i5 2500kOC on P8P67

[SuPrEIVIE]

Finally tried out OCing and got some decent results. First here are specs

i5-2500K with A70
P8P67 standard
8GB DDR3 1333
5850 1GB
TX750 PS


My initial test was 47x100 and i accidentally applied an excessive increment of voltage from the actual base. All of this in bios by the way dont really trust AIsuite at all especially since it periodically says my rails are running at a voltage way off the spec, which really isnt happening. I thought that the base voltage was what was being shown at the entry above the offset row so it showed 1.168V and so I added like +.132V to make it around 1.3V, but when i got into windows aida64/cpuz was showing 1.465V after running prime95! I had LLC at high as well, but then went back to bios to only apply +.030V at offset and turned LLC off. Then i was able to run 47x100 at ~1.3 for 2hrs running prime95 before it crashed. 47x100 was my mark but i dont mind settling with 46X and keep vcore down.


With only +.005 offset i am able to run 46X for 10 hrs plus on p95 at <1.280V! In result hottest core temp ever reached was 62C

Is this good for 24/7 operation? Really looking for atleast 5 year lifespan out of this. Wondering if i should go further and negate the offset to see if it can still do 46x.

One thing i learned through all this is just running prime95 still puts computer to sleep and i had to turn it off.

Also it was running as a bench run so motherboard was on its box untill I called 46x stable and put everything in case, but now im getting this weird input lag from keyboard and mouse in Bios. Really weird mouse pointer behaves strange as well. I had hit delete to enter bios and with lag instead took me to windows.

 

[BTRY B 529th FA BN]

24/7 operation would be a little hard on the electric bill.

Back if off a little, don't kill your rig.

EDIT: also try LinX and Memtest HCI - Run LinX as admin, btw ...or else it will act like it's "running" and not be doing anything at all... and to double check just quickly open Task Manager to see if all the cores are maxed..

 

[Wuzup101]

I did the same damn thing on my Asus P67 sabertooth with a 2500k. I was overclocking by setting the voltage manually at first... found that I was getting good stability at 4.5-4.6 using ~1.3 volts (as read by the cpuz). I wanted to try to use some power saving features, so I decided to use the offset instead of controlling voltage manually. The bios was reading 1.1 something... so I added 0.150... and I fire up prime / cpuz / hw monitor... and find out I'm at 1.415 in cpuz (or something really close to that).

So I'll pose this question as an extention of the OP... how exactly does "offset" work. Does ASUS have a base voltage table for different clocks? I'm running +0.010 offset right now at 45x103. It's definitely running more than 0.010 above what it did when I was running the chip at stock speeds.

 

[Axon]

What version CPUZ? Even with the newest 1.57.1 version I could find, I still find some of the readings are off.

 

[Jmman]

My new rig with the i2500 and the Gigabyte UD4 is 100% stable at 4.5ghz with the stock voltage......I havent really tested its limits though......

 

[Wuzup101]

What version CPUZ? Even with the newest 1.57.1 version I could find, I still find some of the readings are off.

I'm personally using 1.57 (says Feb 2011 in the "about" tab). I believe I downloaded it thursday in preparation for this new build. What are you comparing CPUz to that you know the voltages are off?

 

[Hogan773]

I did the same damn thing on my Asus P67 sabertooth with a 2500k. I was overclocking by setting the voltage manually at first... found that I was getting good stability at 4.5-4.6 using ~1.3 volts (as read by the cpuz). I wanted to try to use some power saving features, so I decided to use the offset instead of controlling voltage manually. The bios was reading 1.1 something... so I added 0.150... and I fire up prime / cpuz / hw monitor... and find out I'm at 1.415 in cpuz (or something really close to that).

So I'll pose this question as an extention of the OP... how exactly does "offset" work. Does ASUS have a base voltage table for different clocks? I'm running +0.010 offset right now at 45x103. It's definitely running more than 0.010 above what it did when I was running the chip at stock speeds.

I found similar with my ASRock. I thought OFFSET was amount above or below STOCK voltage ie for me 1.18V. So I added .07 to target 1.25. Fire it up and I see 1.36 or something. Shut it down again and then now I'm at offset MINUS .05 at 42x......

It must be as you say, the mobos have some voltage calc or table or something to shoot for at various multiplier settings. And they must be pretty aggressive.....

 

[SuPrEIVIE]

Im also running cpuz 1.57 and vcore seems to fluctuate similarly to aisuite program and aida64. I would like to set vcore to default but if you leave offset mode at auto that means motherboard adjust increments on its own so rather than put a manual static vcore I just put the smallest +.005 increment possible. I tried a -.005 range, but in like 23 mintues of prime, small fft on the 4th worker thread got an error no bluescreen though. No big deal since i didnt really see temp decrease benefit. Was suprised to see though after finally closing the lid on the case temps spiked 3C.

On my kb & mouse issue in bios problem, issue was created by having my psx to usb controller adapter connected. Somehow the adapter interferes with input so if anybody gets this issue try unplugging some extra perhipherals while in bios.


Also anybody getting these weird periodic "false alarms" of voltages and RPM of certain fans in aisuite2?

 

[Axon]

I'm personally using 1.57 (says Feb 2011 in the "about" tab). I believe I downloaded it thursday in preparation for this new build. What are you comparing CPUz to that you know the voltages are off?

BIOS settings. I set voltages at, say, 1.3, and I get a reading of 1.325 or whatever.

 

[Wuzup101]

BIOS settings. I set voltages at, say, 1.3, and I get a reading of 1.325 or whatever.

Ah I got you. I didn't see any occasions when CPUz reported a higher value than I had set in BIOS (when I was manually setting voltage).

Im also running cpuz 1.57 and vcore seems to fluctuate similarly to aisuite program and aida64. I would like to set vcore to default but if you leave offset mode at auto that means motherboard adjust increments on its own so rather than put a manual static vcore I just put the smallest +.005 increment possible. I tried a -.005 range, but in like 23 mintues of prime, small fft on the 4th worker thread got an error no bluescreen though. No big deal since i didnt really see temp decrease benefit. Was suprised to see though after finally closing the lid on the case temps spiked 3C.

On my kb & mouse issue in bios problem, issue was created by having my psx to usb controller adapter connected. Somehow the adapter interferes with input so if anybody gets this issue try unplugging some extra perhipherals while in bios.


Also anybody getting these weird periodic "false alarms" of voltages and RPM of certain fans in aisuite2?

I'm actually using +.005 right now too and that seems to be working. And yes, I get the errors from the aisuite2 program all the time... and they are nuts like 5v rail @ 0V...

 

[Dadofamunky]

Finally tried out OCing and got some decent results. First here are specs

i5-2500K with A70
P8P67 standard
8GB DDR3 1333
5850 1GB
TX750 PS


My initial test was 47x100 and i accidentally applied an excessive increment of voltage from the actual base. All of this in bios by the way dont really trust AIsuite at all especially since it periodically says my rails are running at a voltage way off the spec, which really isnt happening. I thought that the base voltage was what was being shown at the entry above the offset row so it showed 1.168V and so I added like +.132V to make it around 1.3V, but when i got into windows aida64/cpuz was showing 1.465V after running prime95! I had LLC at high as well, but then went back to bios to only apply +.030V at offset and turned LLC off. Then i was able to run 47x100 at ~1.3 for 2hrs running prime95 before it crashed. 47x100 was my mark but i dont mind settling with 46X and keep vcore down.


With only +.005 offset i am able to run 46X for 10 hrs plus on p95 at <1.280V! In result hottest core temp ever reached was 62C

I think that's a really good result and I'm going to try that. That's great for 24/7 but you can probably turn on c1E and EIST to reduce power draw. If you pass ten hours of Prime, in my experience that's as good a stability test as you can ask for - 30-40 passes of IntelBurnTest is a good supplemental one.

 

[SuPrEIVIE]

yeah it was, but now running small fft in prime and the 4th core got an error after 1hr and 30+ mins in. I believe the test i used last time was large fft that worked for 10 hrs at the +.005V increment for 46x. Rather than add more vcore range to stabilize it i just dropped to 45x. 4th core hasnt errored yet in small fft for 2hrs and counting so we'll see...

c1E and EIST are on by default i believe so i just left those as so.

 

[aldamon]

So I'll pose this question as an extention of the OP... how exactly does "offset" work. Does ASUS have a base voltage table for different clocks? I'm running +0.010 offset right now at 45x103. It's definitely running more than 0.010 above what it did when I was running the chip at stock speeds.

I found similar with my ASRock. I thought OFFSET was amount above or below STOCK voltage ie for me 1.18V. So I added .07 to target 1.25. Fire it up and I see 1.36 or something. Shut it down again and then now I'm at offset MINUS .05 at 42x......

It must be as you say, the mobos have some voltage calc or table or something to shoot for at various multiplier settings. And they must be pretty aggressive.....

I can't figure out Offset either on my TP67B+. While experimenting at 4.8 GHz, I set it to Offset +000 and IIRC, I was cooking at 1.46V+ under Prime95 load. It appears I need negative values to use Offset so I'm sticking with Fixed and 4.7 GHz for now (1.368V).

I also tried Auto and Spec and those cooked the CPU as well. I think I saw 1.5V at one point and immediately shut down the computer.

