Prime95 - e8500: Core 1 ok, Core 2 encountering errors

After running it on blend for 2 5-hour tests, both times my CPU 1 ran the full duration with 0 errors but the second CPU would encounter an error within 30 mins to an hour. I currently have my FSB set to 1720 in the bios (~4.083 GHz) and the ram unlinked at 800 (it's 800 MHz DDR2). The Vcore voltage is set to 1.41v and everything else (VTT, Memory voltage, etc.) is set to auto. I originally had the voltage set to 1.36, but both processors would encounter errors very quickly so I upped it. Real Temp puts my average distance to TJ Max during full load at ~25 with my processor's core normally running at 72-75 celsius full load, 45-47 idle. Anyone have any input or suggestions as to why one of the cores would be grinding through Prime95 fine for 5-6 hours but the other encountering an error within the first hour?

By the way, the error encountered is: FATAL ERROR: Rounding was 0.5, expected less than 0.4. Hardware failure detected, consult stress.txt file.

Edit: Lucius's response was to a question I asked earlier, but the main post has been updated to reflect my current issue.

Edit 2: I use Windows 7 64-bit, by the way (in case the software isn't compatible or something).
10 answers Last reply
More about prime95 e8500 core core encountering errors
  1. Most of your background processes are probably running on the core that is behind. Stuff like windows processes etc will do this. It's nothing to worry about.

    Run it through Small FFT, Blend is a cyclic run through of CPU and RAM. Small FFT will stress the processor only, holding it 100% for however long you keep it running.
  2. Bump
  3. Sounds like your overclock isn't quite there yet. If you have no more room to add more voltage try bring the clock speed down a little at a time to see if the errors subside. And again i suggest you to run the Small FFT torture test as it'll only stress the processor. This will help eliminate RAM from the equation. A small bump in voltage on the fsb couldn't hurt either.
  4. I found one core would crash before the other, I also found my cores had slightly different temps. If it does this on Small FFT settings, then you need more Vcore or a slightly lower bus speed.

    Treat them as seperate CPUs, and you can only go as fast as the slowest one.

    I'm running mine at 4.37Ghz, but at 4.5Ghz I get Core 1 erroring.
    I was able to do 4Ghz with a bump in Vcore but everything else auto like you.

    You're running kinda hot though, at 4.37 @ 1.52V I don't get over 65/63C and normally at 100% load the cores are 62/60C). I'm using a Coolermaster Hyper NS520 (not the best but works well).
  5. as for temps if there off then dont worry about it tbh there not very accurate :P i have one set at 37c and one at 30c but they are actually well below 0. they are only accurate at higher temps to save from frying the cpu from heat
  6. System specs? What kind of HSF?

    Your temps indicate that you could use some better cooling.
  7. Bumping this because I'm having a similar problem with my e8400 and upon further inspection i found with speedfan that even though I have my vtt bumped to 1.30, core 1 is at 1.28 and core 2 at 1.30. core 1 is the one that keeps failing at around 3 hours in prime95 with core2 running stable.

    Here are oc stats, 3.70 (411x9) 1.30vcore 44c idle 57c load (zalman 9700)
    dfi dk p35 t2rs
    corsair xms pc-6400 1:1 822mhz

    What originally prompted me to look at vcore was that cpu-z is only displaying my vcore at 1.28 when it should be 1.30...

    I'm not sure how to approach this.
  8. More volts.
  9. Since i Up'd it to 1.30 its actually been prime stable for 3 hours so far.. i was hoping to get to 3.85 at 1.30 so i could save 1.326 for 4.0ghz
  10. Every chip is different.
Ask a new question

Read More

CPUs Core Overclocking