O/C'ing my new processor Q's

[Kruzin]

So, I upgraded my CPU from an XP 2000+ (133/266x12.5 Barton) to a 2800+ (166/333x12.5 Barton), and I want to see what I can get out of it.

Any suggestions for speeds I can expect, and a good way to test stability? I dumped the 2000+ because it wouldn't O/C worth a squat...went with the 2800+, cause it's the best I could find in town that my mobo would support (my mobo will do a 333fsb 3000+, but those have gotten really, really hard to find locally)

Last night, I clocked it to 180FSB, and it wouldn't boot at stock voltage (1.65v). I upped it to 1.70v, and it booted and seemed to run ok, but when I tried to run 3DMark03 on it, several of the tests did not display (?). The comp didn't freeze, and niether did the app...some tests just didn't show up on the monitor, resulting in a crazy high 3DM score...though I could see from the CPU tests that did run properly that I was getting speed increases. I thought the O/C might have been the cause of the tests not running properly, but when I went back to stock clocks, I got the same failures with 3DM'03. Due to this, I don't think I can rely on 3DM to test stability.

Suggestions welcome...

[Tjalfe]

I am running a 2500+ @ 1.75V 200MHz FSB, which is 2.2GHz ( 3200+) .. it is stable..

On another computer I am running a P4-E 3.4GHz @ 3.82GHz stable

[Jon P. Inghram]

I don't know why, but on my system Starcraft is extremely sensitive to overclocking.

When I was running my Celeron Tualatin 1.2 at 1.44 GHz (120 MHz fsb) I couldn't find any indication of instability (Prime95, Climate prediction, 3DMark, and so on,) but when I put Starcraft on the dang thing'd lock up every time, either in the inital menu or in game. Bumping the core voltage up a little stopped the crashes, but I'd rather go easy on the CPU so I dropped the voltage back down and dropped the speed to 1.428 GHz (119 FSB) which ran Starcraft without crashes.

[Marshmallowman]

If you can't change the multiplier, you could try modding to the mobile version and use overclocking tools in windows pick fsb and multipler

I think if you have decent 400mhz ram you may be able to set the bus at 200x12.5 if you have GOOD cooling and max out the cpu core voltage

my xp2500 is only good for 11.5 200 used to do 12 x200 with an old bios but could not cold boot(water cooled)

[Kruzin]

JPI- I don't have Starcraft. Any ideas on freeware diags that can be run overnight or all day to test stability on?

MMM- Can't change the multiplier, and aint going searching for a mobile proc. My RAM is PC2700, so I don't know if it'll go as high as 200. Would be nice though .
Cooling doesn't seem to be an issue so far. When running at 180FSB (2.2GHz), the core never went over 45c. And that was with my Thermaltake HSF at only 4500rpm (will go up to ~5800).

This was all sparked by a new case I got a couple weeks back, which has decent cooling, as compaired to the zero case cooling I had before. Was hoping the improved case temps where gonna help O/C the old 2000+, but it didn't...

[Tjalfe]

sisoft sandra has a burn in test, which can stress the CPU / RAM .. I used it myself to test the stability of my P4

http://www.sisoftware.co.uk/index.html?dir=dload&location=sware_dl_x86&langx=en&a=

[Jon P. Inghram]

Originally posted by Kruzin
JPI- I don't have Starcraft.

Prime95 has a torture test mode that can be adjusted to stress the CPU and cache, system RAM, or a mixture while checking the results for errors.

SuperPi (or any of those Pi generators) might be a good test too since it's easy to check the results for validity.

[GNEP]

I'm going to sound a bit like Umf here, but of all the progs that have shown whether I'm 100% stable or not, CPDN (classic version, not so fully sure of BOINC version in how close to the bone it gets yet) - if it ran stable for a week, the computer was as good as it can get. If not, you get an iceball world in the first phase - even a couple of smallish mem errors in a week at load can throw it off. Also gets out the way nicely when you want to do other stuff. Won't test any graphics card o/c though

http://www.climateprediction.net/index.php

[Taz]

Yep any kind of distributed project is a good way to test for system stability. I use Folding@Home myself, but that's because I'm part of a team so would be running it anyway

What does you motherboard have in the way of dividers, you don't want to run the AGP and PCI buses too far out of spec

Oh and this is the post which has transformed me into a Super MURCer And it's only taken 5 years to get there

[The PIT]

Too be honest is there much point these days.

Someone had taken there Prescott from 2.8 to 3.57 for around 9fps extra which I doubt you'd notice. AMD 64's still whipped its ass though.