Sign in with
Sign up | Sign in
Your question

Need help with e8400 Stability

Last response: in Overclocking
Share
June 20, 2012 6:04:52 PM

Hey everyone,

I've been trying to get this overclock stable for a very long time now, longer than even I myself would like to say.

While my OC seems to run perfectly in games with a significant and noticeable performance boost, Either one or the other core will seem to fail Prime95 blend within the 4-6 hour mark. The CPU Passes Intel Burn Test, Small FFTs, and OCCT CPU. I've tried a wide array of vcore settings and it will pass, from Auto, to 1.2 to 1.25. However, it will just not pass blend.

FATAL ERROR: Rounding was 0.5, expected less than 0.4
or
FATAL ERROR: Rounding was 0.403941023, expected less than 0.4 (Some other large decimal)

are the errors I get, and it's always the same exact one. I am not sure if it occurs on the same exact test, however.

I'm running my RAM which is rated @ 1066 at 800 mhz stock with a 400 FSB... So I don't understand how this could potentially cause it to screw up the OC.

I've had a thread on overclock.net but they weren't able to crack the issue it seems, following the countless suggestions, see it if you need an extra reference.

http://www.overclock.net/t/1205374/e8400-oc-issues-unst... f

Here are my current settings:


CPU: Intel Core2 Duo E8400
Motherboard: ASUS P5Q PRO
RAM: Patriot Viper 4GB (2 x 2GB) PVS24G8500ELKR2

FSB: 400mhz
CPU: Multiplier locked to 9.0, SpeedStep etc Disabled
CPU VCore: Auto, 1.2, and 1.25 make no difference (1.25 for the last test)
Load Line Calibration: Disabled (I've toggled between Disabled and Auto before, makes no difference)
DRAM Voltage: 1.96
DDR Timings: 5-5-5-12, and I've also tried 5-5-5-15, no difference.
NB Voltage: 1.26 (Auto causes same error)
FSB Termination Voltage: 1.26 (Auto also still causes the error)

Thank you for taking your time to assist me with my issue. :) 


Edit: An update, I figured I should also add that when a failure occurs, only one core will fail, and the test will continue to run. I've never had BOTH cores on the cpu fail a test. So, my guess is the northbridge would be the culprit at this point?

Funny thing is, I ran a blend last night. When the RAM was OC'd it took 2 hours 33 minutes for one core to fail, when I reverted back to stock (Thereby 1:1 Ratio), One core failed after 33 minutes, the other core contiuned chugging along at 7+ hours...

More about : e8400 stability

a c 172 à CPUs
a c 197 K Overclocking
June 22, 2012 1:44:35 PM

The blend and large fft's test both stress the memory subsystem. Because the small fft's test passes, your problem most likely lies with your memory settings.

Go back to stock settings. Do what you need to set the FSB:RAM ratio to 1:1. I am unfamiliar with the Asus BIOS, so I cannot tell you exactly what to do. Then run the blend tests at stock settings to make sure you do not have anything wrong at stock freqs.

Running at stock freqs does not mean you are running at 1:1. For example, at stock freqs and Auto mem settings, I thing CPU-Z should indicate a 3:4 FSB:MEM ratio. Use CPU-Z to check this.

You will probably need to muck about in the BIOS.

The northbridge feeds both cores, so if one core is working, the NB is OK.
m
0
l
June 22, 2012 5:56:06 PM

I've been running at 1:1, which is stock when the FSB is at 400. RAM is set to 800 Mhz, FSB is at 400, and still get errors.

However, when running the RAM at 800 Mhz with default FSB of 333, which everything is at stock, I don't get any errors at all ever.

Also passes memtest while at all stock settings with no issues.

I've also tried fiddling around with latency settings when running at 800 Mhz, and increasing the voltage up to 2.1, has never seemed to make things different.
m
0
l
!