Whats a good indicator that the CPU is the cause of errors..

[imported_goku]

What is a good indicator that the CPU is the cause of errors in memtest 86 opposed to the memory? I'm getting mixed results in memtest 86 and I really can't say for sure if it's the processor or the memory that is causing the results. Basically I'm at 257MHZ/265/272mhz overclocking scenarios trying to find what is stable when and while putting a divider seemingly alleviates the issue, it also doesn't seem fair to automatically assume it's the memory.

I've got a P4 2.8C that is being overclocked. Could this be my board getting too hot and that is why I'm getting errors in memtest86? Is there a certain way it has to error for it to be memory errors and the other way is board/cpu related?

 

[CTho9305]

With consumer equipment, it would be nearly impossible to tell whether the problem is in the memory, the northbridge, or even the processor's load-store unit / bus interface.

 

[bob4432]

how does it do with prime95 with the same dividers/fsb? how does each react with extra V?

 

[imported_goku]

Prime 95 or what ever processor stressing utility that comes with the ultimate boot CD seems to let more through the cracks. I've been mostly testing with memtest 86 atm. One test I ran disappointed me since it had 15 passes before an error popped up.. Problem with the prime95 program is that if the ram is bad, it will fail to work as well.



If the memory is at fault, wouldn't there be a lot more than just 2 errors after a few passes? Before on my pre-RMA board I would get like a huge list of errors in memtest86 after like 9 minutes and now I'm getting like 1-2 errors after like 2 whole passes (30 minutes each). I'm running the memory at SPD settings with 2.8v, mushkin Redlines XP4000. 2X1GB

 

[jjzelinski]

Originally posted by: goku
Prime 95 or what ever processor stressing utility that comes with the ultimate boot CD seems to let more through the cracks. I've been mostly testing with memtest 86 atm. One test I ran disappointed me since it had 15 passes before an error popped up.. Problem with the prime95 program is that if the ram is bad, it will fail to work as well.



If the memory is at fault, wouldn't there be a lot more than just 2 errors after a few passes? Before on my pre-RMA board I would get like a huge list of errors in memtest86 after like 9 minutes and now I'm getting like 1-2 errors after like 2 whole passes (30 minutes each). I'm running the memory at SPD settings with 2.8v, mushkin Redlines XP4000. 2X1GB

Usually if it's your CPU that's flaking out on you it results in hard failues and lock-ups, not memory test errors. Also it shouldn't be that hard to isolate your equipment's weak points; crank up your FSB, drop your multiplier and use a memory divider (if your board supports these features) to see what your CPU can take.

 

[Kelnoen]

I find a lot of the time when the CPU is at fault (been through literally hundreds of bad cpu's) you'll get garbled text on the screen using the windows memory test program.

 

[bob4432]

Originally posted by: goku
Prime 95 or what ever processor stressing utility that comes with the ultimate boot CD seems to let more through the cracks. I've been mostly testing with memtest 86 atm. One test I ran disappointed me since it had 15 passes before an error popped up.. Problem with the prime95 program is that if the ram is bad, it will fail to work as well.



If the memory is at fault, wouldn't there be a lot more than just 2 errors after a few passes? Before on my pre-RMA board I would get like a huge list of errors in memtest86 after like 9 minutes and now I'm getting like 1-2 errors after like 2 whole passes (30 minutes each). I'm running the memory at SPD settings with 2.8v, mushkin Redlines XP4000. 2X1GB

i have had memory give me a couple of errors very late in testing, but there were always errors with this particular set of sticks. if i would have just run a couple of passes i would have never caught it.

i have also had it where the entire screen gets filled up with errors within a couple of minutes, but always it turned out to be a memory issue as when i replaced them i could let memtest run for days, literally.

i would say it is your memory.

 

[Hadsus]

When I bought a Pentium 3 ghz system a few years ago, it would spontaneously reboot at random times. In the bios, we turned off hyperthreading and everything worked fine. So, the CPU was switched out and the rig became stable again.....the CPU was bad.

 

[flexy]

you can ONLY "assume" its the memory if the CPU is NOT overclocked and the overclock applies ONLY to the memory.

ALso..memtest is a pretty WEAK test, its nothing more than a QUICK test basically making sure your mem is without defects.

I say that because i came alone many scenarios where i OC and memtest passes fine, even hardcore tests like 5 and 8...but cant even BOOT into XP because i am so instable.

I really would rather go with OCCT and/or prime95 and do the tests there. Basically only use memtest to make halfway sure you can boot safely...and do the real testing w/ OCCT (prime95).

ALso...the only way to exclude certain components, again, to run ONE componenent within specs, maybe even UNDERclocked and then do testing. The first OC rule is to find out the limits of each component...and THEN combine.