OC Stability

[zendyn]

I recently bought a new CPU, the XP1800+ with 1.5v core. It overclocks quite well. I'm getting 2080mhz pretty easily. If I had a 6/1 divider, I'm sure it go quite a bit higher. However, my computer seems to be rebooting about once every few days. Can this be due to the PS? Or should I bumb up the core voltage a bit? The PS is currently the one that comes with the Antec Lanboy. 350 claimed watts I believe. How much does the PS influence the overclocking ability? I'm a little new at this and wanted to know if any of you have any ideas. Here are my voltages...

3.3V reading 3.5V
5V reading 4.84V
12V reading 11.86V

I have no idea what the negative values mean or even if they're connected to the right sensor, but here are the values anyways. I think they may be attached to the wrong sensor or maybe the mobo doesn't support these readings through MBM.

-12V reading -7.17V
-5V reading 0.89V

These are the specs to the system.

1800+ JUIHB-DLT3C running at [email protected] FSB 180.4Mhz
Memory: Kingston Hyper-X PC3000
PowerSupply: Antec 350W SmartBlue
Mobo: KX7-333
CPU cooler: AX-7 w/Panaflow H1A
Idle temp: 45
Load temp: 52

If someone here knows why it may be rebooting, I'd appreciate any ideas.

Thank you

 

[Pinky]

Power supply looks like it's holding up, 4.80 on the 5v rail is the magical number, below that and you're going to run into issues... you're at 4.84, I ran many overclocked PSUs at 4.82/4.80 and they were stable, and Antec makes a decent overclockers psu.

If that temp is in-socket (and not the internal cpu diode) then you're pushing 62-65C core temps, that could explain it.

Could also be memory instability, not enough voltage for the fsb and memory timings you're trying to run it at.

Oh, and welcome to the addiction .

 

[hawk072]

It shouldn't be the power supply. That PSU should be plenty to power your system. I'd guess you've reached the limit on that CPU at 1.6v. You may need to back down the speed or up the voltage. It could also be the fact that you are using 180 which is also overclocking your pci and agp bus. Or your memory speed or timings.

I'd run some stability tests and see if they tell you anything. Prime95 would be my first choice, then try memtest to test the memory.

My XP2100 would do 13x166 at 1.6v and everything in windows seemed to be fine. I tried Prime95 and I'd get Illegal Sumout errors every time. I had to bump the voltage up to 1.75v to get it totally stable. Each time I upped the voltage, Prime would run longer. I think it made it somewhere between 6-9 hours at 1.7v. At 1.75v I ran it for 48 hours with no errors.

 

[Pinky]

I'd run some stability tests and see if they tell you anything. Prime95 would be my first choice, then try memtest to test the memory.

My XP2100 would do 13x166 at 1.6v and everything in windows seemed to be fine. I tried Prime95 and I'd get Illegal Sumout errors every time. I had to bump the voltage up to 1.75v to get it totally stable. Each time I upped the voltage, Prime would run longer. I think it made it somewhere between 6-9 hours at 1.7v. At 1.75v I ran it for 48 hours with no errors.

Good suggestions to run some tests.

Another I thought of, some processors cannot run a higher fsb but CAN run a higher clock. Try increasing the multiplier and leave the fsb at 166 (leaves everything in spec, which is what I do versus pushing the limits).

 

[zendyn]

It could also be the fact that you are using 180 which is also overclocking your pci and agp bus. Or your memory speed or timings.

I'd run some stability tests and see if they tell you anything. Prime95 would be my first choice, then try memtest to test the memory.

It could be the AGP/PCI bus speed, but I doubt it. I'm running a 5/2/1 divider so my AGP should be 72Mhz and PCI 36Mhz. It's slightly oc'ed but not by much. I know many people going higher without any problems.

I've tried running Prime95 for around 10h once and no errors. The temp under load stabilized at 52 degrees. I believe the CPU can be pushed even higher and memory capable of going to higher FSB's as well. I guess I'll try raising the DDR voltage and core voltage a little and see what happens. I'll probably need better cooling soon hehehe...

Thank you for all your help.

I've noticed that rarely does OC'ing give you more in performance than the money spent. For the price of a better PS, cooling and memory, we probably could have gotten a higher rated cpu. BUT all said, it's a hobby that I'm starting to like

 

[Pinky]

I've noticed that rarely does OC'ing give you more in performance than the money spent. For the price of a better PS, cooling and memory, we probably could have gotten a higher rated cpu. BUT all said, it's a hobby that I'm starting to like

OMG, you were born to read this thread

http://forum.oc-forums.com/vb/showthread.php?s=&threadid=149942

 

[hawk072]

I've noticed that rarely does OC'ing give you more in performance than the money spent. For the price of a better PS, cooling and memory, we probably could have gotten a higher rated cpu. BUT all said, it's a hobby that I'm starting to like

Many times this is true. I was never a huge fan of overclocking. I've built many systems in the past and this Athlon XP2100 is the first I ran overclocked full time. I've bought compenents in the past planning to overclock and then I couldn't overclock as much as I expected so the the performance gain was not noticable. So I would just leave everything at stock speed to not risk stability.

This time my successful overclock was well worth it. I'm running th same speed as an 2700+ which was well over $200 when I bought my 2100+ for about $85.

I also prefer to isolate the compent I'm overclocking to make it easier to track down what piece of hardware is being pushed too far. That's why I stick which standard FSB speeds and go slow...unless you have an NForce2 board with the PCI lock which is VERY nice.

 

[zendyn]

I read quite a bit of the thread. VERY interesting prespectives on this whole thing. I've been playing around with hardware since my venerable 8086 and have obviously upgraded many times since...

I've learned early on that there is usually no good time to buy computer parts. Technology changes so quickly and standards become obsolete so fast that the only logical thing is to buy what you can afford and what you need/want. Hardware is indeed overpriced as end users end up doing a lot of R&D and testing for the companies.

But it's also business. Everything is worth what people are willing to pay for it. If companies can get away with having ppl paying for their unfinished products and testing them while they improve on it, then that's fine. It would be up to us to be more careful. I'd rather be smart in what I buy than dumb and have manufacturers only come out with good products. I'm sure many would disagree, but that's my opinion.

Anyways, all the ranting and comments aside, I will continue to tweak my setup until I'm happy with what I've done!

Thanks for everyone's help...

 

[zendyn]

I just found out that my CPU is now unlocked! I didn't know that KT333 mobos could unlock TbredB's. I'm pretty sure the last L3 bridge is not closed. Maybe the latest BIOS revision gave it the ability! I'm pretty happy now.

It seems to be running quite stable at 2254.63 Mhz (12.5x180). I'll put it through a few hours of Prime95 and see how it holds up. Does anyone know what's the highest PCI bus speed it would be safe to run at? Right now it's at 36 Mhz. I'm wary about pushing it a lot more since it may mess up my HD data. I don't have a 6/1 divider that I now see could be very useful!