• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

what's dangerous on an .09 chip

Paintballfreak66

Golden Member
Ok I've got my cpu at 237x11 1.55 stable and its a 3700 SD .09

now I've heard don't go past 1.55 on a .09 chip but what if I were to have a thermalright heatsink(this one) and a vantec tornado 92mm fan on top of it(the thing puts out 119CFM lol). Is it just a heat issue with going higher than 1.55 on a .09 or is there something else?
 
The 90nm process came out with the winnies at 1.4v so 1.55v represents a 10%overvolt and in most ppl minds that is safe if using decent air cooling and temps are below 60c....

Now some of the SD cores are rumored to be 1.35v but they should follow same principle as the winnies...I think 1.55v actual is a safe range....15% is safe for higher end air cooling...anything above that needs really high end exoctic cooling...
 
I agree, I see lots of folks with the DFI boards hitting Winchesters and Venices with 1.6 even 1.7v but thats just crazy IMO

And yes the SD3700+ is .09
 
ok but it is a heat issue then. I know the more power, the more heat but I just wanted to make sure that is why people say don't go higher than 1.55 on a 90nm. alrighty thanks for the help.
 
A thing called electron-migration...Basically the high voltage causes gates to eventually fail and bingo you start having errors....Now I thing it still has to do with heat persay but on a more localized level that may not be detected in the overall core temp by the themral diode....

I have never in 5 year now of ocing seen a chip fail from 10% or less vocre boost...
 
Originally posted by: Duvie
A thing called electron-migration...Basically the high voltage causes gates to eventually fail and bingo you start having errors....Now I thing it still has to do with heat persay but on a more localized level that may not be detected in the overall core temp by the themral diode....

I have never in 5 year now of ocing seen a chip fail from 10% or less vocre boost...

Doesn't electro-migration screw up the wires (power rails included)? Maybe you're thinking of gate-oxide breakdown which is directly affected by supply voltage when it exerts too strong of an electric field over the gate oxide?
 
Electromigration affects everything that has a current through it but it really doesnt make a difference when metal moves, cracks or warps (that is what electromigration is) when you have a huge solid piece of metal conducting. Now, when you have wires that are nanometers wide, even thinner and only nanometers apart, then you start having problems. Electromigration is related to the current (not tension) density on the conductor and the heat (<- which is DIRECTLY related to the tension). Like Duvie said, it's more important to think about the temperature of the different hotspots. Even if you have insane cooling, it is not applied directly to the metal conductors, so you are still limited by all the thermal conductivity of the different materials (silicon, heat spreader, thermal paste, etc). You are correct, however, gate-oxide degradation is directly related to the electric field across the gate dielectric.

Edit:

Paintball: Now, to answer your question 😀. Most people dont recommend you going over 1.55v because (if I remember my formulas correctly) at 1.55 (from 1.35) you shave off about 32% of a CPU's MEAN life--so this means your cpu will supposedly only last you around 13.5 years from the original 20ish year rating. This, of course, is just an estimate, some cpus deteriorate much faster due to different opeating conditions and some take even longer.

NOTE that I oversimplied everything, there are lots of different things that affect a cpu's lifespan, but voltage is one of the important ones, and so is heat.
 
Originally posted by: Furen
Electromigration affects everything that has a current through it but it really doesnt make a difference when metal moves, cracks or warps (that is what electromigration is) when you have a huge solid piece of metal conducting. Now, when you have wires that are nanometers wide, even thinner and only nanometers apart, then you start having problems. Electromigration is related to the current (not tension) density on the conductor and the heat (<- which is DIRECTLY related to the tension). Like Duvie said, it's more important to think about the temperature of the different hotspots. Even if you have insane cooling, it is not applied directly to the metal conductors, so you are still limited by all the thermal conductivity of the different materials (silicon, heat spreader, thermal paste, etc). You are correct, however, gate-oxide degradation is directly related to the electric field across the gate dielectric.

Edit:

Paintball: Now, to answer your question 😀. Most people dont recommend you going over 1.55v because (if I remember my formulas correctly) at 1.55 (from 1.35) you shave off about 32% of a CPU's MEAN life--so this means your cpu will supposedly only last you around 13.5 years from the original 20ish year rating. This, of course, is just an estimate, some cpus deteriorate much faster due to different opeating conditions and some take even longer.

NOTE that I oversimplied everything, there are lots of different things that affect a cpu's lifespan, but voltage is one of the important ones, and so is heat.


So you are saying that even if I was able to keep my cpu in the 40s(C) under load, just the fact that there is a higher voltage coursing through the chip is enough to harm it?
 
You can "cheat" physics by activating Cool n Quiet. Most of the time, the CPU is running at 5x multiplier and <1.25Vcore. It will only run full speed @ 1.55V to 1.6V when the CPU load approaches 100% (probably less than 10% of the time).

In summary, you can probably bump up to 1.6V knowing that the CPU will only run at that voltage about 10% of the time with C n Q enabled. If you're not running C n Q, then 1.55V would be the highest "safe" voltage.
 
Back
Top