run at 3.6 GHz for 4 hours Gromacs test.

My guess is that the RAMs could be hot. Funny though, I have a RAM cooler installed...

Is the processor your brother is giving you a Prescott? I wanna compare temps and overclocking ability

Your overclocking results should be higher, since you have a better board

Take your time, dude. We don't want any smoke coming out of your PC, do we?

Hope your overclocking goes for the best.

I tried pulling my FSB higher (240 to 245), but I guess my 4x256MB SAT5B RAMs aren't helping much. Time to start tightening timings then.

I read somewhere that more slots occupied on the board will stress the memory controller further. Is this true?

And if i remember correctly, Tictac used to modify BIOSes, but his site is currently down to inactivity. Anyone kept his modded BIOS for the AI7?

You have the same processor as I do Currently my processor is running at 3.6 GHz stock voltage, 65C full load.

Yours should do about the same, or maybe better with good RAMs. Also, are you sure that your 9600 PRO cannot be used? If I'm not mistaken, 9600s use 1.5v.



@kabukiawie

Just found out about Tweak865 from your post Mind explaining the settings and your experiences with it?

This post has been edited by colinwong89: Apr 26 2007, 03:59 AM

After toying around with Tweak865, I have PAT enabled, although partially from what CTIAW reported

I gained an extra 100MB/s from Sisoft Sandra's Test, however I have yet to test for stability problems

---

Added on April 26, 2007, 4:51 pmI found a few modded BIOSes, however they are only for the IC7 and other ABIT overclocker boards.

Seems there's no one actually modding the latest BIOSes for our boards, all of them seems to have retired or something

This post has been edited by colinwong89: Apr 26 2007, 04:51 PM

Well, a modified BIOS could increase overclockability on a motherboard by editing romsip tables, unlocking hidden options, etc. That was for the NForce2 series boards though.

I wouldn't mind an E6600 if it was free

It was stickied on some French forum, I'll get the link for you soon as I'm currently out right now.

http://www.hwtweakers.net/postx4741-0-0.html

Only a few BIOSes for the P4, but better than none. The page is Italian, btw

@raymond5105
Sorry, the one I saw on the website was for the MAX3 version, not for the normal IC7. Nevertheless you could still have a look at it.

@shinjite
Will do if I manage to find them.

To those people still using stock Intel heatsinks, how fast do your fans rotate? Mine's only about 2700RPM max; did a fast check online to find that the speed is controlled by a thermal sensor.

Any way to set it to maximum speed?

Speedfan can detect my the fan's RPMs and motherboard voltages, but it doesn't give me any options to increase or decrease the fan speed. My fan doesn't spin faster during primes, just stays at 2700 RPM max.

If it's really noisy as you said, well let's just see if I can take it

This post has been edited by colinwong89: Apr 27 2007, 06:56 PM

Crap. Speedfan doesn't support regulating fan speeds for the Uguru boards as it will 'cause a huge overhead'.

Looks like I have to find something else.

Anyone here tried CPUMSR before?

I checked my PC using Sisoft Sandra, and they reported that Fast Strings was disabled. Using that particular program you could enable it.

According to ARP, Fast Strings should be enabled to improve performance. This is supposingly an option in the BIOS too.

Maybe my AI7 leaves it disabled, will run a few tests later to see if performance levels improve after enabling. There are many other options, such as Data Error Checking, Response Error Checking, and Address/Request Error Checking, which are all disabled.