Why don't Intel "K" chips have embedded VIDs for overclocked multipliers too?

[VirtualLarry]

It would take all of the guessing out of voltage-control for overclocking, and make it truly easy for "Grandma" to overclock.

 

[Ferzerp]

They do provide VIDs for multipliers above stock...... Where do you think the voltages for the "auto" setting on motherboards is coming from?

 

[VirtualLarry]

They do provide VIDs for multipliers above stock...... Where do you think the voltages for the "auto" setting on motherboards is coming from?

I always thought that there was a mobo "Algorithm" in the BIOS, based on mfg testing of chips, as to what most chips would need at that speed.

If Intel individually bins K chips, including VIDs for multis above stock, then there would be no need to mess with any voltage settings at all, just leave it completely on auto.

Since that's not the common wisdom around here, I concluded that Intel does not program exact VIDs for overclocked multis.

 

[Idontcare]

If Intel individually bins K chips, including VIDs for multis above stock, then there would be no need to mess with any voltage settings at all, just leave it completely on auto.

Since that's not the common wisdom around here, I concluded that Intel does not program exact VIDs for overclocked multis.

Of course the VIDs exist and you do not have to do anything with the voltage when overclocking...but the reason that is not conventional wisdom around here is because (1) the default VID's are based on an assumed thermal profile that is based on the stock HSF whereas most OC'ers are using 3rd party HSF's and running their CPUs well below TJmax at any given clockspeed, and (2) needlessly turning our CPUs into mini-furnaces by allowing way too much over-volting seems silly.

Pretty much everyone tweaks their voltage settings when overclocking so they can minimize operating temperature and power consumption. We can do this because of our investments into superior HSF as well as the otherwise limited suite of stress testers and applications we use to determine CPU stability at any given voltage/temperature/clockspeed.

But yeah, if you don't care about operating temperatures or power-consumption then you can just leave the system to Auto control the voltage and the VID's will take care of it.

 

[VirtualLarry]

Are you suggesting that Intel tests each CPU, and programs all of the multis up to 57x or whatever they go to, with a VID that will allow the chip to OC using the stock cooler?

 

[Kenmitch]

Are you suggesting that Intel tests each CPU, and programs all of the multis up to 57x or whatever they go to, with a VID that will allow the chip to OC using the stock cooler?

I don't think they do it that way. The VID stops at 1.5v's or so but the chip may not even make it to 50x. At 48x my 2550k throttled with 1.38v's with the stock cooler.

 

[Idontcare]

Are you suggesting that Intel tests each CPU, and programs all of the multis up to 57x or whatever they go to, with a VID that will allow the chip to OC using the stock cooler?

Yes they do. But the chip will thermal throttle from hitting TJmax long before it gets to the 57x multiplier.

 

[rgallant]

"make it truly easy for "Grandma" to overclock.'

to do list for monday

-prune rose's
-pick up poi-dent
-clean cat litter
-bake carrot cookie's
-overclock cpu to the max.

 

[VirtualLarry]

Yes they do. But the chip will thermal throttle from hitting TJmax long before it gets to the 57x multiplier.

I still find that hard to believe. All of that shmoo plot testing, for each individual CPU, that's a lot of test time taken up by the 'K' chips. And they don't sell for THAT much of a price premium.

 

[Hulk]

I still find that hard to believe. All of that shmoo plot testing, for each individual CPU, that's a lot of test time taken up by the 'K' chips. And they don't sell for THAT much of a price premium.

We know they bin their chips so we know they must do some pretty comprehensive frequency vs. voltage testing. I'm sure they have it down so well by now that each chip get like a 2 seconds test and that tell them everything they need to know about that particular piece of silicon.
And all of that info is embedded in the chip ... somewhere and in some cryptic form that we'll never know. It's kind of like the Holy Grail of overclocking knowledge.
Even if you don't know the exact voltage at various frequencies, because that would depend on cooling, ambient temperature, motherboard, etc... You WOULD know which chips are the cherries and which ones are the rotten apples.

 

[pm]

I still find that hard to believe. All of that shmoo plot testing, for each individual CPU, that's a lot of test time taken up by the 'K' chips. And they don't sell for THAT much of a price premium.

Not that I'm agreeing with IDC (because I honestly don't know) but you wouldn't need to shmoo. You could just pick points. If you already knew what the shmoo was going to look like - based on measurements that are fast like idle leakage, and other known parametric testing values, you'd pretty much already know what the shmoos look like so you just test points to confirm.

 

[Ferzerp]

Not that I'm agreeing with IDC (because I honestly don't know) but you wouldn't need to shmoo. You could just pick points. If you already knew what the shmoo was going to look like - based on measurements that are fast like idle leakage, and other known parametric testing values, you'd pretty much already know what the shmoos look like so you just test points to confirm.

Especially when you have 0 reason to validate stability outside of stock speeds, so you can interpolate or extrapolate as you will, knowing that no one is going to really be using the VID beyond a few bins above stock anyway.

At the core of it, the OP is really just asking for higher binned procs because if they take the time to validate it at 4.5Ghz for example, they'd be idiots to sell it as a 3.5Ghz cpu (but he only wants to pay for a 3.5Ghz cpu).

 

[VirtualLarry]

I'm not asking for higher-speed-rated procs, I'm curious about whether Intel bothers to do verification testing for any multiplier above the stock turbo multis for the 'K' chips, in order to set the VID for those multis.

I mean, that would be great for overclockers, to be able to set a multi and forget it, with the CPU and the mobo VRMs setting the proper voltage without having to tweak anything.

But I guess I just find that hard to believe, because it would seem to take too much time on Intel's multi-million $$$ test equipment, and test time is a factor in the cost of the CPUs, and the retail cost of 'K' chips isn't that much more than their non-'K' variety (especially considering that they have the validated higher-end IGP as well).

But I guess pm is suggesting that they do validation testing, and then test a few other data points, and sort of extrapolate the VIDs for all the multis from there?

 

[FalseChristian]

"make it truly easy for "Grandma" to overclock.'

to do list for monday

-prune rose's
-pick up poi-dent
-clean cat litter
-bake carrot cookie's
-overclock cpu to the max.

LOL! I like your sense of humour!