AthlonXP 2600+ Barton Overclocking Problems

[sheenster38]

I've been having problems recently (within the past 2-3 months) with my comp randomly shutting off. It usually occurs while I'm playing games or something that requires graphics, so logically, I assumed it was my graphics card that was having problems. But a couple of days ago, I ran Prime95 (stress testing for those who don't know what it is) and when I got back home around 7 hours later, the computer was shut off again. As I've read before, Prime95 does not focus much on the graphics aspect of the computer during its stress testing. At the time, my computer was being tested under 170x11.5 @ stock vcore, whereas 166x11.5 is stock for a 2600+.

My specs:

Abit NF7-S v2
AthlonXP 2600+ Barton
Geforce 3 Ti200
Thermalright SLK-900A (120mm)

All of the hardware at the moment is running at stock speed again just incase anything wants to fry itself. After the comp shuts off randomly, the hsf seems rather cool (which is expected for an SLK-900A), but the ramsinks on the graphics cards are very hot.

I have no idea at the moment what is going on anymore. I'm trying to prevent the comp from shutting off randomly and maybe overclock the cpu if it doesnt turn out to be a cpu overheating problem.

Let me know if you need any more details or have any questions whatsoever. Thanks in advance for the help.

 

[c627627]

Cheapo power supplies have been known to cause this

or

Autoshutdown is kicking in because your CPU is getting too hot.


Welcome to the forums.

 

[||Console||]

will 200X11.5 not boot ? (prob will need more V)
what speed is your ram?
the rails on your psu are they stable (run prime and watch)
have you ran mem test ?
do you have the agp/pci lock turned on ?
maybe more vcore ?

 

[Ruiner]

I vote for the automatic thermal protection....it happened to a friend of mine with that board.

The diode for auto protection reads higher than the one in bios/mbm. (The onboard one for thermal protection is in a hotter part of the core than the bios one, to keep users from freaking out when they see it). Set the auto off temp up a notch or two and see what happens.

You may have to reseat your HSF.

 

[enduro]

I vote either psu or shutoff due to heat. Use a multimeter and check your rails while gaming or running prime. Also, like what was said above, up your temp shutoff point, or reseat your hs and use quality TIM if you haven't all ready.

 

[sheenster38]

c627627: Thanks for the welcome I don't think that my cpu is getting too hot because the high shown on MBM5 is only 52C. I've only known cpu's to shut off at around 60+C or so. It might be a power supply problem, I'm using the one that came with my case (Antec Lanboy, I believe). The ps is the Antec SL300S 300W.

||Console||: I've never tried running my comp at 200x11.5 provided that it can't even handle Prime95 stress tests on 170x11.5 I'm using 2 sticks of Buffalo PC3200 256MB. About the rail thing, I'm not exactly sure what you're referring to, so if you can clarify, that'd be great. I'm going to run memtest86 today to see if it's a memory problem as one of my friends suggested. I don't believe my AGP/PCI lock is on, but I'll check as soon as I finish this post. And finally, from my knowledge, if the vcore is too low for the cpu speeds, the computer shouldn't even boot properly. I may be wrong, so I'll try increasing the vcore later.

Ruiner: How do you change the auto thermal protection? I've never had any experience with doing that, so if you could explain briefly. I was also going to try and reapply some arctic silver to the SLK-900A and the gpu fan.

enduro: Again, no idea what these rails are.

Thanks for all the advice, I'll check on all of this and post back again tonight with my results.

 

[Ruiner]

52C isnt too high for the cpu diode, but if your auto shutoff is 60C, you are probably hitting it. Look in 'PC health' section in bios....that's where the setting is on my shuttle. It may be called 'thermal protection' or 'automatic shutdown'.

 

[sheenster38]

I checked the BIOS under 'PC Health' and it's already been disabled. I just took off the hsf's for my cpu and gpu and reapplied some arctic silver. Still gonna run memtest86 and see what turns up.

Oh yeah, after I took off my slk-900a, I noticed almost all the sides were covered in dust. I just vacuumed it all off and my cpu temp dropped to 48C.

 

[sheenster38]

I just tried pushing my fsb up to 200 @ the stock vcore. And expectedly, the comp didn't even hit the boot screen. I ended up raising the vcore repeatedly in .25v increments up to around 1.775v and the comp still didn't boot. On the way up to 1.775v, the comp shut itself off twice.

 

[enduro]

Rails meaning your voltage rails. So on your molex connector, you have a red, yellow, and to black (ground) cables. Take a digital multimeter, and measure the yellow by placing the red multimeter tip into the yellow socket and the black tip into a ground socket. This will give you your 5V reading ( I believe) and the red will give you your 12V rail, by doing the same thing. If they are around .15 or .2V off from the standards, then you need a new psu, or you need to adjust the pots if you can. EX: 12V rail reads 11.80V, then you need to do something about that.

 

[sheenster38]

Ok, thanks for the explanation. I'll test it out tonight.

I just ran Prime95 a couple hours ago @ 166x11.5 (stock speed) and the comp ended up shutting itself off again. I still have to do the memtest later too.

Edit: MBM5 displays 4 readings labeled +5.00, +12.00, -5.00, and -12.00. Are those the voltage rail readings? If so, are these accurate readings?. The +5.00 category is reading 5.00V and the +12.00 category is reading 11.86V.

 

[enduro]

No, you can't read them from Windows accurately. Some people get 3V on the 5V rail. Do you have a multimeter?

 

[sheenster38]

Yeah I do, I was just curious as to whether those readings from MBM5 were accurate.

 

[enduro]

Gotcha. Also, if you want to read the 3.3V rail, take your P4 power connector, and place the red tip into any orange cord, you have to do it from the top though. Place ground in black ( of course, but I need to say it). Anyway, good luck with that, and make sure you do it while your cpu is under load.

 

[sheenster38]

Apparently I don't have a multimeter at the moment, but I'm borrowing one from a friend so I'll test it this weekend.

Anyone know any reasons?

 

[enduro]

Have you reseated you heatsink yet? Also, set the cpu heat protection to 80°C just for the heck of it, and see if that helps a bit.

 

[sheenster38]

Yeah I already reseated the heatsink. I cleaned off all the old thermal paste residue and reapplied the arctic silver. I set the shutoff temp to 75C (max shutoff temp) just now, I doubt it'll change anything, but it's worth a shot.

 

[enduro]

Yeah, you have to try at least once. Some of the most unlikely stuff has worked on my rig. I've tried many things that haven't had any affect either. You just have to search the forums, and dig through all of the "fixes" for problems, sometimes the weirdest solution is the one that works the best.

 

[sheenster38]

Hmm, well I tried running prime95 again last night with the updated settings. So far, none of them have helped. My comp shut off again during the middle of the night.

Any other solutions anyone can think of?

 

[sheenster38]

I just ran memtest86 for 18 loops and all of the test loops passed. So I'm assuming that the problem has nothing to do with RAM.