 

[Hogan773]

I can't figure out Offset either on my TP67B+. While experimenting at 4.8 GHz, I set it to Offset +000 and IIRC, I was cooking at 1.46V+ under Prime95 load. It appears I need negative values to use Offset so I'm sticking with Fixed and 4.7 GHz for now (1.368V).

I also tried Auto and Spec and those cooked the CPU as well. I think I saw 1.5V at one point and immediately shut down the computer.

As I said in a previous post I use offset MINUS .05 (not .005, but .05) at 42X which gives me 1.21/1.22 area under Prime.

I've used this as my 24/7 OC for past month.

I often leave HW Monitor on while the computer is on during the day. Usually the Max column will show 1.26V although I am never quite sure at what point it spiked to 1.26.....as it runs 1.22 under Prime and I can never actually see a point of 1.26, although I guess that is one of the functions of HW Monitor....to MONITOR the hardware for me and tell me later......

I'm trying to see what I can get for a 24/7 OC in the 44-45x range in terms of voltage. Crapped out on minus .05 and minus .04 last night as both BSODed about 4 mins into Prime. Seemed like minus .03 might be OK although I only let it run about 15 mins and then decided to turn it off and go to bed.

 

[SuPrEIVIE]

I left 45x at +.005V offset running on prime 95 for 10 hrs and got 0 errors so im going to call it done. Temps seemed a bit high to me like avg 63 with upperlimit of 66C. Anyway i notice that as it hits 45x for non max out tasks, voltage is ~1.3V like .04-.05V above what it has on prime tasks. Also the temp differential in this scenario between cores is much greater than what i had seen on LGA775 like 9C difference.

 

[Hogan773]

I left 45x at +.005V offset running on prime 95 for 10 hrs and got 0 errors so im going to call it done. Temps seemed a bit high to me like avg 63 with upperlimit of 66C. Anyway i notice that as it hits 45x for non max out tasks, voltage is ~1.3V like .04-.05V above what it has on prime tasks. Also the temp differential in this scenario between cores is much greater than what i had seen on LGA775 like 9C difference.

when I tried 44x and minus .03 it STARTED Prime at around 1.23 or something, and after the temps rose about 4 mins into it, the voltage increased to 1.26/1.27.

When you say "non max out tasks" is there a specific burn test or something you're referring to? Or just daily use? I could see that under full load, the voltage may be lower due to VDroop. Incidentally, my OCing is done with Load Line Control at "5" which is the lowest setting on ASRock (meaning it isn't adding any Load Line compensation). I've heard that with LLC you can get momentary spikes above the max values you're targeting when you move from a loaded to a non-loaded state.....

I still have to decide if the extra voltage (closer to 1.3) is worth it for an extra 200mhz vs voltage in the 1.22 range. I wish I knew whether 1.3V was so easy on the processor that it was taking the potential life from 10 years to 9.7, versus 10yrs to 5. I was comfortable with the 1.22V OC because I figured it's so close to stock. I just don't know how "steep" the damage curve is when going from 1.2 to 1.3 to 1.35 etc. I assume its not linear but I have no idea actually. All you EEs out there - ideas?

 

[SuPrEIVIE]

yes non max out tasks like in games and daily use, I havent tested to see how the vcore is while doing these tasks at a longer sustained time frame. I also have LLC at zero as I felt it overcompensates on vroop.

I researched a bit and in another forum one guy spoke with an intel representative about the max vcore tolerated by these chips and it was said to be 1.52V. Now that has no indication on how it impacts its lifespan, but to think of it it im sure at that rating it would die within the 3 yr warranty.

 

[Hogan773]

yes non max out tasks like in games and daily use, I havent tested to see how the vcore is while doing these tasks at a longer sustained time frame. I also have LLC at zero as I felt it overcompensates on vroop.

I researched a bit and in another forum one guy spoke with an intel representative about the max vcore tolerated by these chips and it was said to be 1.52V. Now that has no indication on how it impacts its lifespan, but to think of it it im sure at that rating it would die within the 3 yr warranty.

???

I don't want to do something that would have it be "sure to die within 3yrs"

That's why I'll probably play it safe.

I just wish I knew how the damage curve looked - ie is going from 1.2 to 1.3 basically the same, but its when you're going from 1.35 to 1.4 that is starting to really hurt it, and 1.4 to 1.5 is very damaging, etc. Basically is 1.3 NEARLY AS SAFE as 1.2? (or not)?

 

[MustangSVT]

if you do oc sb to 4.6ghz, what speed does it throttle down to in idle or on low usage?

 

[Dadofamunky]

That would be 1.6 with C1E and EIST turned on.