q6600: lower VID = better?

gurusan · 15 Mar 2008 at 09:36

From my experience VID has not had much of a correlation to how good a chip clocks.

I had a 1.325 VID q6600 B3 and it did 3.84ghz with waterchilling, 4.05ghz bootable.

and had a Q6600 G0 with 1.265VID that did 3.6ghz stable, 4.05ghz benchable.

had an e6300 with 1.35vid that did 3.22ghz @ 1.35vcore.

currently have an e6300 with 1.325vid that does 3.3ghz @ 1.35vcore

Justintime · 15 Mar 2008 at 13:02

Heres another crackpot theory

The VID might be set in accordance with how random samples from the batch perform, so even though a chip might be capable of say 2.4Ghz at 1v, going by average they might set it to 1.325v just to ensure all cpus of the batch run stably. The lower VID cpus might be from better overall batches but that does not mean higher VID cpus as a rule will not clock better, overclocking can be pleasantly random at times

fornowagain · 15 Mar 2008 at 13:28

fornowagain said:
One problem though, a lower default Vcore at those higher voltages will be using more power than one that started out higher.

hernaldo said:
Source?

I've never heard anything that would suggest there's any truth at all in that?

cheers

Semiconductor power P, P=kLV²F

Where:
L = Load (from software)
V = Core voltage
F = Chip frequency
k = Empirical constant for the chip.
Pd = Total cpu power
Ps = Stock Power Output

Therefore delta P, i.e. change in power. The constants cancel.

Pd = Ps x (F2/F1) x (V2/V1)²

or

OCed Watts = Ps x (OCed Mhz / Default Mhz) x (OCed Vcore / Default Vcore)²

Lower the default voltage at the same working voltage uses more power from the vcore delta squared.

Have a go, online version. You may be surprised just how much differance the vcore delta makes to the wattage.

http://newstuff.orcon.net.nz/wCalc.html

Devious · 15 Mar 2008 at 13:32

So there !

muon · 15 Mar 2008 at 17:59

fornowagain said:
Semiconductor power P, P=kLV²F

Where:
L = Load (from software)
V = Core voltage
F = Chip frequency
k = Empirical constant for the chip.
Pd = Total cpu power
Ps = Stock Power Output

Therefore delta P, i.e. change in power. The constants cancel.

Pd = Ps x (F2/F1) x (V2/V1)²

or

OCed Watts = Ps x (OCed Mhz / Default Mhz) x (OCed Vcore / Default Vcore)²

Lower the default voltage at the same working voltage uses more power from the vcore delta squared.

Have a go, online version. You may be surprised just how much differance the vcore delta makes to the wattage.

http://newstuff.orcon.net.nz/wCalc.html

But the stock power output for the lower VID cpu will obviously be lower. So this is all irrelevant. The power consumption between processors at the same voltage and frequency should be equal plus an error term (which should be distributed under a normal distribution). Unless, the error term is somehow related to the VID (some kind of selection bias destroying the assumption of a normal distribution), on average, power consumption will be identical.

fornowagain · 15 Mar 2008 at 19:08

Thats the point they are different, maximum VID is factory set to determine the thermal profile, no two are the same. How this relates to the criteria for actual power vs published TDP is more than a little vague. Of course there must be a minor variation in stock power, trouble is consumed power is far greater from vcore delta, simply because of the squared term. To pick more holes in it, what's missing is any reference to the static power or leakage, should really be superimposed and change with the curves for frequency and voltage.

Intel said:
Each processor is programmed with a maximum valid voltage identification value (VID),which is set at manufacturing and can not be altered. Individual maximum VID values are calibrated during manufacturing such that two processors at the same frequency may have different settings within the VID range.

The TDP is not the maximum power that the processor can dissipate.
This table shows the maximum TDP for a given frequency range. Individual processors
may have a lower TDP. Therefore, the maximum TC will vary depending on the TDP of the
individual processor.

ebilcake · 17 Mar 2008 at 14:30

It can help but not always, my Q6600 will do 3.6GH at 1.375V (vid of 1.225) but loads at over 70C with good air cooler and ariflow, yet one with much higher vid may load almost 10C lower at the same speed with higher voltage.

I think it's more of a benefit to users who watercool.

DavidB · 17 Mar 2008 at 16:00

You are on the edge there with regards to top end VID mate! Intel does not recommend anything above 1.5V! You risk damaging your cores!

I think your confusing vcore and vid mate. Also, Intel dont recommend overclocking at all , hense its not covered by your warranty.

fornowagain · 17 Mar 2008 at 16:15

Quite, vcore is always (by design) lower than VID. The upper voltage they quote 1.55v for 65nm is for transient spikes and even they say that's the voltage it can survive 'with' degradation to its lifespan.

