what does random reboots and corrupt files w/OC mean?

Jeremy · Post by **Jeremy** » Sat May 01, 2004 6:00 pm

once again i tried overclocking my desktop. it's a 1.8A so i'm really surprised with my lack of success.

i boosted it to 132MHz (the max i can use) for 2.37GHz. the pci/agp was set to 1/4 for 33/66, and the memory ratio was 2.0 for effective 264MHz. when windows boots up it says tons of files are missing or corrupt and turning back to 100 makes it work fine again.

i tried values from 122 to 130 all using 1/4 fsb for pci, and 2.0 for memory so it would never pass the original frequencies of 33/66/266. every single time it eventually gets a random reboot after a few hours of FAH, and the temp never goes above 45-48c in all cases.

would teh bottleneck be teh CPU, memory, or a peripheral? the one thing i'm unable to do is boost the core voltage if that means anything.

thanks for help,
Jeremy

CableDude · Post by **CableDude** » Sat May 01, 2004 6:52 pm

Overheating?

Cypher · Post by **Cypher** » Sat May 01, 2004 6:54 pm

You did lock your agp/pci bus at 66/33?
Are you sure your MB is indeed locking it?
I ask because missing or corrupt files points to data corruption due to the IDE being out of sync. I had that when I set mine too high to see what happens. It corupted the OS and I basicly had to reformat.

If it's crashing under load like that than it's not stable.
You should be able to get something out of it even if you can't increase your voltage. Not much because the voltage increases the current to the CPU when it's taxed by OC'ing it.
It could be a device that's folding under the higher settings or other things as you already suspect.
Under start and recovery in system properties there's an option for system failure. Try unticking those three options and see what happens than.
Hope that helps you a little.

Cypher · Post by **Cypher** » Sat May 01, 2004 6:56 pm

You did lock your agp/pci bus at 66/33?
Are you sure your MB is indeed locking it?
I ask because missing or corrupt files points to data corruption due to the IDE being out of sync. I had that when I set mine too high to see what happens. It corupted the OS and I basicly had to reformat.

If it's crashing under load like that than it's not stable.
You should be able to get something out of it even if you can't increase your voltage. Not much because the voltage increases the current to the CPU when it's taxed by OC'ing it.
It could be a device that's folding under the higher settings or other things as you already suspect.
Under start and recovery in system properties there's an option for system failure. Try unticking those three options and see what happens than.
Hope that helps you a little.

*EDIT
Could be a heat issue like Cabledude said.

YARDofSTUF · Post by **YARDofSTUF** » Sat May 01, 2004 7:39 pm

He's saying the pci is at 1/4

Ya u need it locked at 33mhz can you adjust the vcore a bit?

What mobo?

Jeremy · Post by **Jeremy** » Sat May 01, 2004 8:34 pm

for sure i know that the pci and agp buses never went above 33/66, and went as low as 30/60 at the 120fsb setting which also had similar problems.

the board in question is a gigabyte GA-8IRX, bios F11. it was mid-high end at the time, but as i said the core voltage can't be changed as far as i know. in addition, the GB monitor that came with it notes the vcore goes from 1.500 to 1.490v under load if that means anything, although the fan remains constant (below; the moment i shut off FAH, the vcore goes to 1.500v). my case fans are on normal power plugs so it doesn't show up.

*also about the heat issue, the temp reading never goes beyond this, and with my bios set to beep at 60c (the lowest setting), it has never done that yet.

YARDofSTUF · Post by **YARDofSTUF** » Sat May 01, 2004 8:45 pm

Is the ram set to host clock or host clock + 33? I would set it at just host clock

U sould be able to OC that that pretty damn well.

Cypher · Post by **Cypher** » Sat May 01, 2004 8:51 pm

Don't worry about the vcore drooping. Intel reccomends that as part of their design.
Your rails look good. The 12V is a little low but it's within spec.
The RPMs on your CPU are pretty low. A higher RPM fan would do you good either way.

Is the ram set to host clock or host clock + 33? I would set it at just host clock

Here's a link on your MB for you:

http://www.ntcompatible.com/story8686.html

Try here and check for latest drivers and Bios.

Jeremy · Post by **Jeremy** » Sat May 01, 2004 9:03 pm

about the ram clock...basically i can set it as auto, 2.0, or 2.66 vs the FSB. when i treid overclocking i set it to 2.0 (it displays the frequency below it, and it read correctly at 240-264MHz; auto resulted in a frequency that was far out of spec for my memory rated to run at 266). i shouldn't need to overvolt the ram or the AGP because they are still within original specs.

i have a spare p4 supply, would that improve anything for potential power issues?

the 2500rpm fan is based on stock cooling. in the past it went as high as 3500rpm automatically, but after fiddling with the rear exhaust, it's able to stay lower.

the newest bios which i have, f10, simply supports new C-stepping cpus up to 3.06GHz.

thanks for the suggestiosn guys!

