7600K: Keep Failing Prime 95 Test But Always Pass Intel Burn-In Test??? hua?

I am running my 7600k at 1.27vcore and 4.9 GHz. and I am able to pass the Intel Burn-In test fine. However, I have problems with both Prime95 and Orthos. The computer does not lock up or crash, rather two cores stop operating almost as soon as I start the test when I run Prime95. I get the message"FATAL ERROR: Rounding was 0.5, expected less than 0.4. Hardware failure detected, consult stress.txt file".

So I thought okay, I will bump the vcore. Well I bumped it all the vcore all the way up to 1.35v which is way higher than what most people are running the 7600k at 4.9GHz, and still I have the same problem. In order to not get any of the cores to crash I have to bump it way to 1.4vcore which is ridiculous. People are running 5.1 GHz. at that vcore. So basically, 1.27vcore is stable at 4.9 GHz on Intel Burn-In but 1.35v wont pass Prime95 at the same clock speed... That makes no sense. Again, not once has the computer ever locked up or crashed. Temps are fine, all 80C or less.

what is the vcore actually showing in HWmonitor or HWinfo64? Is it showing your set value, or something less?

Doesn't surprise me. P95 is very demanding on the CPU. On my Skylake CPUs which run similar work 24/7 I run them at 4.2 GHz 1.25v, Prime95 stable. Quite simply the PC really is Prime95 unstable at high clocks and low voltages, but if you never run anything like Prime95 workloads, the question can be asked if you need to be Prime95 stable. I do run P95 like work, so I have to use a much lower clock than others might, or alternatively jack up the voltage with all the heat problems that come with it. Lower clock is preferable to me.

R_Pierce said:

what is the vcore actually showing in HWmonitor or HWinfo64?

Click to expand...

The values I stated. Those came from CPU-Z which reads the same as HWmonitor. I dont know if it is showing my set value or something less because I overclock using offset voltage so the numbers in the BIOS dont tell me what the set voltage is, only what the offset is, which at the moment is set to -0.040v which is about 1.27vcore. I just ran RealBench at 4.9ghz at 1.296v and it was stable for two hours.


On a side note, I notice that offset voltage is far less stable than manual override. With manual, the voltage doesent really change under load. However, with offset, when my benching program has the cores maxed out, my voltage keeps bouncing around from 1.26v to 1.296v which is making IntelBurnTest fail my computer. Anyone know why the voltage keeps fluctuating under load instead of remaining steady? With my 3570k I used offset voltage to overclock and it remained quite steady under load.

What board do you have? How about updating your signature to show the proper hardware.

What version of P95 are you using? Beginning with version 27.7 AVX1 was included in the Prime95 instruction testing set. Later versions of Prime95 beginning with 28.5 have AVX2 instructions which is even more demanding. Your IBT does not have AVX.

And it appears you did not win the silicon lottery on top of that.

What motherboard are you using? Your Sig reflects older equipment.

SPL Tech said:

The values I stated. Those came from CPU-Z which reads the same as HWmonitor. I dont know if it is showing my set value or something less because I overclock using offset voltage so the numbers in the BIOS dont tell me what the set voltage is, only what the offset is, which at the moment is set to -0.040v which is about 1.27vcore. I just ran RealBench at 4.9ghz at 1.296v and it was stable for two hours.


On a side note, I notice that offset voltage is far less stable than manual override. With manual, the voltage doesent really change under load. However, with offset, when my benching program has the cores maxed out, my voltage keeps bouncing around from 1.26v to 1.296v which is making IntelBurnTest fail my computer. Anyone know why the voltage keeps fluctuating under load instead of remaining steady? With my 3570k I used offset voltage to overclock and it remained quite steady under load.

Click to expand...

Offset voltage fluctuates with the varying load, that is normal. If your having a problem still with Prime95 up the voltage a little more.

With Kaby Lake and Z270 some motherboard have AVX offset. It lowers the multiplayer when AVX is used. This allows you to keep your 4.9ghz OC for most things but when AVX is detected it will drop your multi by 1-3x for stability.

What is your LLC set to? I tested my motherboards 5 levels and found 3 of them will have a lower vcore when running P95 than ITB, which causes P95 to fail when ITB passes. The higher settings will keep the vcore the same across all stress tests.

SPL Tech, also, you never mentioned which difficulty level you employed in running IBT. There is everything from Standard to Maximum. And how many iterations did you set it to when stress testing? Did you leave it on the default 10? IBT can be configured so as to be relatively easy to pass or relatively hard.

Sorry, I was away for awhile. I updated my signature to the computer I am using. I was running IBT at both the standard and high level. It passes both. My LLC is set to auto. My mobo does not have level 1 - 5, rather it has a specific voltage number you can add (e.g. add 0.012vcore). I just keep it at auto. I am using P95 version 28.9 build 2.

I lowered my overclock to 4.8 and 1.28vcore offset. It still passes IBT add day everyday, but P95 one of the cores stops within seconds of starting it. It seems really unlikely that I should be needed upwards of 1.3vcore just for 4.8 as everyone else is running far lower voltages at that clock speed, even without deliding their CPU (which I did).

SPL Tech said:

Sorry, I was away for awhile. I updated my signature to the computer I am using. I was running IBT at both the standard and high level. It passes both. My LLC is set to auto. My mobo does not have level 1 - 5, rather it has a specific voltage number you can add (e.g. add 0.012vcore). I just keep it at auto. I am using P95 version 28.9 build 2.

I lowered my overclock to 4.8 and 1.28vcore offset. It still passes IBT add day everyday, but P95 one of the cores stops within seconds of starting it. It seems really unlikely that I should be needed upwards of 1.3vcore just for 4.8 as everyone else is running far lower voltages at that clock speed, even without deliding their CPU (which I did).

Click to expand...

I would not say everyone because since the i7 7700k was released I've seen a lot of folks complain just like you about voltage and overclocking past 4.8GHz.

wingman99 said:

I would not say everyone because since the i7 7700k was released I've seen a lot of folks complain just like you about voltage and overclocking past 4.8GHz.

Click to expand...

The 7700 needs more vcore than the 7600 does. Two different CPUs.

honestly, 1.32 vcore is nothing.