Efficient way to test incremental overclock?

Toggle sidebar Toggle sidebar
M

Marc999

Distinguished
Apr 19, 2010
48
0
18,530
Here I am testing each incremental Blck 10 mhz step. with IBT Max. that takes 1/2 hr. between reboots to test for highest stable Blck frequency, while holding cpu multiplier constant. Its working but is it ever labor intensive lol.

Then I have to do the same Blck incremental steps again when I test for highest stable memory frequency. Then use memtest as well.

Then I can finally attack the vcore, the best for last.

Then I can run long term 24 Prime95 blend test and 8 hr. Prime95 small fft

It's a guide I'm following which is great, yet I'm wondering if these 1/2 hr.incremental tests are overkill. Is their a quicker test to check for major stability issues?
 
Sort by date Sort by votes
saint19

saint19

Polypheme
Mar 20, 2009
15,746
26
60,540
Hi and welcome to Tom's Hardware Forum.

What are your components? Test each 10MHz is a little "extreme" I think, depending of your components you could overclock the CPU about 200MHz without touch voltage and should be stable.

Let us know the components that you have and the temps that you get on idle and load for your rig.
 
Upvote 0 Downvote
M

Marc999

Distinguished
Apr 19, 2010
48
0
18,530




I'm running an i7-860 on a gigabyte p55_ud4p, with 8gb g.skill 1600 ram cl 9.
Cm Hyper 212 Evo
Hyper-threading ON. ( I use it a lot via handbrake)
Power saving items in Bios back on.
Turbo OFF


I finally hit a road block with Bclk @ 181 @ 21 CPU multiplier = 3.8 ghz
Vtt/qpi = 1.25
Vcore = Normal + dynamic vcore @ 0.05 =1.275

Vcore @ 1.3 will bsod, or linpck error ( Intel burn test- maximum, 3 loop)
The bsod error is 0124, which could indicate Vtt increase needed but I tried that. Vtt @ 1.270. Likely need more than that.

Having said that, my temperature during stability testing via IBT or Prime95 small fft has been high @ 78-81 degrees Celsius.

I'm thinking if these stability tests can run for awhile and be stable, 3.8 GHz with hyper-threading on isn't too shabby and much better than stock 2.9 GHz.with turbo.
I presume real world use I.e. games and video compression/rendering wouldn't get as hot as these tests?


 
Upvote 0 Downvote
S

schmuckley

Distinguished
BANNED
Jul 18, 2011
809
5
19,165
3 loops of Intel Burn Test..
Hyper-Pi 8m..
UCbench with latest instruction set + custom threads
Cinebench 11.5
IBT will heat a Lynnfield up pretty good..I'd probably avoid that @ higher voltages.
Nothing heats a CPU up more than IBT..owait..no AVX..so maybe not so bad.
10 Mhz on 1156 is quite a bit..I do 10-15Mhz increments as well.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom