Does this OC look ok?

dwighet

Lads this is my first attempt at overclocking my new rig..
I5 760
Gigabyte GA-P55A-UD4
corsair XMS3 1600 cl7
noctua nh-d14
I would apreciate if you guys in the know could take a look at a my core temp and cpu-z stats and let me know if im heading in the right direction...

Coretemp was done with prime 95 running for 2 hrs...Should it be run for longer do you think?
core temp.jpg

memory.jpg

cpu.jpg

Solitaire

For a major I'm-Finished stability run you should use Prime95 on all cores for 8+ hours and/or use an even more hardcore program, keeping more of an eye on heat than stability really.

What RAM is that btw? Hope you don't have it much past 1.65V!

dwighet

The ram is corsair xms3 1600 cl7
volted @ 1.66..
What would the max temps I should be looking for when I get a chanceto sit and whatch for 8 hrs....??
Sounds like a good excuse for a 8hr booze:D over the weekend...

Solitaire

Hmm... as low as possible, but realistically the further below 70°C the better! Above 80°C you should be re-thinking your OC and in particular your overvolt

PogMoThoin

OCCT > Prime95

Solitaire

OCCT isn't normal usage though, IIRC its more like LinPack, which is for testing to server-grade stability standards. You can't work out the heat level from it either as those "supertorture" tests are designed to circumvent TDP and pull 120%+ normal power levels... It could wind up artificially forcing speed reduction and voltage increases to maintain stability during that one test.

dwighet

Solitaire wrote: »

Hmm... as low as possible, but realistically the further below 70°C the better! Above 80°C you should be re-thinking your OC and in particular your overvolt

You mention overvolt...do you mean the ram? I went1.66 for the ram as corsair recommend 1.65 and the bios will go 1.64/1.66...should I take it down a tad do you think?...sorry for all the ??????`s but this craic is new to me..

PogMoThoin

If You're stable for 2 hours of OCCT, You've found Your 24/7 overclock. No need to test for 8 hours, it will push Your pc to the absolute limit.

dwighet

PogMoThoin wrote: »

If You're stable for 2 hours of OCCT, You've found Your 24/7 overclock. No need to test for 8 hours, it will push Your pc to the absolute limit.

sounds good...I will run it now on large data set and priority high for 2 hours and see what happens...

I will post the results for your post mortem;)

Thanks lads..

dwighet

well my 2 hrs is up on OCCT and here are the results...Im pleased..
What are your thoughts? is it a keeper?

cpu 1occt.jpg

cpu2 occt.png

cpu3 occt.jpg

occt temps.jpg

The temps off coretemp seem a small bit higher than occt

dwighet

Solitaire wrote: »

OCCT isn't normal usage though, IIRC its more like LinPack, which is for testing to server-grade stability standards. You can't work out the heat level from it either as those "supertorture" tests are designed to circumvent TDP and pull 120%+ normal power levels... It could wind up artificially forcing speed reduction and voltage increases to maintain stability during that one test.

Thats all gobbledeegook to me:eek:

will give prime a good sitting over the weekend when time is on my side


cheers

PogMoThoin

I'd be very happy with that

dwighet

PogMoThoin wrote: »

I'd be very happy with that

My mrs thinks im going weird looking at the monitor watching numbers and smiling...
They just dont understand do they...;)

Hank Scorpio

What cpu voltages are you using in the bios?

Looks like a great overclock btw

dwighet

nuxxx wrote: »

What cpu voltages are you using in the bios?

Looks like a great overclock btw

1.1937v in the bios....
its working well.... temps are good...
no probs so far and I have been throwing everything at it to see if its going to **** up on me....all good.

nvus

dwighet, when you can run Prim95 & Furmark at the same time for more than an hour you know you've made a champ PC


P.S. when do we get te see pics of your new monster?:cool:

dwighet

nvus wrote: »

dwighet, when you can run Prim95 & Furmark at the same time for more than an hour you know you've made a champ PC


P.S. when do we get te see pics of your new monster?:cool:

Prime 95 and furmark at the same time...cool..Will give that a go after work..

I will post some pics as soon as my new fans and cable extentions land from abroad...the guts of it is a bit messy at the moment:o

game4it70

Thats very low Vcore for 4ghz :eek: :cool:

What are your temps like under load?

What have you used to test for stability?
If its stable at the vcore you could have a really good clocker.

PogMoThoin

game4it70 wrote: »

Thats very low Vcore for 4ghz :eek: :cool:

What are your temps like under load?

What have you used to test for stability?
If its stable at the vcore you could have a really good clocker.

Read page 1, He's stable alright. He's got a golden chip.......like my e0 that does 3.8Ghz undervolted

Solitaire

From the heat my i5 is jacking out at stock I suspect I haven't been so lucky...

dwighet

nvus wrote: »

dwighet, when you can run Prim95 & Furmark at the same time for more than an hour you know you've made a champ PC


P.S. when do we get te see pics of your new monster?:cool:

Just did the prime95 and furmark for a hour together...no worries there.
Temps got up to 68 but thats ok I think as the coal fire is on...

I could jump the cooler and case fans up to 1300rpm to drop a few degrees but i like things quiet..

Hank Scorpio

Solitaire wrote: »

From the heat my i5 is jacking out at stock I suspect I haven't been so lucky...

What stock temps are you getting?

Solitaire

Anywhere between 65°C and 90°C depending on ambient, which varies between 10°C and 25°C depending on heating. Stock speed, stock volts, stock cooler. And this is an open-air testbed!

Every time I catch it surpassing ~80°C I drop it into hibernation to give stuff time to cool off, otherwise its running 24/7.

game4it70

dwighet wrote: »

Coretemp was done with prime 95 running for 2 hrs...Should it be run for longer do you think?

PogMoThoin wrote: »

Read page 1, He's stable alright. He's got a golden chip.......like my e0 that does 3.8Ghz undervolted

I did :P.
To be fair 2 hours of prime is nowhere near long enough to be sure its stable.
The more common way to test is at least 8 hrs of prime and/or 20+ runs of linX with "all" memory.

dwighet

game4it70 wrote: »

I did :P.
To be fair 2 hours of prime is nowhere near long enough to be sure its stable.
The more common way to test is at least 8 hrs of prime and/or 20+ runs of linX with "all" memory.

I did the 8 hrs of prime last weekend....Was all good.
Havnt`t done the linX....will give it go now for the craic....

game4it70

dwighet wrote: »

I did the 8 hrs of prime last weekend....Was all good.
Havnt`t done the linX....will give it go now for the craic....

Nice one so very good chance your stable.
Do keep an eye on temps when running linX as its more aggressive than prime and will add a few C more to your cores.

dwighet

game4it70 wrote: »

Nice one so very good chance your stable.
Do keep an eye on temps when running linX as its more aggressive than prime and will add a few C more to your cores.

There you go


Temps did get to 72c briefly but its very warm here in the sitting room

game4it70

How much ram are you running?

When using linX you need to have memory set to "all" so it stress's it fully.

Suberb Vcore imo and still decent temps.

dwighet

game4it70 wrote: »

How much ram are you running?

When using linX you need to have memory set to "all" so it stress's it fully.

Suberb Vcore imo and still decent temps.

oops...forgot to select all....will do it again

The ram is 4gig

game4it70

dwighet wrote: »

I selected "all"

I suppose its just down to what process's/programs are running in the background that are taking up the resourses.I thought it would have used a bit more ram but shouldn't make any difference in your case.

Have you tried using uneven multipliers?
Its strange but i5/i7 most of the time clock better with uneven multi.

191 base clock by 21 multi is a very common to achieve 4ghz although it will make you ram run slower or alot faster depending on what ram divider you choose.
If you have time it would be worth trying it as you may need less Vcore for it than 20x200.

I would certainly try for 4.2 (21x200) with that cpu as you shouldn't need to much more voltage looking at how little you need for 4ghz.



Lol i was slow posting so just saw your edit.
Let me know how you get on.

dwighet

game4it70 wrote: »

I suppose its just down to what process's/programs are running in the background that are taking up the resourses.I thought it would have used a bit more ram but shouldn't make any difference in your case.

Have you tried using uneven multipliers?
Its strange but i5/i7 most of the time clock better with uneven multi.

191 base clock by 21 multi is a very common to achieve 4ghz although it will make you ram run slower or alot faster depending on what ram divider you choose.
If you have time it would be worth trying it as you may need less Vcore for it than 20x200.

I would certainly try for 4.2 (21x200) with that cpu as you shouldn't need to much more voltage looking at how little you need for 4ghz.



Lol i was slow posting so just saw your edit.
Let me know how you get on.

I tried getting 4.2 a few days ago and had no worries @ 1.28v but the temps were getting warm so thought Id wait until my extra fans arrive(im only running 2 intake and 1 exhaust at the moment) also got some arctic silver thermal paste on its way.......

Heres the linX test done with all memory

game4it70

You see the way your Gflops have dropped during test,where you doing something like opening a program or browsing at those times?
If so thats why they dropped.
After the first test the flops should be steady and within about .7ish of each other.If they fluctuate to much it can mean slight instabilty.

I see top temp of 64c on core 2.Is that core always approx 10c lower than the top temp when cpu is usage is loaded?
Its unusual to see that unless you have a stuck sensor of badly applied tim.
The average i5 will only have about 5c difference in the cores.

The top core of 74c is very good and real world day to day usage will never get it near as hot as linX.

I actually think the noctua nh1 paste that you got with the cooler is very good and rate it as one of the best i've used.
And boy oh boy have i tried a few and so many remouts in a quest to lower temps.

Congratz btw on a nice OC.

PogMoThoin

Arma 2 must run lovely on that at 4Ghz
The only game that really needs a quad.

dwighet

game4it70 wrote: »

You see the way your Gflops have dropped during test,where you doing something like opening a program or browsing at those times?
If so thats why they dropped.
After the first test the flops should be steady and within about .7ish of each other.If they fluctuate to much it can mean slight instabilty.

I see top temp of 64c on core 2.Is that core always approx 10c lower than the top temp when cpu is usage is loaded?
Its unusual to see that unless you have a stuck sensor of badly applied tim.
The average i5 will only have about 5c difference in the cores.

The top core of 74c is very good and real world day to day usage will never get it near as hot as linX.

I actually think the noctua nh1 paste that you got with the cooler is very good and rate it as one of the best i've used.
And boy oh boy have i tried a few and so many remouts in a quest to lower temps.

Congratz btw on a nice OC.

I was on the net for bit while the test was on...

game4it70

dwighet wrote: »

I was on the net for bit while the test was on...

Grand that covers the drop in the Gflops.

What about the 10c difference in cores?

dwighet

game4it70 wrote: »

Grand that covers the drop in the Gflops.

What about the 10c difference in cores?[/QUOTE]

F&$k knows.....
Will run it again with everything shut down so the bulk of the ram is getting used...
see what happens then..

game4it70

dwighet wrote: »

F&$k knows.....
Will run it again with everything shut down so the bulk of the ram is getting used...
see what happens then..

If its around the same temp as the other cores while idle and it only gets that far apart when at full load then its probably just a bad sensor.

dwighet

game4it70 wrote: »

If its around the same temp as the other cores while idle and it only gets that far apart when at full load then its probably just a bad sensor.

nah theres 5c diff at idle.....but hey its not a perfect world we live in so no sleep will be lost over it.....
my games are running sweet and redtube is working well....happydays:D

P.S. I appreciate your input....
Good man and merry xmas to ya and your family...

ATB Dwighet

game4it70

dwighet wrote: »

nah theres 5c diff at idle.....but hey its not a perfect world we live in so no sleep will be lost over it.....
my games are running sweet and redtube is working well....happydays:D

P.S. I appreciate your input....
Good man and merry xmas to ya and your family...

ATB Dwighet

Same to you,enjoy.

Hank Scorpio

fwiw my i5 is similar in the difference of core temps. 2 of them are often 7 degrees lower even in idle.

wouldn't worry about it tbh

Hank Scorpio

What cpu vtt voltage are you using mate?

dwighet

nuxxx wrote: »

What cpu vtt voltage are you using mate?

QPI/VTT voltage is on auto...

game4it70

dwighet wrote: »

QPI/VTT voltage is on auto...

You should set that yourself as auto may or may not overvolt it,just depends on the mobo.
Qpi/Vtt voltage adds heat to the cpu too so lower is always much better.
It must be kept within .5v off your ram voltage to avoid damage to your cpu.

1.20v-1.25v is a good place to start @ your overclock.

dwighet

game4it70 wrote: »

I suppose its just down to what process's/programs are running in the background that are taking up the resourses.I thought it would have used a bit more ram but shouldn't make any difference in your case.

Have you tried using uneven multipliers?
Its strange but i5/i7 most of the time clock better with uneven multi.

191 base clock by 21 multi is a very common to achieve 4ghz although it will make you ram run slower or alot faster depending on what ram divider you choose.
If you have time it would be worth trying it as you may need less Vcore for it than 20x200.

I would certainly try for 4.2 (21x200) with that cpu as you shouldn't need to much more voltage looking at how little you need for 4ghz.



Lol i was slow posting so just saw your edit.
Let me know how you get on.

Well mate....got my new fans(6x noctua 120`s) and some arctic silver matrix thermal goo yesterday......Fitted the new fans.....4x intake and 3x exhuaust all using the L.A.N running the fans at 1100 rpm..... Took your advice at o`cing at 4.2 ghz....... not aproblem..... Im getting 68c on prime after 6 hrs which I ran last night...... 73c on linX on 20 runs set on all....
and vcore is set on 1.25v...... :eek: not bad for a i5 760 me thinks..think I was blessed with a very good cpu;)....