Cypher · Post by **Cypher** » Sat May 01, 2004 9:15 pm

I dont' think the other PSU would really matter. Than again it may, it's hard to tell. You really can't trust those SW programs nor the Bios when it comes to accuracy. The best way it to use a multi-meter to test it.
You should have a nice OC'er there. *scratches head* I wish I could of more help to you here.

Jeremy · Post by **Jeremy** » Sat May 01, 2004 9:48 pm

one thing i forgot to mention was that the thermal tape on the CPU is a little non-attached and screwy. when i first got the system, the seller had put a willamette p4 in it and it had advertised a northwood so i took it back. i think they may have swapped the cpu and added nothing else.

in spite of that, the temps seemed ok still so i didn't bother adding anything. i wonder if getting some new goop would help.

*also changed psu, 12v line is 11.9-11.96 now.

Cypher · Post by **Cypher** » Sat May 01, 2004 10:28 pm

Jeremy wrote:one thing i forgot to mention was that the thermal tape on the CPU is a little non-attached and screwy. when i first got the system, the seller had put a willamette p4 in it and it had advertised a northwood so i took it back. i think they may have swapped the cpu and added nothing else.

in spite of that, the temps seemed ok still so i didn't bother adding anything. i wonder if getting some new goop would help.

*also changed psu, 12v line is 11.9-11.96 now.

That's it than hopefully. That tape is shot once you remove the HS.
I'd put some thermal compound on that. I like AS5, or you can get whatever you can score local.
Make sure you remove all the waxy residue from the tape. Best way is to use 90% isopropyl alcohol and a lint free cloth. <hint> Instant coffee filters work great.

If the residue is rather stubborn use a little acetone than go over it again with the IPA.
*Make sure you're grounded though. CPU's are very vulnerable to ESD.

Touching something metal should do the trick as long as you don't move around allot and generate static. Don't ground to your case either. It would only act as a Faraday cage. Allot of people do that and don't realize it does that.

This could be what's been holding you back. :2cool:

Jeremy · Post by **Jeremy** » Sat May 01, 2004 10:59 pm

alright maybe that'll do it! i actually removed the heat sink last week as well without temperatures changing so i'll guess that it was already bad to begin with.

AS5 is expensive! i did a quick search and they're above $10cdn here! is it because of the high silver content?

Cypher · Post by **Cypher** » Sat May 01, 2004 11:12 pm

Yeah that and it's rep for being a top performer. Besides it does last quite a while.
Honestly the difference between different TIMs, (Thermal Insulation Materials) is generally a few degrees give or take. Get what you can find for a good rate where you're at.
Whatever you decide to get, Artic silver's site has great instructions on how to apply it properly.
http://www.arcticsilver.com/instructions.htm
It pretty much applies across the board.

Jeremy · Post by **Jeremy** » Sat May 01, 2004 11:51 pm

well i found AS3 on ebay for $10 shipped. is it much difft than AS5? my server is running without thermal compound so i figured i may as well get it.

Cypher · Post by **Cypher** » Sat May 01, 2004 11:59 pm

Not much. Both are really good though. AS5 has a thicker consistency and may have like 1 degree on it or so. That's a good find for $10 shipped up there.

You should see a nice decrese in temps compared to what you're running. Just remember that it takes a few days to fully cure.

Jeremy · Post by **Jeremy** » Mon May 03, 2004 12:11 am

just ordered a tube of arctic silver 3 online...hope this solves all my problems!

-Jeremy

Jeremy · Post by **Jeremy** » Wed May 12, 2004 8:11 pm

anything higher, i got corrupt registry hives or rebooting cycle. this is with AS5 just put on. is it possible that the pci divider isn't actually kicking in, or the AS5 has to burn in for a few days? the settings are PLL/16, PLL/20, PLL/24, and PLL/32 (with a 133 bus, PLL/16 read 33/66 for pci/agp in the bios)

sentra8777 · Post by **sentra8777** » Thu May 13, 2004 6:34 pm

sounds like over heating or even damgaged processor. I had a 3.2gig not oc but it was defected and thats why i keeped getting random reboots.

