Prime95 Locks *Solved*

Vjiper

Junior Member
Mar 10, 2003
16
0
0
I have a Celeron Tualatin 1.3 GHz. It will run Prime95 mostly stable (crashing once every few days) at 13x131 @ 1.825V (1.74V actual). However, when I increase the FSB to 133MHz Prime95 locks up within a matter of seconds. 132MHz also locks right away. At lower FSB and voltage settings, Prime95 would sometimes cause an illegal sumout error depending on my settings. I could fix this by increasing the voltage and everything would run just fine.

I think this is very strange since last night I was able to encode a .mpeg2 cartoon without any problems. I was encoding with TMPGenc and frameserving with VirtualDub. I have also been able to run games without any problems. I ran the Sandra burn-in test thinking that my overclock might be the problem but that failed to lock-up my computer.

I am running generic PC133 at the most agressive timings my board will allow. When I back off on the timings, I still lock-up in Prime95 at 132 and 133 FSB.

I think I might be able to fix this with a higher voltage but 1.825V is as high as the BIOS will allow.

Any thoughts would be appreciated.

 

RalfHutter

Diamond Member
Dec 29, 2000
3,202
0
76
First - You are running too much Vcore. You will seriously shorten the life of that poor CPU!!!

Prime95 is erroring because your OC isn't stable. Very few 1.3 Tualerons will run that fast.

If I was you, I'd back off your Vcore and FSB until you find where you can run Prime95 stable with about 1.625V (actual) and leave it there.
 

Vjiper

Junior Member
Mar 10, 2003
16
0
0
That's not what I wanted to hear.

If I continue to run at 1.825V (1.76V actual) will my processor die within the next few months or will it take a few years for the electrostatic migration to kill it?

How much voltage over stock is considered too much for most processors?

I think my 2 sticks of RAM might be the problem. I'm going to try running one stick and see what happens.
 

Vjiper

Junior Member
Mar 10, 2003
16
0
0
I played around with my RAM and found some interesting information. 2x128 doesn't run Prime95 at 133MHz. 1x128 runs Prime95 at 133MHZ.

I ran all my tests at 13x133=1733. The RAM timings were CAS2. CAS3 was unable to make Prime95 run stable with 2x128MB at 133MHz FSB. I tried to run the Prime95 benchmark successfully three times in a row. If it failed, I upped the voltage and tried again.

128MB stick of K-Byte PC133 SDRAM
Voltage - Actual - Result
1.525 - 1.47 - failed to load windows
1.550 - 1.50 - failed to load windows
1.575 - 1.53 - failed to load windows
1.600 - 1.56 - failed to load windows
1.625 - 1.58 - locked running Prime95 benchmark
1.650 - 1.61 - locked running Prime95 benchmark
1.675 - 1.63 - success (most likely needs to be increased to 1.700V)
1.700 - 1.64 - did not test
1.725 - 1.68 - did not test
1.750 - 1.69 - success
1.775 - 1.74 - success
1.800 - 1.74 - success
1.825 - 1.79 - success

128MB stick of unknown PC133 SDRAM
Voltage - Actual - Result
1.600 - 1.58 - failed to load windows
1.625 - 1.58 - locked running Prime95 benchmark
1.650 - 1.61 - locked running Prime95 benchmark
1.675 - 1.63 - initial success, then rebooted while replying to this thread
1.700 - 1.66 - runs Prime95 and hasn't rebooted

2x128MB in slots 1 and 3
Voltage - Actual - Result
1.775 - unknown - posts
1.800 - unknown - fails to post
1.825 - 1.76 - locked running Prime95 benchmark

I will try 2x128MB in slots 1 and 2 but I doubt it will make any difference (I think I have tried this before).

Do you think one stick of 256MB would work? I think it will but I would like to hear some thoughts before I go and buy a stick.
 

RalfHutter

Diamond Member
Dec 29, 2000
3,202
0
76
If your PC133 RAM (even no-name brand or multiple sticks) won't even run at 133Mhz than it's either the worst RAM in the world and should be RMA'd or it's not the RAM, it's the CPU. Like I said before, getting 133Mhz out of a 1.3Ghz Celeron is not common, most won't do it. I'd still be suspicious of the chip.

Try this another way. Set your RAM speed at 100Mhz in the BIOS and try OCing to 133FSB again (with the memory speed locked at 100Mhz). If you're still having poor results until you feed the poor CPU a bunch of Vcore than your RAM's not your problem, it's your CPU. If you are able to get your CPU to run 133FSB fairly easily than it is your RAM and you should go buy a decent stick, preferably a 256MB stick.
 

Vjiper

Junior Member
Mar 10, 2003
16
0
0
I now have the chip running at 13x133. 1.700V wasn't enough so I increased the voltage to 1.725V and I ran Prime95 overnight without any problems. That's about .25V above stock voltage but probably closer to .205V over stock. I don't think this is too high.

I tried running both sticks at 100 while the cpu was at 133 but didn't have any luck. I'll give it a try again just to make sure.

If that doesn't work, I'll pick up a stick of 256 and see if that fixes my problems.

Thanks for all of your help.
 

Vjiper

Junior Member
Mar 10, 2003
16
0
0
I found the problem. HEAT. Or more likely, too much of it. At least I found the problem. Now to get some cardboard and have fun with ducts.