n00b overclocking question

Printable View

03-13-2010, 07:23 PM
Ezekiell

n00b overclocking question

Hi,
I got recently a 965BE C3, 4GB of G.skill 12800 Eco and a Gigabyte 890GPA-UD3H mobo. I'm currently running the ram @ 1333mhz 6-7-6-18 1t @ 1.5v with a 5.33 mem multiplier and CPU freq @ 250 ( I'm doing this because the board will only boot otherwise with ram @ 1333mhz CL9 or 1600mhz @ CL8)

I'm currently at 3875mhz @ 1.45v (not sure if it's 100% stable yet, only did 1H OCCT) and NB freq @ 2250mhz, stock volt.

When you increase the NB freq (my goal is 2.5ghz), which voltage do you need to increase for stability? Is it the NB voltage or the CPU-NB voltage? And what is the max safe volt on air

thanks
03-13-2010, 07:29 PM
BeepBeep2

CPU_NB voltage is what you increase.

NB VID is what it currently shows in AMD Overdrive. Make sure to flash the latest BIOS here: http://www.gigabyte.com.tw/Support/M...ctID=3342&ver=

What kind of temps are you getting on your cooler? Save volts are usually 1.52-1.53v for 24/7, 1.55v is pushing it, you will probably see CPU degradation within a few months at 1.55v (Not much but it will be noticable). Safe volts on the CPU_NB for 24/7 are probably 1.425v, 1.45v for testing...theoretically it's the same as the CPU but nobody has really pushed it that far and you won't need it that far. I got 2.8 CPU_NB stable on only 1.375v. Right now I'm running 4 Ghz (20x200) @ 1.52v, DDR3-1600 7-6-6, CPU_NB @ 2.8 1.375v

1 Hour OCCT is enough for me and all I do to stability test, maybe Prime95 Blend 1 or 2 hours but...

The 890GPA-UD3H only has 4+1 Phase CPU PWM, I'd probably not trust it for multiple hours (ie 5-6 hours+ or something)
03-13-2010, 08:01 PM
Titan7171

Quote:

Originally Posted by BeepBeep2

CPU_NB voltage is what you increase.

NB VID is what it currently shows in AMD Overdrive. Make sure to flash the latest BIOS here: http://www.gigabyte.com.tw/Support/M...ctID=3342&ver=

What kind of temps are you getting on your cooler? Save volts are usually 1.52-1.53v for 24/7, 1.55v is pushing it, you will probably see CPU degradation within a few months at 1.55v (Not much but it will be noticable). Safe volts on the CPU_NB for 24/7 are probably 1.425v, 1.45v for testing...theoretically it's the same as the CPU but nobody has really pushed it that far and you won't need it that far. I got 2.8 CPU_NB stable on only 1.375v. Right now I'm running 4 Ghz (20x200) @ 1.52v, DDR3-1600 7-6-6, CPU_NB @ 2.8 1.375v

1 Hour OCCT is enough for me and all I do to stability test, maybe Prime95 Blend 1 or 2 hours but...

The 890GPA-UD3H only has 4+1 Phase CPU PWM, I'd probably not trust it for multiple hours (ie 5-6 hours+ or something)

Dont klnow about that cpu degradation over 1.55v........I been running this 965 c3 since release at 1.57v at 4.2ghz.......no cpu degradation here also ran my 965 c2 for 6 months at 1.57v never seen a problem.

Beter safe than sorry for this guy anyway though:)
03-13-2010, 08:44 PM
BeepBeep2

Quote:

Originally Posted by Titan7171

Dont klnow about that cpu degradation over 1.55v........I been running this 965 c3 since release at 1.57v at 4.2ghz.......no cpu degradation here also ran my 965 c2 for 6 months at 1.57v never seen a problem.

Beter safe than sorry for this guy anyway though:)

Well, Flank3r noticed a ~30 Mhz drop in stability after he fed his C2 955BE 1.55v for too long :yepp:...better safe than sorry I suppose.
03-13-2010, 08:50 PM
chew*

I did a controlled experiment at 1.6vcore.