TrevGlas · Post by **TrevGlas** » Thu May 13, 2004 7:12 pm

http://www.arcticsilver.com/arctic_silv ... ctions.htm

Follow these instructions for best heat transfer... straight from AS themselves

P4 instruction start at #10

Jeremy · Post by **Jeremy** » Thu May 13, 2004 8:10 pm

in fact that's where i started; i followed it precisely. i'm thinking it's not the cpu. i removed the fan, and it ran perfectly forever not reaching 50c. then i put the fan back in, bus to 125, and it crashes. and the settings say the memory is running at 2*fsb (250 in this case), and the pci running at 1/4 the fsb (about 31/62). could it be just a really really bad cpu? i see most ppl reach 2.4 stock with no adjustments for vcore.

A_old · Post by **A_old** » Fri May 14, 2004 11:09 am

i didn't read, but did you already run a memory integrity test? i.e. memtest or something.

Jeremy · Post by **Jeremy** » Fri May 14, 2004 5:56 pm

Amro wrote:i didn't read, but did you already run a memory integrity test? i.e. memtest or something.

i haven't, but when i boot up the bios does say "ram frequency ### MHz" and at the standard 1.8GHz, it says 266, and when overclocked to 125 for example it would show "250" after i override the ram ratio to 1:1 (125:125) rather than 3:4 (100:133). so it's actually running a tad slower than before, and if it ran perfectly stable at 266, the memory should run stable at 250 right?

brembo · Post by **brembo** » Fri May 14, 2004 6:59 pm

Corrupt files smack of memory issues. I have one stick of RDRAM that is mighty picky about how install it, and I know almost immediatlly when I've goofed it. The comp will shutdown, make oddball read/write errors and hose the system files if I don't get that stick back in the slot *just so*.

lansens · Post by **lansens** » Fri May 14, 2004 7:19 pm

brembo wrote:Corrupt files smack of memory issues. I have one stick of RDRAM that is mighty picky about how install it, and I know almost immediatlly when I've goofed it. The comp will shutdown, make oddball read/write errors and hose the system files if I don't get that stick back in the slot *just so*.

Agreed..........

Radium reboots in most cases are going to be caused by two things, RAM or PSU. You don't often hear about faulty procs except after extreme overclocking. It can happen yes, but you just don't hear much about it. I noticed you said that you reused the thermal tape on your heatsink. Not good at all.

You need to run memtest at stock speeds to see if you get any errors. Run prime95 as well. Report the errors you get. Don't forget, run the tests at stock speeds first.

Jeremy · Post by **Jeremy** » Fri May 14, 2004 7:27 pm

ok i'll give that a shot. note that the thermal tape is in the garbage now, AS5 was in about 2 days ago.

Jeremy · Post by **Jeremy** » Fri May 14, 2004 8:44 pm

holy crap...
did a brief run...30mins. at 1.8/266/33/66 cpu/memory/pci/agp no errors.
at 2.3/254/32/64...a million errors keep rapidly appearing.

so something is up with the ram. any suggestions? the fact that it's running at 254 confuses me as to why it doesn't run.

Cypher · Post by **Cypher** » Fri May 14, 2004 9:10 pm

Test the Ram to eliminate that.
http://www.memtest.org/ *Dos based ran off a floppy.

http://hcidesign.com/memtest/ *Windows based.

If the reg hives and files are corrupt your OS may be hosed. I'd reformat that without a second thought.

Jeremy · Post by **Jeremy** » Fri May 14, 2004 9:14 pm

actually that *was* using memtest86. i just let it run for 30mins to get a basic idea; nothing happened at stock (like 6 pass, 0 error, looked fine to me), then once i oc'd the errors piled up from the beginning of the test with identical CAS timings, just running at 1:1 to the fsb (eg 130:130) vs 3:4 before (100:133)

Cypher · Post by **Cypher** » Fri May 14, 2004 9:34 pm

Than that's it.
It's your Ram that's holding you back.

Is it the 2100? If so you need at least some 3200.

lansens · Post by **lansens** » Fri May 14, 2004 9:54 pm

cyPHer_138 wrote:Than that's it.
It's your Ram that's holding you back.
Is it the 2100? If so you need at least some 3200.

Jeremy · Post by **Jeremy** » Fri May 14, 2004 9:55 pm

before i plunk down $$$ for memory which has become insanely expensive as of late, could you possibly explain why the memory isn't functioning with the new frequency ratio? if you set the bus to 130MHz (OC situation), the ratio is 1:1, then the memory runs at 260MHz, or pc2100 spec. and when running normally with a 100mhz bus, the ratio is 3:4, which results in 266MHz, also in spec. that seems a little weird, although the chipset officially does not support a 533mhz bus.

but yeah, seeing as a 512MB stick would cost a small fortune, i guess i'll have to live with a stock 1.8a.