[MasterG]

I've decided to swap out my current mobo for another Biostar, this time the Tpower I45. The D2-A7 is a great motherboard (and cheap), but it is pretty limited in terms of overclocking and it doesn't allow for that precise control like on the I45.

At the moment, I'm running right around 4GHz at 1.36V which is somewhat less than what Intel lists as the max core voltage for the E8500 (1.45V). I've heard of people running past this magic number, but the thread title says it all. How much is too much?

(comp specs below)

EDIT: Intel lists several specs regarding voltage. For Vcc it lists the max as 1.45V. For VID it list 1.3625V. Which of these is regarding the processor core voltage?

[citronym]

I personally am not comfortable running a wolfdale past 1.367v for any prolonged periods. 4ghz on your piece is a modest overclock. Try creeping your clock up a little more (I would keep it under 1.4v loaded after droop, wont hurt it for testing purposes) and see if you start needing exponentially more voltage for each increment. That is a sign that you've reached the ceiling on your current configuration.

[redduc900]

1.45v is the "Absolute Maximum" Vcc (Core Voltage) as specified in the Intel Datasheet for the 8000 series 45nm chips (section 2.6.1)...

Absolute Maximum and Minimum Ratings
http://download.intel.com/design/pro...hts/318732.pdf

... and Intel "recommends" 1.3625v (for longevity) as the maximum core voltage.

[rge]

I can tell you how mine did running both voltages. When i first got my E8400 last year, it was 15hr small ffts prime stable at 4 ghz and 1.29V. I benched it once for 15 mins at 1.5V, then ran it at 1.36 cpuz and 4.2ghz without problem for 6 months. After 6 months checked prime, still stable 15+hrs.

Then, bored with that, ran it 4 months at 1.425bios, 1.39cpuz 4.3 and then started getting random crashes...checked orthos and crashed repeatedly in minutes. Then needed 1.36 for 4 ghz stable 15hrs (.065v more, so degraded quite a bit). That was stable even up until couple weeks ago with testing 60 runs linpack. Then over past week benched at 1.5 cpuz for about 3 hours. Checked stability after, and had to increase one more notch to be stable.

On my chip, 24/7 at 1.36 was ok, 24/7 at 1.39 was not and benching for 3 hours made me lose 1 notch of stability.

Not that I care, I will be upgrading to nehalem or buying an E8600 to play with soon...but if you are trying to make the chip last, I would stay under 1.36ish like citronym.

Though others have claimed to run theirs at higher voltages without issue and chips vary...that is my experience with my chip.

[MasterG]

That's probably sound advice. I'm only running air too so I probably shouldn't push my luck. Even at my current OC, I'm still only maxing out at about 49C. Not too shabby for air cooling.

[OldSkool]

Originally Posted by rge

I can tell you how mine did running both voltages. When i first got my E8400 last year, it was 15hr small ffts prime stable at 4 ghz and 1.29V. I benched it once for 15 mins at 1.5V, then ran it at 1.36 cpuz and 4.2ghz without problem for 6 months. After 6 months checked prime, still stable 15+hrs.

Then, bored with that, ran it 4 months at 1.425bios, 1.39cpuz 4.3 and then started getting random crashes...checked orthos and crashed repeatedly in minutes. Then needed 1.36 for 4 ghz stable 15hrs (.065v more, so degraded quite a bit). That was stable even up until couple weeks ago with testing 60 runs linpack. Then over past week benched at 1.5 cpuz for about 3 hours. Checked stability after, and had to increase one more notch to be stable.

On my chip, 24/7 at 1.36 was ok, 24/7 at 1.39 was not and benching for 3 hours made me lose 1 notch of stability.

Not that I care, I will be upgrading to nehalem or buying an E8600 to play with soon...but if you are trying to make the chip last, I would stay under 1.36ish like citronym.

Though others have claimed to run theirs at higher voltages without issue and chips vary...that is my experience with my chip.

+1 QFT 1.36v is the absolute long term max for 45nm.

[Jotosuds]

Originally Posted by OldSkool

+1 QFT 1.36v is the absolute long term max for 45nm.

is that voltage in Vdrop (in CPU-Z) or Vdroop (CPU-Z w/Prime95)?

[citronym]

Originally Posted by Jotosuds

is that voltage in Vdrop (in CPU-Z) or Vdroop (CPU-Z w/Prime95)?

The latter.

[OldSkool]

Originally Posted by citronym

The latter.

No, that's voltage after vdroop without stressing. The highest voltage voltage your chip will see during operation. That's what you want to go by.

[Jotosuds]

Originally Posted by OldSkool

No, that's voltage after vdroop without stressing. The highest voltage voltage your chip will see during operation. That's what you want to go by.

so the vDrop voltage in CPU-Z will be highest immediately after stress-testing for a while?

hypothetically let's say at bootup it's reading 1.360v, during 30 minutes of P95 it's at 1.260v, and then right after P95 ends let's say it's at 1.371v

so that would be too high according to what you're saying right?

[OldSkool]

Well, here's the thing, when you are stressing, you are running at a higher stress than most computers will ever see. 1.36 as you illustrated is a 24/7 voltage, whereas, the 1.37 is simply a quick peak after stressing. Under that pretense, it really only matters what the 24/7 voltage is non-stressed. Basically, a quick peak to 1.37 for like a second won't matter, it's when you are exposing the processor to a constant flow of stressful voltage, that you need to be concerned.

Think of it like a stream running over a rockbed, the more constant flow of water, the more the errosion. Every processor will eventually die, just like the rock will eventually errode, how fast that happens is directly related to the amount of flow. Maybe that's a corny analogy, but it works.

[MasterG]

So here's another question. How far can you push chipset voltages without it being a problem. I'm now running a Biostar TPower I45.