PDA

View Full Version : [Speculation] overclocking and vcore



Doug2507
02-04-2017, 06:04 AM
Got a tech question for you Raja, thought maybe asking Elmor and Nick about it also.

vCore...

VIII series across the range are reasonably consistent with vcore required for 'x' freq, however compared to another board in particular (small/yellow) chips are requiring roughly 50mv on average more than that other board. Obviously taking reading from the board and not SW (cap of vCore rail). Also noted was the other board gave consistently higher temp readings for same applied vCore. Only a couple of degrees when running lowest vCore for required clock but when the vCore delta is added to the other board to match voltages temps are considerably higher. (adding the 50mv delta to make both the same vCore regardless of freq. stability)

Can you explain the theory behind how this is possible? Is this due to the other board having higher leakage and using more amps?

And on a side note, a chip I tested ran R15 5g@1.25v LLC6 on Hero Alpha. That same chip ran R15 5g@1.23 LLC5 on an M8E. Ambient the same for both. Any reason?

Raja@ASUS
02-04-2017, 08:15 AM
Got a tech question for you Raja, thought maybe asking Elmor and Nick about it also.

vCore...

VIII series across the range are reasonably consistent with vcore required for 'x' freq, however compared to another board in particular (small/yellow) chips are requiring roughly 50mv on average more than that other board. Obviously taking reading from the board and not SW (cap of vCore rail). Also noted was the other board gave consistently higher temp readings for same applied vCore. Only a couple of degrees when running lowest vCore for required clock but when the vCore delta is added to the other board to match voltages temps are considerably higher. (adding the 50mv delta to make both the same vCore regardless of freq. stability)

Can you explain the theory behind how this is possible? Is this due to the other board having higher leakage and using more amps?

And on a side note, a chip I tested ran R15 5g@1.25v LLC6 on Hero Alpha. That same chip ran R15 5g@1.23 LLC5 on an M8E. Ambient the same for both. Any reason?



I'd need to test this myself in order to comment. Too many variables.

Doug2507
02-04-2017, 08:32 AM
Would be good if you could. Seems to be a bit unexplained within the community. Pretty sure you know what the 'other board' is....

Look forward to hearing back from you on this one.

Raja@ASUS
02-04-2017, 08:38 AM
I dont have the other board nor do I intend to get it, sorry. Usually find the explanations are quite simple if one knows what to look for, though.

Arne Saknussemm
02-04-2017, 08:52 AM
less resistance for same voltage = higher current

Maybe team yellow are using lower resistivity material...

Same resistivity shorter fatter traces?

Maybe sample size of "the community" is statistically insignificant? :o

Doug2507
02-04-2017, 09:17 AM
Feel free to pm me as it sounds like you have an idea why this happens. Just looking for theoretical knowledge...

Community being those who OC. Big discussion about it on one of ze German forums. ;)

Even if that were true arne it doesn't explain a 50mv difference between them for same freq...

Arne Saknussemm
02-04-2017, 10:22 AM
it sounds like you have an idea why this happens.

Sorry...didn't mean to give that impression LOL

Just remember my physics classes for voltage/current/resistance V=I.R

But there are so many variables from power getting to board to entering CPU that I wouldn't really speculate on why this is happening...assuming it's true.

Just saying that in the simplest scenario...if you are getting more heat at lower voltage on yellow then I assume more current is flowing at that lower voltage which would mean resistance is lower.

If resistance is lower it might be the material used is higher quality...less resistivity....or if same materials more or less then physically connections are shorter or connections thicker or...

But this is assuming a simple circuit and the VRM/CPU is not that a simple circuit...

And all this before my second cup of tea ;)

Raja will probably be spitting his second cup of tea laughing :o

Doug2507
02-04-2017, 10:42 AM
Haha, I'm on my 3rd!

Heat at same freq using minimum vcore required, so roughly 50mv delta, is near enough the same. Just a degree or two different. If you set them both at the same vcore on dmm then yellow is significantly higher temp.

That's what makes me question the true voltage on each board.

Plus in a lab environment x cpu is going to require y voltage/current/power etc etc to runderstand at z freq. Maybe that would change due to the quality of the supply? Ripple etc etc?

There's massive speculation in the oc community about the difference in vcore but neither manufacturer seems willing to post any theory / reasoning behind it. It is what it is but an explanation from the educated as to why would be nice.

Doug2507
02-04-2017, 11:42 AM
I'll simplify...

Test used, frequency, ambient and cooling same for both. Vcore taken directly from board with DMM. (Software shows the same fyi)

Red team - 1.31vcore 60deg

Yellow team - 1.25vcore 60deg

Theoretically possible or not? If so, how?

Raja@ASUS
02-04-2017, 12:02 PM
The boards will still clock the cores to the same freq, so it doesnt really matter. Not going to say anything more about it, as it isnt important.

Arne Saknussemm
02-04-2017, 12:08 PM
Where are those voltage measurements taken Doug...just out of interest...is that the voltage after the VRM that's going to the CPU?...and I guess I'd have to ask is the read point in a physically similar position in the circuit on both boards?

If temps are exactly the same...assuming that can be taken to mean that for same load and same power consumption...yellow is using less volts because it is sending more current.

Volts/current/resistance so yellow may have less resistance between VRM and CPU if that is the read point...

I think....:confused:....

I feel like I'm in my fifth form physics exam LOL

EDIT: The main point is that right....anyone saying yellow is clocking same chip higher?

Doug2507
02-04-2017, 12:30 PM
Like for like vcore yes there's 100mhz difference which for ambient oc it makes a difference.

Yes, after vrm. Could mesure from pin but woupd need to ready whitepaper to check which one.

Haha, ok raja. So basically no comment. Np.

Raja@ASUS
02-04-2017, 03:42 PM
There are loads of VCC pins. No need to check the land diagram as you wouldn't actually measure this from a "pin" anyway.

In any case, as I said above, the final OC will be the same if the right person is performing the testing. ;)

Telstar
02-04-2017, 06:44 PM
A friend of mine just got the Apex and the cpu he is testing needs the same vcore than the yellow z170 we all know about.

Doug2507
02-04-2017, 09:19 PM
And I still can't get a direct answer as to why the difference. Simple question, is it theoretically possible? Yes or no?