***Official***Z68X/P67A-UD7 Club/Discussion/Info**

[Falkentyne]

I wouldn't worry about that post, and LLC on the UD5 should be different as it was for B2 boards.

What do you want to see? The thing is that my temps are going to be much different, and temps and power draw especially monitored by software like that is prob going to be much different machine to machine. I can measure actual power draw fromt eh 8 pin connector through some device I bought, but unless you have that too it wouldn't matter.

You are asking if LLC level1 with teh same measurement on the DMM as LLC Level2 under load will be the same?

Honestly it you are staying under 1.5v from the trends of the DMM test I did you are going to be ok, i wouldn't worry about degradation over a months or a few months period. I don't know what you want me to do exactly? I am a bit confused lol


Yea i just saw that, i guess they removed it for P67, no idea why

Hi Sin,
just got up, ok what I meant was, if you can check the temps (and power draw with realtemp) with LLC2, with the vcore value as measured by the DMM (example: 4 ghz @ 1.35v hardware measured load DMM), and compare it with the temps and power draw (with realtemp) as measured with the vcore from software with LLC1 (4 ghz @ 1.35v software measured load hwinfo32; that would probably require 1.4v BIOS For LLC1 so the software reports 1.35v at load, and 1.25v BIOS for LLC2, so the DMM reports 1.35v at load). All under small fft prime95. I know that sounds odd and crazy, but I was just curious about the results.

So basically what I meant was, i wanted to see if the software target load voltage with LLC1, matching the hardware target load voltage with LLC2, gave the same temperatures and the same power draw.

The exact reason is: I've now found two posts that are making me think something funky is going on.

The person who did the DMM tests with the UD4 board, told me that if he got (example: i dont remember the FULL details), 1.35v load, hardware measured LLC, he was stable, however if he got hardware measured 1.35v load LLC OFF (standard), he BSOD 124....

The measured hardware load voltage is exactly the same, AND LLC off requires a higher voltage set in BIOS to compensate for the droop, yet he BSOD'd only with LLC off.

I hope this makes sense....

But yeah, the power draw of the LLC1 test, and the temps, should be higher, than the LLC2, since the software report of LLC1 vcore would be the same as the hardware report of LLC2 (which is what I want), but the hardware report of LLC1 would be much higher

[sin0822]

why would you want to compare two different LLC settings using two different measuring methods?

So you want me to measure LLC level2 with a DMM, and LLC level1 with software, and you want me to match the software to the DMM?

[Falkentyne]

Yep exactly ^^
(since we already know what the hardware values will be for the wrong software readings). Just want to see if temps and power draw match up.

[sin0822]

ok ill try it out for you. But still your LLC might not be the same as the UD7, i remember B2 boards they were totally different.

[Falkentyne]

Yea, I know. Just want to see what you'll get for your temps. I'm sort of shocked that your LLC standard is showing a more than 0.12v droop from sensors (1.23v->1.1v), while on mine the still incorrect sensors show 1.25v->1.164v. Plus the strange UD4 DMM and stability report....LLC and standard (after vdroop) are not the same stability even with the DMM showing same vcore; LLC more stable)

I'll ask Bullet1, too.
I hope I can get a DMM soon. Do you have to have the board outside the case to take a reading?

[andressergio]

Yea, I know. Just want to see what you'll get for your temps. I'm sort of shocked that your LLC standard is showing a more than 0.12v droop from sensors (1.23v->1.1v), while on mine the still incorrect sensors show 1.25v->1.164v. Plus the strange UD4 DMM and stability report....LLC and standard (after vdroop) are not the same stability even with the DMM showing same vcore; LLC more stable)

I'll ask Bullet1, too.
I hope I can get a DMM soon. Do you have to have the board outside the case to take a reading?

im on P67A UD7 B2 tell me what you want me to measure with vdimm and i do it mate

*sin link me the points plz

[Falkentyne]

I think you need a B3 board or a crossflashed b3 bios first, andre, because LLC was changed on the B3. (at least it was for UD5 (massive change) and UD4).

But first you would need to know the difference between software load vcore and DMM measured vcore.

Now, lets pretend that your board did 1.2v BIOS LLC2->1.3v DMM measured at load (software showing 1.23v), and 1.35v BIOS LLC1->1.3v SOFTWARE at load, and 1.35v measured at load.

I'd want to see if the realtemp current power draw, and the temps for LLC2 1.3v hardware load, are the same, or lower, than LLC1 1.3v softare load (1.35v real). The LLC1 should have higher current draw and temps, than the LLC2, since the DMM is showing the sensors to be underreporting.

The temps and power draw should NOT be the same.

[the_real_7]

Ok guys pulled out my good ole fluke multimeter and followed sins guide for the voltage points , now my setup is B3 and bios F3H , this is not my normal settings as im in 2133 7 9 7 21 benching mode and my 2600k is taking way more power than when its memory is at 1600mhz 6 6 6 18 , and all my system require to be prime stable , which takes alot of juice

btw would like to thank sin0822 for his hard help in the forum

4.8 ghz 2133mhz 7 9 7 21 timings LLC1 setting
BIOS IDLE LOAD
VCORE 1.480 1.467 1.432

VDIMM 1.68 1.681 1.681

CPU PLL 1.800Auto 3.306 3.295

VTT 1.160 1.151 1.151

This is why our board is kicks ass

[andressergio]

I think you need a B3 board or a crossflashed b3 bios first, andre, because LLC was changed on the B3. (at least it was for UD5 (massive change) and UD4).

But first you would need to know the difference between software load vcore and DMM measured vcore.

Now, lets pretend that your board did 1.2v BIOS LLC2->1.3v DMM measured at load (software showing 1.23v), and 1.35v BIOS LLC1->1.3v SOFTWARE at load, and 1.35v measured at load.

I'd want to see if the realtemp current power draw, and the temps for LLC2 1.3v hardware load, are the same, or lower, than LLC1 1.3v softare load (1.35v real). The LLC1 should have higher current draw and temps, than the LLC2, since the DMM is showing the sensors to be underreporting.

The temps and power draw should NOT be the same.

sorry i dont have a B3 im in south america and cant RMA it, im using bios F8f wich is perfect for it wont use the B3 version crossflashed bios unless i need them

will try to read point as are not easy to access on ma setup

cheers
Sergio

[sin0822]

the choke on the top of the board works pretty well, but i read off an MLCC cap on teh back of the board the difference is minimal, but i soldered a wire to the capacitor and attached it to a molex connector. Any leg of any choke facing the CPU socket will suffice.

[Falkentyne]

Ok guys pulled out my good ole fluke multimeter and followed sins guide for the voltage points , now my setup is B3 and bios F3H , this is not my normal settings as im in 2133 7 9 7 21 benching mode and my 2600k is taking way more power than when its memory is at 1600mhz 6 6 6 18 , and all my system require to be prime stable , which takes alot of juice

btw would like to thank sin0822 for his hard help in the forum

4.8 ghz 2133mhz 7 9 7 21 timings LLC1 setting
BIOS IDLE LOAD
VCORE 1.480 1.467 1.432

VDIMM 1.68 1.681 1.681

CPU PLL 1.800Auto 3.306 3.295

VTT 1.160 1.151 1.151

This is why our board is kicks ass

Real_7,
Why is your load voltage so low?
you're using LLC1, right?

Sin's load voltage at LLC1 was almost the same as the BIOS setting (about 0.01v higher), but your voltage is -0.05v below the BIOS voltage.

Your board is showing 1.432 with the DMM for load voltage with LLC1? Shouldn't it be showing 1.481-1.490 ?

That result looks like LLC is disabled because sin's LLC disabled result was like that.

Are either of you guys using a B2 version of UD7 crossflashed with B3 bios?
I feel like collapsing right now because I'm so confused

[Falkentyne]

@Sin:

I'm confused as to why Real_7's vcore reading with his DIMM is so much different than yours (he has vdroop with LLC1 but you don't).

Are you using a B2 UD7 cross-flashed with B3 bios? Or actual B3 board?

[andressergio]

Real_7,
Why is your load voltage so low?
you're using LLC1, right?

Sin's load voltage at LLC1 was almost the same as the BIOS setting (about 0.01v higher), but your voltage is -0.05v below the BIOS voltage.

Your board is showing 1.432 with the DMM for load voltage with LLC1? Shouldn't it be showing 1.481-1.490 ?

That result looks like LLC is disabled because sin's LLC disabled result was like that.

Are either of you guys using a B2 version of UD7 crossflashed with B3 bios?
I feel like collapsing right now because I'm so confused

using LLC1 will never do that load mate thats LLC2 land...

[sin0822]

no i do have vdroop with level1, i set 1.5v and it puts out 1.52v and then drops to like 1.509v load.

LLC level1 droops under load. I don't know what voltage he set, i think he is showing what BIOS voltage reads.

now that i think of it a PSu that is stressed could make a difference, but theoretically it shouldn't. I was using my sparkle 1kw unit without the GPUs. It also depends what you load with.

[Falkentyne]

I sent him a PM, hopefully we'll find out what's up.
His vcore should have been between 1.481 to 1.490 measured with the DMM.
The 1.432 looks like a software reading, or it looks like LLC just wasn't working on his board.

If both you guys are using B3 stepping boards, and Gigabyte didn't pull a change a droop resistor at the last minute, we'll get to the bottom of this

This is important to me cuz I'm OCD and afraid of degrading chips....

Someone said that crossflashing stops certain things from working (PLL Override ? LLC?) until you reflash with the non crossflashed version of the same BIOS.

[sin0822]

idk man, i checked the board, and all the resistors around the PWM are the same, they didn't change em, i specifically looked for it.

The thing is that this LLC level1 mirrors the other guy's results on the UD4-B3 as well. LLC level2 mirrors LLC level2 on the B2 UD7.

[the_real_7]

no i do have vdroop with level1, i set 1.5v and it puts out 1.52v and then drops to like 1.509v load.

LLC level1 droops under load. I don't know what voltage he set, i think he is showing what BIOS voltage reads.

now that i think of it a PSu that is stressed could make a difference, but theoretically it shouldn't. I was using my sparkle 1kw unit without the GPUs. It also depends what you load with.

sent him a PM, hopefully we'll find out what's up.
His vcore should have been between 1.481 to 1.490 measured with the DMM.
The 1.432 looks like a software reading, or it looks like LLC just wasn't working on his board.

If both you guys are using B3 stepping boards, and Gigabyte didn't pull a change a droop resistor at the last minute, we'll get to the bottom of this

This is important to me cuz I'm OCD and afraid of degrading chips....

Someone said that crossflashing stops certain things from working (PLL Override ? LLC?) until you reflash with the non crossflashed version of the same BIOS.

Hey guy's saws your forum and I got his pm . . . yes Im on lc1 and that corrected after 1.480 set in bios I get a 0.050 droop to 1.432 on prime load on DMM , but ive always gotten that on this and my B2 board . . I thought sins check on the B3 had change this droop im getting , but it still remains , and this is not only me original Sergio on his chip before had gotten same with his chip , my meter is of high quality so im sure it is not wrong . . .I tried 3 different voltage with loads , and repeated same thing no matter of vtt . . is it degrading well I hope not, I have need more vcore to run my 2133

[sin0822]

what about your PSU?

Try reading off the MLCC capacitor on the back that is the only difference.

Ill check with my other CPU tomorrow and see if its any different. 0.05 is a lot to be drooping, it just doesn't sound right. lol

Edit you are drooping 0.035, i don't count from where you set in BIOS, i count from idle to load, there is no point because it sets a definite offset from what you set. Vdrop and vdroop are different. I don't count vdrop because it has no performance impact if you account for it.

Anyways I hope your CPu is not degrading you were pushing a lot of voltage through it, so it might have been, i saw early degradation with these CPus above 1.5v that is why i tell people not to use it.

LLC level1 has definitely changed from B2 BIOS F8X to B3 BIOS F3H.

[Falkentyne]

That's why I asked Sin to run that Software vs Hardware vcore comparison with temps, so we can see if something's up.

I know SOMETHING is going on.

The UD4 readout showed LLC (same as our LLC1) giving same vcore as BIOS, which was the same as Sin's conclusion.

Your readout is giving 0.05v vdroop, on a B3 stepping board.

How can two of the exact same boards give such different values for DMM measurement?

This is why you gotta also look at the temps...if the CPU is actually USING different voltage than what the DMM or sensors are saying, the DTS temperature in realtemp should show it ...

[sin0822]

voltage really means really not much man, its all about the current pull. I don't trust software to produce current and wattage numbers, i just don't lol. I will do it for you i am just so tired tonight.

[the_real_7]

what about your PSU?

Try reading off the MLCC capacitor on the back that is the only difference.

Ill check with my other CPU tomorrow and see if its any different. 0.05 is a lot to be drooping, it just doesn't sound right. lol

Edit you are drooping 0.035

Ive tested psu running like a tank on all rails plus its a hx850 a single rail powerhouse
I have no access to the back of my board sin but what are the chances my software matches what my dmm is reading cause on my b2 they where exact settings and all my other point are right on the money . . . I use pll over volt . . hate the droop but if it is correct , Im sitting in a good place . .

[Falkentyne]

Well if the cap on the backside of the board is reading differently, then that would mean that, if that voltage is higher, then that's the voltage coming from the VRM's, while what Real_7 is reading is what the CPU is actually getting (after that vcore has been drooped).

Sort of logical, actually. But won't know without a MLCC cap reading.
But the reading from the MLCC cap and the choke shouldn't be that much different; only about 0.01v.

There are only two possible explanations:
1) Gigabyte revised something on earlier or later B3's (unknown which) to adjust the vdroop without changing the PCB revision.

2) Real_7's reading is vcore after vdroop while Sin's is before vdroop.

But #2 doesn't make sense because Sin has a lot of vdroop with LLC set to standard. Yet the sensors show MUCH more vdroop...

One thing for sure, if I got a DMM, I won't have access to the backside of the board, either...

[sin0822]

Ive tested psu running like a tank on all rails plus its a hx850 a single rail powerhouse
I have no access to the back of my board sin but what are the chances my software matches what my dmm is reading cause on my b2 they where exact settings and all my other point are right on the money . . . I use pll over volt . . hate the droop but if it is correct , Im sitting in a good place . .

See i don't use CPu pLL Overvoltage, but i doubt that would make a difference. I have tested many many times, multiple times on like every BIOS with P67, and what software reads(i used easytune) it is always 100% off lol. Software hasn't matched up hardware on P67 for me ever.

That is why I did the readings.

Software matches hardware kind of when it comes to trends, but not 100%, i find that ti sways a bit different with different levels of LLC.

The MLCC cap on the back of the board is the best place to read the voltage, as the flow goes from the DrMOS to the inductor, then all the inductors to the capacitors and then the capacitors discharge, and the MLCC caps are part of those capacitors. We aren't reading from some type of feed back loop, the MLCC cap is the voltage right before it goes to the processor. All the caps on the back, well almost all of them have the same voltage reading.

Going through a capacitor shouldn't make a difference tho.

My readings are done during idle, task manager at 0% on all cores, then at load with 100% on all cores, and i wait for the voltage to stabilize after a few seconds.

[stasio]

BIOS F3 is out for all B3 boards (UD3-B3 on the way).

Edit:

Modified BIOS to flash back from B3 to B2:
http://www.xtremesystems.org/forums/...postcount=1672

BIOS F8 is out for B2.

[Z3LL]

BIOS F3 is out for all B3 boards (UD3-B3 on the way).

Edit:

Modified BIOS to flash back from B3 to B2:
http://www.xtremesystems.org/forums/...postcount=1672

hows the official f3 bios compared too the f3 beta bios for the B3 ud7?