Removed lid from cpu, ran it at cool temps in prime 95 @ 4.2.

Shut system down. Let everything get to normal temps.

Posted next day at normal temps = dead chip almost instantly.

Take away the buffer ( IHS ) and the chip can die rather easy.

With the buffer there is potential to degrade the chip although I doubt it will spike it and kill it instantly I can assure you that damage can be done above 1.55v at the average users ambient temps.
03-13-2010, 09:05 PM
BeepBeep2

Quote:

Originally Posted by chew*

I did a controlled experiment at 1.6vcore.

Removed lid from cpu, ran it at cool temps in prime 95 @ 4.2.

Shut system down. Let everything get to normal temps.

Posted next day at normal temps = dead chip almost instantly.

Take away the buffer ( IHS ) and the chip can die rather easy.

With the buffer there is potential to degrade the chip although I doubt it will spike it and kill it instantly I can assure you that damage can be done above 1.55v at the average users ambient temps.

What? I've spent at least 30 minutes with this CPU doing valids and 1M @ 1.6v in normal ambients (even idling around 40c at times), and up to 1.63v in the cold...:confused:

I don't really understand how the (lack of) IHS makes the chip die so quickly???
03-13-2010, 09:11 PM
chew*

Quote:

Originally Posted by BeepBeep2

What? I've spent at least 30 minutes with this CPU doing valids and 1M @ 1.6v in normal ambients (even idling around 40c at times), and up to 1.63v in the cold...:confused:

I don't really understand how the (lack of) IHS makes the chip die so quickly???

It's a buffer it basically makes the DIE spread heat instead of being concentrated thus the name Integrated heat spreader.

Remove that buffer and the chip becomes far less tolerant to voltage and can spike easily.

For reference I tested this from stock volts to 1.6 where it was finally enough to spike the chip and kill it.

My intention was to kill it and also find out at what voltage for personal data.

If it can spike and kill instantly at 1.6 it can surely degrade with the buffer.

Usually voltage above 1.55 is enough to peg the chips at 55C anyway which is the overall average temp that most chips stability breaks down under high clocks @ normal ambients.

Actually compiling some interesting ln2 data as well but the data is not complete yet.
03-13-2010, 10:03 PM
Daveburt714

I'm actually a pretty timid OC'r. Guess thats why I never put up any really impressive numbers... :p:

I tend to do things VERY slowly, and when I get to a point where increasing voltages only shows a marginal performance boost, I call it good.

Temps can definetly be a problem, and anything over 53c on the cores has been unstable for me on almost every Phenom I've had....

Granted I've never worked with extreme cold, but under water and aftermarket air I've found that temps don't seem to be the only limiting factor.

My WC rig now can keep temps under 50c up to 1.6v's. Anything over 1.525 on the core, and 1.3 on the IMC doesn't seem to help though, reguardless of temps.

C2's seemed to be a little more tolerant of high V's, but I have 2 C3's and neither one of them really benefits from CPU-NB v's >1.3v or Core v's >1.525 even at temps under 50c...

From what I've seen things seem to be really different when using extreme cooling (Dice/Ln2), but for folks using more conventional cooling just keep in mind that temps may not be the only limiting factor... ;)
03-13-2010, 10:06 PM
BeepBeep2

Quote:

Originally Posted by Daveburt714

...
My WC rig now can keep temps under 50c up to 1.6v's. Anything over 1.525 on the core, and 1.3 on the IMC doesn't seem to help though, reguardless of temps.

C2's seemed to be a little more tolerant of high V's, but I have 2 C3's and neither one of them really benefits from CPU-NB v's >1.3v or Core v's >1.525 even at temps under 50c...