V3g3tto · 17 Mar 2008 at 16:15

My VID is 1.325v, here are the results after CPU lapping and now its bedded in (Note: I found the longer the CPU is bedded in, the less voltage is required,makes no sense to me, but is fact with my CPU):

Current: 3375mhz (375x9)@ 1.45v
Would reach 3.55Ghz, haven't tried pushing it recently though, but i could get it near 3.6Ghz @ ~1.5v
Idle Temp: 33C (mild-cool room temp)
Prime Small FFT's temp: 69C

VID is directly linked to how your CPU will overclock, fact, lets end the discussion there lol. To get higher speeds, you need more voltage, overvolting anything is dangerous. More voltage = more heat.
Only thing stopping me going past 1.5v is cooling. But it ain't worth it.

wuyanxu · 17 Mar 2008 at 18:45

so given this chance, you guys are saying i should swap with my friend (who have VID of 1.2125v) when he sells his q6600.

what im concerned about is heat, you guys said lower VID will mean more current. and heat in any given circuit is R*I^2, is this means more current is worse than more voltage?

Devious · 17 Mar 2008 at 18:54

What vcore do you require to get 3.9Ghz with your current Q6600 stable ?

wuyanxu · 17 Mar 2008 at 19:35

not stable, i need 1.8v which drops down to something more than 1.6v, it's not even able to complete 3Dmark06, and i didn't dare running Prime95. only ran SuperPi on it.

would getting the lower VID give me more chance in reaching 4Ghz with possibly lower temperature?
currently running 3.4Ghz or 3.1Ghz (depends on how i feel like) i'd like to run 3.6Ghz 24/7. would lower VID help me do that?

Devious · 17 Mar 2008 at 19:44

eek! thats crazy high vore for air or even water.

My chip has the same VID of 1.2125 and im running it stable @ 3.8Ghz 24/7 with just 1.376v actual. Your Cooling should also be able to cope with that. My temps are 45c under prime95 small FFT, you may be in the 50's tops, im sure it will be way better than the chip you have.

wuyanxu · 17 Mar 2008 at 19:57

Devious said:
eek! thats crazy high vore for air or even water.

My chip has the same VID of 1.2125 and im running it stable @ 3.8Ghz 24/7 with just 1.376v actual. Your Cooling should also be able to cope with that. My temps are 45c under prime95 small FFT, you may be in the 50's tops, im sure it will be way better than the chip you have.

got it. thanks!

lol, i know it's crazy high voltage, that's why i only had 1 run at it, and didn't dare run Prime on all 4 cores, if i did, the heatsink might melt!

IAmATeaf · 17 Mar 2008 at 22:03

wuyanxu said:
got it. thanks!

lol, i know it's crazy high voltage, that's why i only had 1 run at it, and didn't dare run Prime on all 4 cores, if i did, the heatsink might melt!

So the real question should be will a lower VID allow me to get past 3.4Ghz shouldn't it?

My VID is 1.2625, I run my Q6600 at 1.3v in the BIOS to get to 3.2GHz prime stable.

wuyanxu · 17 Mar 2008 at 22:26

i can run at 3.6Ghz Prime stable alright, but that would require me to turn top and rear fan to Medium, or to take the side panel off. otherwise it will go over 70c during Prime small FFT.

IAmATeaf · 17 Mar 2008 at 23:15

wuyanxu said:
i can run at 3.6Ghz Prime stable alright, but that would require me to turn top and rear fan to Medium, or to take the side panel off. otherwise it will go over 70c during Prime small FFT.

Well running my quad at 1.3 and 3.2Ghz it gets to around 63 under prime, that's with all my case fans set to the lowest, in fact the fans are so low that if I take the side panel off the temps on the core drop by around 8-10 degrees or so.

V3g3tto · 18 Mar 2008 at 09:40

70C under prime95 is fine, it's pretty much the max. core temp you could sustain for long periods safe. But that's irrelevent as no commercial software would ever push your CPU that hard.

Mine is 33C idle, 69C Small FFT's, mainly due to me having a HUGE tempertaure difference between cores, you're looking at upto 12C difference between core 0 and core 2.

Cyber-Mav · 18 Mar 2008 at 10:35

my [email protected] using stock voltage of 1.2850v gets up to 60c in temps under snm using a freezer 7 cooler.
under prime small fft the hottest core peaks at 52c.

im guessing the lower the vid the better the clocks may be? i have a vid of 1.2850 so i guess the guys with a lower vid will either have lower temps or the potential to clock higher on the core.