Sign in with
Sign up | Sign in
Your question

Sandy Brdige 2500K OC Build

Last response: in Overclocking
Share
February 21, 2011 2:08:57 AM

I just got time today to overclock mine. Now I have it OC'ed to 4500 MHz it tops out at 60C under Prime 95 Stress Test if you go by the CPU temp in ASUS Utilities... However I had to use Offset Voltage instead of manual. ASUS and CPU-Z say the range of the vcore is 1.29-1.32 I tried upping it manually going from 1.2v but always got a BSOD although at 2.8v it just froze at the login screen and at 2.9v it almost finished booting up before BSOD 1.12-1.28 and 1.29-1.35 were a bust using manual settings. at idle it is 1.007v

If you need I can give you logs from Prime 95 and the last 40 mins of the RealTemp Log (last 40 mins as I didn't realize it didn't autolog until it had been running for 20 mins) idk what good they'll do you...

Anyways I was wondering if it was safe to go higher? I can also post my parts list again if needed.
I had it stable at 1.2v at 4 GHZ using offset probably go lower but manual wouldn't work. Oh also something weird I noticed was even @4500 MHz it's only pulling 80W or something like that I'll double check but lol.
Also what's the latest news on the SATA bug is ASUS issuing a recall for the motherboards?

At one point I had read that 1.35 was the safe vcore settings. But now I'm hearing 1.52 I'm also talking about a mostly on system I put it to sleep when it's doing absolutely nothgin so it whould have an 8 hour break every 24 hours or so if not every 48 hours or so.
a b à CPUs
a b K Overclocking
February 21, 2011 3:02:34 AM

I would use Core Temp for more accuracy and individual core temps. If those are at 60ºC under Prime95 load, you're doing well.

You could probably go higher than 4.5GHz but voltage (and the "multiplier wall" of your particular CPU) would be the limiter. Sandy Bridge can register up to 1.52v, but it would be better to keep it under 1.4v or so just in case.

Note that an hour's worth of Prime95 isn't really enough of a stability test.

I use two tests to assure myself of complete stability:
1. 24 hours of Prime95 Torture Test using "In-place large FFTs" setting.
2. 50 runs of IntelBurnTest v2.50 on Maximum setting.

All manufacturers have issued recall orders for their boards. However, replacements won't be available until mid-March or early April. You don't need to RMA until they are available.
m
0
l
February 21, 2011 3:21:34 AM

Thanks I will wait on RMA then. Um is there a program that you would recommend to give individual core temps and such (free that is) that I could have log all the observations and then post here for eval? I used thst to double check it worked before wasting more time upping OC'ing it more.
m
0
l
Related resources
a b à CPUs
a b K Overclocking
February 21, 2011 4:14:05 AM

Core Temp keeps track of each core's current temp, its lowest temp, and its highest temp. It does have a logging feature, but I don't need it because it already keeps track of the low/high/current temps for each core.

You don't actually need to post logs here, as they tend to have too much info to be helpful. Just keep track of your highest temps with Core Temp. As long as you stay below 70ºC or so at full load, you'll be fine.
m
0
l
February 21, 2011 4:21:29 AM

The highest core Temp I got was Core 3 which was 70C.
I believe ASUS utils (which is where I got the 60C from) displays the TCASE. Cores 1, 2 and 4 were in the ranges of 60-67C during the test acording to Real Temp
m
0
l
a b à CPUs
a b K Overclocking
February 21, 2011 4:44:10 AM

You can go higher than that, but not much. I would definitely stay at or below 75ºC peak full load temp for an everyday overclock.
m
0
l
February 21, 2011 5:01:29 AM

Alrighty I was reading that the Tcase temp was 75.6 on intels's site for no more than that obviously I'm also thinking that my cooler may be insufficient because at 4000 MHz doing X264 video encoding Full CPU load the Temps Maxed out at 45 TCase
It’s this one http://www.coolermaster.com/product.php?product_id=5983
and it seems good. Anyway the system drops immediately from 60-70C per core to 28-32C per core the moment I stop the test. And nothing but cool to lukewarm air comes out of it there were a few times the air that was coming from the case was a bit warm but not unbearable.

“As long as you stay below 70ºC or so at full load, you'll be fine.” When you say that do you mean 70C per core or 70C TCASE ?
m
0
l
a b à CPUs
a b K Overclocking
February 21, 2011 5:34:57 AM

Per core. That also goes for my "stay at or below 75ºC peak full load temp for an everyday overclock" comment. That's what I personally hold to, just to make sure I don't fry my chip too soon.

The TjMax value (max safe temp per core) is 98ºC, and the CPU will start auto-throttling in the low 90sºC on individual cores. At least, that's what happened on my system with the stock low-profile Intel cooler when I deliberately overclocked and over-volted too high to test it out.
m
0
l
February 21, 2011 7:56:36 AM

Well I have it at 4.7 MHz stable now BUT only after using offset which puts the volts at 1.4 ( manual always gives BSOD and I can’t figure out why) and during the stress test it went up to 75C on core 3 73C on core 4 68C on core 1 and 69C on core 2 I stopped it after 50 mins. I also have a few days to return the cooler above idk how much more than that I could spend though... max would be maybe 74ish bucks so maybe 35 ish more any recommendations?
m
0
l
a b à CPUs
February 21, 2011 8:47:42 AM

Im overclocking a 2500k on an MSI mainboard. I got mine up to 5000Mhz fairly stable, but I was shoving almost 1.5v into it. That is what CPUz was showing under load - 1.49/1.5v. I have a Mugen2 cooler with 2x120mm fans (one is on 100% all of the time) and my temps at 5Ghz would touch 73/74 degrees, then come back down and stabilise around 68-69. But I have opted to aim for about 4.8Ghz, because I would rather run my CPU at about 1.35v I think. I left it last night at 4.5Ghz with 1.3v and little or no tweaking. Getting it from 4.8 to 5Ghz requires a lot more power.

Half the reason for me butting in here is to ask What is PLL voltage? Is it safe to mess with it. I read somewhere that "anything up to 1.9v is fine" and I set mine at one point to 1.94v, but I have no idea what it is doing. Or what would be a good level to set it at.
m
0
l
a b à CPUs
a b K Overclocking
February 21, 2011 4:04:20 PM

Yeah, I've had my 2500K at 5.1GHz and 5.2GHz, but it wasn't stable. To get complete stability, I had to clock it down to 4.8GHz

I use two tests to assure myself of complete stability:
1. 24 hours of Prime95 Torture Test using "In-place large FFTs" setting.
2. 50 runs of IntelBurnTest v2.50 on Maximum setting.

Voltages from a guide I go by:
VTT/VCCIO: 1.1v default, 1.2v if OCing near or above 5.0GHz, 1.4v ceiling
VCCSA: 0.9v default, 1.3v if OCing to high BCLK, 1.4v ceiling
PCH: 1.0v default, likely don't have to touch this ever
CPU PLL: 1.75v?? default, 1.9v stabilizes high OCs, 2.0v ceiling (my ASRock P67 Pro3 defaults to 1.832v for this setting)
m
0
l
a b à CPUs
February 21, 2011 5:12:17 PM

Leaps-from-Shadows said:
Yeah, I've had my 2500K at 5.1GHz and 5.2GHz, but it wasn't stable. To get complete stability, I had to clock it down to 4.8GHz

I use two tests to assure myself of complete stability:
1. 24 hours of Prime95 Torture Test using "In-place large FFTs" setting.
2. 50 runs of IntelBurnTest v2.50 on Maximum setting.

Voltages from a guide I go by:
VTT/VCCIO: 1.1v default, 1.2v if OCing near or above 5.0GHz, 1.4v ceiling
VCCSA: 0.9v default, 1.3v if OCing to high BCLK, 1.4v ceiling
PCH: 1.0v default, likely don't have to touch this ever
CPU PLL: 1.75v?? default, 1.9v stabilizes high OCs, 2.0v ceiling (my ASRock P67 Pro3 defaults to 1.832v for this setting)


Thanks for the info man. this will be quite useful for experimenting again :)  VTT/VCCIO - thats not the same as VCore is it? I have a "CPU I/O" or something. Is that the same thing as "VTT/VCCIO" does anyone know?
:S

Also what is your VCore @ 4.8Ghz?
m
0
l
a b à CPUs
a b K Overclocking
February 21, 2011 6:39:14 PM

VCCIO basically means "voltage I/O" which is likely the same as "CPU I/O" in your case.

I overclock using one of the presets in the ASRock's UEFI, which uses Auto for Vcore. In programs it registers as 1.5012v (Core Temp) or 1.504v (CPU-Z) at full load. I tried unsuccessfully for days to get a manual 4.8GHz overclock stable, but for the life of me I couldn't figure it out. I could get 4.7GHz, but not 4.8GHz for whatever reason. Even using that same CPU voltage didn't work. It would usually BSOD or hard-lock when coming down from load, not while at load. So now I just use the "Turbo 4.8GHz" preset.
m
0
l
a b à CPUs
a b K Overclocking
February 23, 2011 5:37:26 AM

Leaps-from-Shadows said:
Yeah, I've had my 2500K at 5.1GHz and 5.2GHz, but it wasn't stable. To get complete stability, I had to clock it down to 4.8GHz

I use two tests to assure myself of complete stability:
1. 24 hours of Prime95 Torture Test using "In-place large FFTs" setting.
2. 50 runs of IntelBurnTest v2.50 on Maximum setting.

Voltages from a guide I go by:
VTT/VCCIO: 1.1v default, 1.2v if OCing near or above 5.0GHz, 1.4v ceiling
VCCSA: 0.9v default, 1.3v if OCing to high BCLK, 1.4v ceiling
PCH: 1.0v default, likely don't have to touch this ever
CPU PLL: 1.75v?? default, 1.9v stabilizes high OCs, 2.0v ceiling (my ASRock P67 Pro3 defaults to 1.832v for this setting)


and what guide would that be.. :pfff: 
m
0
l
!