3930k - Best Way of Stability Testing ?

Flisk

New member
[font=tahoma,geneva,sans-serif]Hi to all here on OC3D,

I wanted to ask u guys or Tom if u get a minute to help, whats best way to test stability on X79 platform, I mean rock solid stability.

Whats actual procedure u use while overclocking ?

For example, I am doing something like :

CPU stability testing :

1. Setting some stuff in bios

2. Runing prime small ffts for 15min

3. Runing linx 4GB for 15min

4. Runing prime custom blend with 14GB of memory (I've got 16GB in total) for 2 hours

- I am not sure if after passing this I can say I got very stable system

Memory Stability testing :

1. I have no idea here, I tried memtest but it passes 3 times and I still get errors in VMware so its not good way. Now I am trying also blend with 14GB for 2-3 Hours but still soemtimes after like 2.5h I get some error so do I have to run it for like 10-20 hours every time I change something ? I guess there have to be some better way to test ram stability cos this way it would take me month to overclock it
biggrin.png


Now I am running 4.5Ghz @ "Extreme" LLC 1.405v in BIOS/1.398v in CPU-Z idle and 1.368v in Load - Prime Small FFS , I am quite sad because I was expecting something like 4.7-4.8Ghz and also volts sounds high for 4.5Ghz

- Borad : X79-UD7 , Cooling NH-D14, 140mm 110cfm Vipers, TIM AC MX-4

IMG_1116.jpg


18fa3644_45GhzSmallFFT-15min.png


Any help or tips would be great, thx guys.[/font]
 
I run blend on prime for 24 hours. if it passes is when i would say it is fully stable!

I also run small fft for 12 hours

so in total 36 hours

Thats a bit extra but it makes me satisfied!

I h8 bsods!
 
well first of all be realistic, dont push it too far

if it posts then your geting there,

boot = sign of stable oc

atleast 12 hours prime 95 blend

then just rinse and repeat with higher clocks, use steady intervals to find the sweet spot
 
I would strongly recommend Intel Burn Test, which in my opinion, is way faster at sorting out unstable overclocks.

Also, overclock the processor, get it stable, THEN try to overclock the ram.(if u want to, I run mine at stock)

Apart from that, u may have been a tad unlucky in the lottery with the chip, but remember, at 4.5 it's still insanely fast.

Personally I have a rule of thumb that I should never have to go past 1.38v, and I managed that even running offset-mode. 4.5ghz.

Good luck on the clocking

/D

EDIT: Intel Burn Test is somewhat hot on the processor. remember to watch the temperatures.
 
Personally I had the processor draw over than 20W more under IBT than Prime.

It also helps people realize how crap stock fans are.. my old processor, i7-920 stock, hit 93 degrees after 5seconds of IBT...
 
i have mine at 4.6ghz at 1.336volts according to cpuz stable after 2 hours of intel burn test (this is rough voltages i havnt fine tuned it yet).

I have been using alot of gaming and video converting software etc and no crashes yet in 4 days.

i may try and lower the volts or push for 4.7 next week.

2la3u4l.png


and cinebench result

2uxu7aq.png
 
Thx to all of u guys for input .

I got serious today and get myself multimeter
smile.png
Changed LLC to 90% which looks as sweetspot and I'am trying to get 4.6Ghz stable .

IMG_1159.jpg


Also I just found out that CPU-Z volts is kinda useless indicator, for example now it says 1.368v in Load and real volts are 1.424 , it's not little off ... its damn far far away from reality
biggrin.png


4.6Ghz @ 1.424 real volts looks promising, max temps are around 69 which sounds fine to me.
 
What do you mean by real volts? sorry for the noob question.

I have my bios set to 1.34 and digi + power control set to high. There is very high and extreme over that so its just over normal and I don't think I need more than 1.3xx volts for a stable 4.6
 
By real volts I mean volts measured by "multimeter" or how is it called in English its that thing on the left side of picture in my previous post.

And multimeter is connected directly on motherboard vCore contacts.
 
Back
Top