What CPU_NB speed are you running? I get scaling all the way up till 1.45v on the CPU_NB... :confused:
03-13-2010, 10:23 PM
Daveburt714

Quote:

Originally Posted by BeepBeep2

What CPU_NB speed are you running? I get scaling all the way up till 1.45v on the CPU_NB... :confused:

Both my C3's (965/555 unlocked) both seem to top out around 2.6Ghz IMC.
Trust me... I've pushed the V's on both of them all the way up to 1.4 and they both seem to quit scaling around 2.6.

Like I said, I'm a wuss. :p:

But when I bring the IMC up .5v's and still can't squeeze another drop out of it (stable) I call it quits...

Call me chicken if you want... But I've never had a chip degrade... :D

I've been playing with the 555 tonight. It's IMC is a little weaker than the 965 but it's running Prime stable 2.51 @ 1.25v's...

I've been pushing tonight, but even @ 1.325 I can't get it stable @ 2600... :shrug:
03-13-2010, 10:26 PM
BeepBeep2

Quote:

Originally Posted by Daveburt714

Both my C3's (965/555 unlocked) both seem to top out around 2.6Ghz IMC.
Trust me... I've pushed the V's on both of them all the way up to 1.4 and they both seem to quit scaling around 2.6.

Like I said, I'm a wuss. :p:

But when I bring the IMC up .5v's and still can't squeeze another drop out of it (stable) I call it quits...

Call me chicken if you want... But I've never had a chip degrade... :D

I've been playing with the 555 tonight. It's IMC is a little weaker than the 965 but it's running Prime stable 2.51 @ 1.25v's...

I've been pushing tonight, but even @ 1.325 I can't get it stable @ 2600... :shrug:

Hmm, it could be the board...but I'm purely speculating. I can do 2600 on about 1.225v :yepp: :(
03-13-2010, 10:30 PM
Daveburt714

You know... I've been wondering about that!

Our boards are/were very similar though.. 790X-UD4P vs 790XT-UP4P.

My 555 wouldn't unlock to a quad stable on an Asus M3A79-T Deluxe, but it's rock solid as a quad on this GBT.

What was you best IMC clock on the 790X Beep?

*** EDIT: My appologies to the OP! Didn't mean to hijack your thread...
03-13-2010, 11:00 PM
Titan7171

Quote:

Originally Posted by BeepBeep2

Well, Flank3r noticed a ~30 Mhz drop in stability after he fed his C2 955BE 1.55v for too long :yepp:...better safe than sorry I suppose.

Flanker is on air.......Im guessing heat might have been an issue:confused:

Quote:

Originally Posted by chew*

I did a controlled experiment at 1.6vcore.

Removed lid from cpu, ran it at cool temps in prime 95 @ 4.2.

Shut system down. Let everything get to normal temps.

Posted next day at normal temps = dead chip almost instantly.

Take away the buffer ( IHS ) and the chip can die rather easy.

With the buffer there is potential to degrade the chip although I doubt it will spike it and kill it instantly I can assure you that damage can be done above 1.55v at the average users ambient temps.

I definitely think 1.55v on air for 24/7 is way to high......I myself would only run up to 1.53v on air........1.55v was just to much for the HS to cool But with water you have much more room to play with

Quote:

Originally Posted by chew*

.

Usually voltage above 1.55 is enough to peg the chips at 55C anyway which is the overall average temp that most chips stability breaks down under high clocks @ normal ambients.

.

Yes Im glad oyu said this, its what Ive seen also for air:yepp:
03-13-2010, 11:12 PM
chew*

It died in my testing on water ;).
03-13-2010, 11:14 PM
BeepBeep2

Quote:

Originally Posted by Daveburt714

You know... I've been wondering about that!

Our boards are/were very similar though.. 790X-UD4P vs 790XT-UP4P.

My 555 wouldn't unlock to a quad stable on an Asus M3A79-T Deluxe, but it's rock solid as a quad on this GBT.

What was you best IMC clock on the 790X Beep?

*** EDIT: My appologies to the OP! Didn't mean to hijack your thread...

3000 in cold in sub-zero ambients, about 2800 stable inside 2900 benching............:shrug: