dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
7694
share rss forum feed

me1212

join:2008-11-20
Pleasant Hill, MO

[OC] safest OC for a 2500k, on a Asrock p67 extreme4 gen 3?

I got the build done yesterday(my first custom build) Im using an antec kuhler 620 for the cpu cooler, cause it was $20 less then the h60, but I've read some stuff about people having troubles with high overclocks. I'm planning to do a 4.5ghz is that within safe ranges with my cooler? Also havethe haf 932 case if that info helps.

thanks.


Nanoprobe
Looking for cures in memory of Mom
Premium
join:2003-05-11
Crab Nebula
kudos:2
A 4.5GHz. overclock should be no problem with that cooler.


bobbagels
Just Another Scorpion Mechwarrior
Premium
join:2000-11-15
Matawan, NJ
kudos:2

1 edit
reply to me1212
yea, 4.5 should be a breeze.
OC it and bench it.
»majorgeeks.com/Prime95_d4363.html

Check your temps with Real temp and voltage with CPUZ

»www.cpuid.com/softwares/cpu-z.html
»www.techpowerup.com/realtemp/

if you want to do some comparisons or need a good guide lokk here.

»clunk.org.uk/

PS- when I overclocked mine I didnt go right to 4.5 I went up gradually (I followed Clunks guide)
Then I learned about offset voltages and am now running at +0.60 VCore which gives me 1.02 V at idle and 1.30 V at full load.(on air) So the benefits with that are obviously less stress on CPU and less heat all around

Hope this helps
--
Last night I saw a naked cowgirl, she was floatin across the ceilin


matt5z

@spcsdns.net
reply to me1212
You are running outside spec, so safe is relative, but what you want to watch is temps and voltage. We can all give a ghz number, but if the cooler is seated wrong or something, game over as far as safe.

So stay under IMHO 85C on the cores, myself 80C.

I would keep voltage under 1.45 myself at the most.

me1212

join:2008-11-20
Pleasant Hill, MO
reply to me1212
Ok, thanks for the info and links guys! IS there any guide made specifically for the 2500k, or 2600k since they are so close in how they work, for ocing? I've googled some and some say for 4.5 no more than 1.2V is needed, some say use no less than 1.3. some say leave the auto voltage on some say off. Microcenter has a nice over clocking warranty, so I hear, but i'd rather not have to use it and just do it right the first time.

matt5

join:2001-10-06
Lagrangeville, NY
No one setting works for all chips unless it's WAY high, as every chip is different.

»www.overclock.net has tons of info

2500k and 2600k overclock the same way...


bobbagels
Just Another Scorpion Mechwarrior
Premium
join:2000-11-15
Matawan, NJ
kudos:2
reply to me1212
Have a look at the guide I posted above
clunk.org.uk/
Its for 2500K & 2600K
Its a good guide that has helped many
But like matt5 says, there is no one setting for all.

Also the link he posted is a good site for overclocking as well.
--
Last night I saw a naked cowgirl, she was floatin across the ceilin

me1212

join:2008-11-20
Pleasant Hill, MO

1 recommendation

reply to me1212
Click for full size
Ok, I gave it a shot. Not sure if I did it right, ran prime95 for 20 minutes. does it look ok?


matt5z

@spcsdns.net
20min of prime is nothing, you also need to make sure you did round off checking, I would for a final run, run at least 12hours of prime.


bobbagels
Just Another Scorpion Mechwarrior
Premium
join:2000-11-15
Matawan, NJ
kudos:2
reply to me1212
Well temps look good and so does voltage But like Matt5Z says prime it for a little longer. I donno about 12 Hrs, that might be necessary and it might not. I have read in a number of places if it passes for 2 hours its all good. But i guess a 12 Hr run wouldnt be bad either

PS- after its stable and your satisfied look into offset voltages. Trust me, its the way to go. Why run it at 1.3V all the time if you dont have to?
--
Last night I saw a naked cowgirl, she was floatin across the ceilin


DKS
Damn Kidney Stones
Premium,ExMod 2002
join:2001-03-22
Owen Sound, ON
kudos:2
reply to me1212
said by me1212:

Ok, thanks for the info and links guys! IS there any guide made specifically for the 2500k, or 2600k since they are so close in how they work, for ocing? I've googled some and some say for 4.5 no more than 1.2V is needed, some say use no less than 1.3. some say leave the auto voltage on some say off. Microcenter has a nice over clocking warranty, so I hear, but i'd rather not have to use it and just do it right the first time.

I use autovoltage on my 2600K and run 4945 GHz without a problem.
--
Need-based health care not greed-based health care.

me1212

join:2008-11-20
Pleasant Hill, MO

2 edits
reply to bobbagels
Click for full size
Just after I posted that I did some stuff with offsets voltages. its usually around 1.256-1.264 under a load, and even gaming, unless its minecraft in which case it can go to 1.3 but I attribute that partially to the java VM

edit: here is a pic of what prime 95 does to it voltage wise. no idea, other than the java vm, why minecraft makes it do more. what is the stock voltage anyway? never bothered to check myself before oc'ing, but want to know.

Chrno

join:2003-12-11
reply to DKS
1.53vcore for a 32nm processor is extremely high. Intel states of maximum of 1.5vcore for their 40nm processors. However, they used the same number for their 32nm offerings which leads me to believe the actual safe vcore for the 32nm SKUs should be much lower than 1.5v If you run your processor at this type of voltage over a sustain period of time be prepare for degradation or damage to the processor.

matt5

join:2001-10-06
Lagrangeville, NY
It's the same if you run the CPU to the thermal limit, but lol he does not believe it.

OP stock voltage varies by the chip, so we can not tell you.


DKS
Damn Kidney Stones
Premium,ExMod 2002
join:2001-03-22
Owen Sound, ON
kudos:2
said by matt5:

It's the same if you run the CPU to the thermal limit, but lol he does not believe it.

OP stock voltage varies by the chip, so we can not tell you.

Nope. Not proved yet. And no one has presented evidence, just "chicken little" fear.
--
Need-based health care not greed-based health care.

matt5

join:2001-10-06
Lagrangeville, NY
The evidence is in not (as it would seem) so common sense and...

I gave you charts and everything... also less common sense, as temp goes up, resistance goes up too, you need more pressure (voltage) to overcome the increased resistance, and that increases electromigration, that decreases the time before the CPU fails.

for more info
»en.wikipedia.org/wiki/Electromigration

To use Black's equation, the component is put through high temperature operating life (HTOL) testing. The component's expected life span under real conditions is extrapolated from data gathered during the testing

So to speed up killing the chip, they run it hot. What more could you need.

Now I am sure you will say... but intel but the throttle at a safe temp blah blah.

Intel also speced the chip to run what 3.7ghz @ like 1.24 volts? One can assume the throttle is nothing more than a fail safe for a fan failure (after all it would suck for a server to go down, or a chip to fry cause a $5 fan died). One time throttling won't kill the chip, extended runs at full temp will hurt.

Oh ya your car likely has a fuel shut off over the redline, I dare you to drive it at that RPM non stop bumping the fuel shut off.

Note I am not responsible for any damage you do by running the car at red line.

Run it as you like, but stop telling other people that super high volts are okay. hell head over to »www.overclock.net and ask them about your voltage and temps 90% of those guys will say too hot... and that whole site is about overclocking and pushing.

OP your choice, 99% of people unless doing a suicide run (yep that is what it's called) will not over volt that much and over temp that much. At least not for 24/7 usage.


DKS
Damn Kidney Stones
Premium,ExMod 2002
join:2001-03-22
Owen Sound, ON
kudos:2
said by matt5:

So to speed up killing the chip, they run it hot. What more could you need.

Now I am sure you will say... but intel but the throttle at a safe temp blah blah.

I don't have to Intel does it.

Intel also speced the chip to run what 3.7ghz @ like 1.24 volts? One can assume the throttle is nothing more than a fail safe for a fan failure (after all it would suck for a server to go down, or a chip to fry cause a $5 fan died). One time throttling won't kill the chip, extended runs at full temp will hurt.

Assume nothing. Evidence of such an occurrence?

Oh ya your car likely has a fuel shut off over the redline, I dare you to drive it at that RPM non stop bumping the fuel shut off.

Not necessarily. A rev limiter is not a throttle cut off. As I have said elsewhere, your automotive comparison fails. That's one of the reasons Intel has SpeedStep. It throttles back based on usage.

Run it as you like, but stop telling other people that super high volts are okay. hell head over to »www.overclock.net and ask them about your voltage and temps 90% of those guys will say too hot... and that whole site is about overclocking and pushing.

I can run all day at 4.950 GHz and 1.5 volts. It's no "suicide run". The computer will handle it just fine. That's one of the reasons a sealed CPU water cooler works in this situation. It keeps temps under control. As I have posted in other threads, temps are well below the critical limits.
--
Need-based health care not greed-based health care.

me1212

join:2008-11-20
Pleasant Hill, MO
reply to Chrno
1.5v is the max intel says wont rip apart your cpu(well something to that effect), 1.35v is what they recomend you limit it to, or so i've read. hence why I try to keep mine under 1.35V.

me1212

join:2008-11-20
Pleasant Hill, MO
reply to me1212
I've noticed that whenever I use offset, unless it adds voltage, it causes errors in prime 95 after about 20 min, always on core 0. is that supposed to happen? if I dont use offset I get voltages of 1.256-1.272 most of the time, but can spike up to like 1.32V for a second or two then drop back down. and trying to used fixed voltages always causes BSOD not matter what settings I use with it, but I think that may have something to do with my wireless card, always disables it on next start up when that BSOD happens.

Am I doing something wrong or is my chip just one that needs more voltage?

matt5

join:2001-10-06
Lagrangeville, NY
What you are seeing is vdroop, Say you set 1.3, under load, you will see more like 1.25, this is because of as I said vdroop, you are setting the MAX voltage it will *ever* hit, not the load voltage, when the cpu comes on and off load you get spikes as the VRM can not adjust the voltage in .000000000000000000000000001 seconds...

If you are getting errors, you likely need more vcore, core 0 is more hungry for voltage than the others... welcome to overclocking.

Like I said, everyone is saying o I get this that... it all depends ON THE CHIP. Yours might be golden, or it might be meh, or suck for top end overclocking... you will find out as you try to overclock.

1.27 is still pretty low and you have room to bump it up.


bobbagels
Just Another Scorpion Mechwarrior
Premium
join:2000-11-15
Matawan, NJ
kudos:2
reply to me1212
it has to do with VDroop
Try setting Load Line calibration to regular. If that dosent help set it up to the next level (high i think it is)

You gotta play with it and find the setting thats right for you.

I have my LLC set to regular with a +0.60 offset and im pretty solid at 4.5 Ghz

Also, common practice is to get your machine running stable at stock. Then OC, then install extra cards such as wireless card. but none the less, you are where your at now so I would just continue on playing with the settings (LLC and Offset V)

Aside from the Clunk web site I recommended which is great for the basic and first OC I know there isnt much info there for offset .
Here are a few sites with lots of info for you.
»forums.overclockers.co.uk/forumd···php?f=89
»hardforum.com/forumdisplay.php?f=80
»www.xtremesystems.org/forums/sho···ormation

And again Matt is correct you have plenty of room to bump up the VCore if need be
--
Last night I saw a naked cowgirl, she was floatin across the ceilin

me1212

join:2008-11-20
Pleasant Hill, MO
reply to matt5
Ok, thanks for the info guys. I've got it set to auto right now, not seeing any errors or anything so far, tops out at 1.288V when prime95, 1.32V on minecraft. I can deal with 1.32V, not too bad i've seen some needed 1.34 or more for 4.5Ghz so mines not too bad. I'll play sround wit hthe vdroop and voltage settings. and if the spikes are just from coming off a load and the vrm not adjusting instantly thats fine, beats having it at 1.32V all the time.

my mobo uses different terms for vdroop and llc and so i'll read the manual and see if it says whats what in there.

matt5

join:2001-10-06
Lagrangeville, NY
reply to me1212
I would not take off LLC, as they is implemented by intel to control the voltage spikes, would you rather set the MAX voltage that the cpu will see... or the min and have no idea what it is spiking too?

vdroop, is good, VRMs can never adjust instantly, they can not predict the cpu work load.

me1212

join:2008-11-20
Pleasant Hill, MO
Click for full size
I would rather see what it can max to then see that its min can be and not know what its max is/can be.

And I think I finally found a stable setting. Yeah i only ran it for about 44 min, but everything else I have tried failed in under half that time. and according to cpu-z its max voltage never goes above 1.32V I think thats acceptable

matt5

join:2001-10-06
Lagrangeville, NY
When testing, I would use IBT, what took 20min in p95 to fail likely would have failed in the first pass of IBT...

I would do a p95 run overnight at least, when you are happy with the speed you have as a final test (I would also run IBT)

me1212

join:2008-11-20
Pleasant Hill, MO

1 edit
IBT meaning intel burn test I assume? and I will but not until next weekend, college starts backup tomorrow.

EDIT: also im gonna game some this week so if it fails during that i'll know that it needs tweaking that way too.

matt5

join:2001-10-06
Lagrangeville, NY
Yes Intel Burn Test...