Z170 Mobo recommendation?

[spaceholder]

I've been researching this for awhile but I can't come down on a clear winner for a Motherboard. I'm building a new gaming machine with the following in mind:

-M.2 Samsung evo 850 512GB
-Intel 6500
-AMD graphics for Freesync - most likely a cooler running RX480 design but we'll see. My GTX 650TI will hold me over for awhile. I dont believe in SLI/Crossfire, I believe in the single GPU graphics solution and will continue with that approach for this machines lifespan.

At the moment I play QuakeLive and CS:GO, though I will play Overwatch, Doom4, Quake Champions etc, so while performance isnt 100% needed now it will be soon. I don't plan on overclocking, at least for now.

The 3 important aspects of the mobo to me are:
1. Sound, ideally Realtec 1150 or some other premium audio. It's actually important for gaming as many earlier chipsets mess up positional audio. I'm done with discreet soundcards.
2. Ethernet - I'd like to go with an Intel gigabit chip this go around. My current board is a ASRock Fatal1ty Z87 Killer with the Killer E2200 which was not well reviewed by people checking up on its performance. A big part of the reason I'm rebuilding is the random DPC latency spikes. I want to have the best small packet sorting performance possible and the first gen Killer chipset doesn't seem like a good candidate. If I have to the newer Killer chips like the E2400 seem to be well reviewed for handling small packets.
3. DPC latency. The two things most affected by it are twitch FPS gaming and video playback. I cant watch a Twitch.tv stream without getting hiccups and I'm sick of it. I've done everything I can to diagnose the DPC stuff but the problem seems to be endemic to the board design/drivers available.

I'm leaning towards the Z170 chipset just for the increased performance ceiling with RAM speeds and generally more robust nature of the mobos. If need be the H series chipsets could do.

I'm not against micro ATX but I do have an Antec p183 and I care about silent computing. At that point you may as well get a mobo with extra I/O if you ask me. Also I usually run 5-6 hard drives so I have a pretty strong preference for regular boards.

Recommendations? I'd like to avoid ASRock as their cheaper boards all seem to have DPC issues and Asus has consistently screwed me over for 4 generations that I used their mobo's. Other than that I'm open to w/e.

Thanks guys

 

[WilliamM2]

I only buy Asrock and Asus, so I guess I can't help much. The Asus Z170 Pro gaming meets all your requirements though, and it's on sale at Newegg as well.

Enter code EMCELEJ28 and get it for $124 after rebate. Love mine.

http://www.newegg.com/Product/Product.aspx?Item=N82E16813132567

 

[UsandThem]

I would recommend the Gigabyte GA-Z170-UD5 that I use. It has been a great board for me, and has all your listed requirements.

The bad thing though is it is now priced almost the same amount as when it was first released ($189). I have seen it several times on Newegg for $139 - $155.

 

[lukart]

I only buy Asrock and Asus, so I guess I can't help much. The Asus Z170 Pro gaming meets all your requirements though, and it's on sale at Newegg as well.

Enter code EMCELEJ28 and get it for $124 after rebate. Love mine.

http://www.newegg.com/Product/Product.aspx?Item=N82E16813132567

Spot on, I pretty much on the same page with you.
The Asrock Extreme series are one of my favorites, they pack ton of features and ports with very decent pricing.

Besides, if you like overclock then their boards are the way to go.

 

[spaceholder]

I only buy Asrock and Asus, so I guess I can't help much. The Asus Z170 Pro gaming meets all your requirements though, and it's on sale at Newegg as well.

http://www.newegg.com/Product/Product.aspx?Item=N82E16813132567

I ended up buying a Asus Z170 Pro gaming but so far I've been plagued by DPC issues. I see spikes in the 650 to 2000 range every few minutes.

I'm gonna try disabling intel graphics, uninstalling the audio radar bs program but somehow I doubt that will fix it. I have a nasty feeling I'm going to end up buying a $300 mobo, or switching to AMD just to avoid this DPC nightmare. I dont mind rebuilding a PC, but fuck I hate updating windows 8.1

 

[UsandThem]

Curious, how exactly will buying a $300 motherboard or moving to AMD eliminate the DPC issues you currently see?

 

[BonzaiDuck]

DPC latency? there are free utilities out there which might help you reveal any driver conflicts behind that.

I look back over decades and I think the only board I ever had trouble with immediately after purchase was an Asrock Extreme 3 Z77. The box was in shrink-wrap and everything tidy, but it was the first time ever that I decided to examine the socket with a magnifier -- first time! Two or three pins were damaged, and one appeared loose and not inserted in the hole where it belonged.

I just wrapped it up and RMA'd it for a refund.

 

[spaceholder]

Curious, how exactly will buying a $300 motherboard or moving to AMD eliminate the DPC issues you currently see?

Because review websites show the $300 boards as having low DPC latency? To clarify: not ALL $300 motherboards will feature low DPC, but the lowest and most consistent numbers seem to be found in the ultra premium price segment. AMD might help because this problem is rooted in central devices and guess what? A different company makes those devices and drivers for AMD. I've also heard from one or two people that its a smaller problem in AMD land.

Thats why those things might help.

DPC latency? there are free utilities out there which might help you reveal any driver conflicts behind that.

I look back over decades and I think the only board I ever had trouble with immediately after purchase was an Asrock Extreme 3 Z77. The box was in shrink-wrap and everything tidy, but it was the first time ever that I decided to examine the socket with a magnifier -- first time! Two or three pins were damaged, and one appeared loose and not inserted in the hole where it belonged.

I just wrapped it up and RMA'd it for a refund.

I'm not talking about a defective board. I'm talking poor design that causes a problem that cant be corrected. Many reviews mention uninstalling included managment and enhancement utilities helping but not completely eliminating DPC issues.

So far unisntalling Intel's integrated drivers seems to have helped the most lowering my DPC to 80-110, but it still spikes up to 500+ from time to time. Last one was caused by HDaudio.sys .... even though I unisntalled the audio manager program while keeping the drivers.

 

[wingman04]

I don't have all those problems I have a budget Gigabyte Z170 HD3.

 

[MadOver]

Running 24/7 Z170 OC-Formula here, rock solid, overclocked obviously

 

[BonzaiDuck]

Running 24/7 Z170 OC-Formula here, rock solid, overclocked obviously

Sometimes there are shared resources on lower-tier boards, and you either have to give up something like front-panel USB, or accept limited lanes on a PCI-E slot. I'd get the free software to analyze DPC latency and get to the bottom of it. It could also be conflicting drivers.

I say this because that Pro board is likely good as any in its class.

 

[wingman04]

Sometimes there are shared resources on lower-tier boards, and you either have to give up something like front-panel USB, or accept limited lanes on a PCI-E slot. I'd get the free software to analyze DPC latency and get to the bottom of it. It could also be conflicting drivers.

I say this because that Pro board is likely good as any in its class.

Motherboard resources is mostly marking hype. PCs can only do one task at a time it is all just smoke and mirrors when it comes to sharing tasks. The PCI-E buss actually runs data in order from one device to the next.

 

[BonzaiDuck]

Motherboard resources is mostly marking hype. PCs can only do one task at a time it is all just smoke and mirrors when it comes to sharing tasks. The PCI-E buss actually runs data in order from one device to the next.

. . . Except that on my Gen2/3 motherboards, you can't use the onboard Asmedia USB3 if certain PCI-E slots are filled, or a third slot only gets you x4 lanes. Marketing hype or none -- there's no way around it.

The other thing to look at is driver conflict. That's where that DPC Latency Checker software helps. OR I can say that it helped me sort out the problem of one core running perpetually at 100% usage on a system.

 

[UsandThem]

There's no doubt some boards have higher DPC latency, which you can see in reviews or using a program like BonzaiDuck mentioned. I believe [H]ardocp now checks DPC latency in their motherboard reviews, so it should be easy to see if there is a concern.

It just comes down to the design of the motherboard, and what components are on it. Just because a motherboard is over $300, it doesn't exempt it from high DPC latency. For example, my motherboard has sold from $140 to $190, and its latency isn't high enough to cause any stuttering.

I think in large part, DPC latency is just another thing for enthusiasts to look at along with video card clocks, RAM timings , etc.

And I'm not saying it's a bad thing, or simply discounting the concern, but people like us on the various hardware forums tend to dig into all the data and charts we can get. We want to squeeze every ounce of performance out of our 'babies'. For the average person, they just want their PC to work properly, and not to crash.

 

[BonzaiDuck]

There's no doubt some boards have higher DPC latency, which you can see in reviews or using a program like BonzaiDuck mentioned. I believe [H]ardocp now checks DPC latency in their motherboard reviews, so it should be easy to see if there is a concern.

It just comes down to the design of the motherboard, and what components are on it. Just because a motherboard is over $300, it doesn't exempt it from high DPC latency. For example, my motherboard has sold from $140 to $190, and its latency isn't high enough to cause any stuttering.

I think in large part, DPC latency is just another thing for enthusiasts to look at along with video card clocks, RAM timings , etc.

And I'm not saying it's a bad thing, or simply discounting the concern, but people like us on the various hardware forums tend to dig into all the data and charts we can get. We want to squeeze every ounce of performance out of our 'babies'. For the average person, they just want their PC to work properly, and not to crash.

Sure . . . and I only focused on the phenomenon when I discovered that my server had been running one core at 100% for several months. Now, I might want to investigate this factor with my Sabertooth Z170 S. I just wouldn't have much of an incentive besides my attention to this thread. Everything is so damn quick! $170 sure beats shelling out $300+. I can live with the 12-phase power-design. I just can't see what features I'm missing that I wanted in the first place. I could wish that I had a board with room for 2x M.2 instead of one, but that's minor, I think.

 

[UsandThem]

Sure . . . and I only focused on the phenomenon when I discovered that my server had been running one core at 100% for several months. Now, I might want to investigate this factor with my Sabertooth Z170 S. I just wouldn't have much of an incentive besides my attention to this thread. Everything is so damn quick! $170 sure beats shelling out $300+.

Most of my comment wasn't directed towards you, outside of you mentioning the handy DPC utility to use. I didn't notice the OP replied to my question back on October 14th. I'm not sure if he stuck around (many new users sign up, ask a question, and disappear), so I just wanted to point out there are very good 'budget' boards out there, and there is no need to go to a $300+ motherboard just because of DPC latency.

Many of us "old timers" aren't "chasing the dragon" anymore, and only people who are in the 'enthusiast' category really need the top-of-the-line boards. Today, even the most entry-level z170 boards will give a person a good over-clock, and decent expandability. Once you start going up, there's usually better audio, expansion, cooling, build quality, etc.

Of course if someone wants a $300 motherboard, that's their call. I exited the 'enthusiast' category back around 2006, so even though I still love new tech, I don't replace my whole system every 12-18 months like I usually did back in 1997 to 2006. Now a person can easily keep a computer for 3-5 years without being left too far in the dust.

 

[wingman04]

Sure . . . and I only focused on the phenomenon when I discovered that my server had been running one core at 100% for several months. Now, I might want to investigate this factor with my Sabertooth Z170 S. I just wouldn't have much of an incentive besides my attention to this thread. Everything is so damn quick! $170 sure beats shelling out $300+. I can live with the 12-phase power-design. I just can't see what features I'm missing that I wanted in the first place. I could wish that I had a board with room for 2x M.2 instead of one, but that's minor, I think.

I have a Mid overclock 4.5GHz on my Gigabyte Z170 HD3 budget board 6 phase. Overclocked with ease my first try, I just looked at all the average core voltages on the web and set my voltage accordingly. After it ran prime95 for 14 hours, then I tried to lower my Vcore in very small steps and prime95 error'd, so I selected the original setting to have a buffer. I use Dynamic VID 0.075

 

[BonzaiDuck]

I have a Mid overclock 4.5GHz on my Gigabyte Z170 HD3 budget board 6 phase. Overclocked with ease my first try, I just looked at all the average core voltages on the web and set my voltage accordingly. After it ran prime95 for 14 hours, then I tried to lower my Vcore in very small steps and prime95 error'd, so I selected the original setting to have a buffer. I use Dynamic VID 0.075

. . . Which is an interesting epilogue to our exchanges on another recent thread you'll remember. By my scrutinizing estimation, I believe that I am running about 30mV above that threshold for 4.5, at my additional estimate of a stock LLC value for the Auto setting. I keep wondering whether I should trim that back 10mV -- even 15mV. But at more modest clock settings that began anyway by adjusting Vcore downward from "Auto" or default, the surplus would have no ill effect. The higher you run the multiplier, the closer you get to excessive voltage, the higher the temperatures and the more parsimonious one might incline with Vcore.

You mention that the board is "6 phase." Are you just counting the phases for the CPU, or those reserved for the iGPU as well? It's no less about quality of components as it is about the rule of thumb of a phase-power-design spec. It's just pretty likely that top-tier expensive boards have it both ways.

My fam-damn-ily has very simple desires, so when they need an upgrade, I make modest motherboard choices. Building my own with the future-proofing or performance satisfaction over time figured in, it's always going to be a notch or two above.

Enthusiasts of whatever degree are always going to feel tempted to spend money under an impulsive curiosity.

If it's about a lifecycle, I could imagine one balancing his guilt for hoarding, the options and trouble for selling on the used market or making gifts to friends and family, or simply dumping it in the truck for a trip to the county recycling center.

I couldn't say whether people who buy a $600 board are less disciplined. There are many budget lines crossing over the space of indifference curves. On the other hand, why pay for something -- some feature -- that you have no desire to use?

 

[wingman04]

. . . Which is an interesting epilogue to our exchanges on another recent thread you'll remember. By my scrutinizing estimation, I believe that I am running about 30mV above that threshold for 4.5, at my additional estimate of a stock LLC value for the Auto setting. I keep wondering whether I should trim that back 10mV -- even 15mV. But at more modest clock settings that began anyway by adjusting Vcore downward from "Auto" or default, the surplus would have no ill effect. The higher you run the multiplier, the closer you get to excessive voltage, the higher the temperatures and the more parsimonious one might incline with Vcore.

You mention that the board is "6 phase." Are you just counting the phases for the CPU, or those reserved for the iGPU as well? It's no less about quality of components as it is about the rule of thumb of a phase-power-design spec. It's just pretty likely that top-tier expensive boards have it both ways.

My fam-damn-ily has very simple desires, so when they need an upgrade, I make modest motherboard choices. Building my own with the future-proofing or performance satisfaction over time figured in, it's always going to be a notch or two above.

Enthusiasts of whatever degree are always going to feel tempted to spend money under an impulsive curiosity.

If it's about a lifecycle, I could imagine one balancing his guilt for hoarding, the options and trouble for selling on the used market or making gifts to friends and family, or simply dumping it in the truck for a trip to the county recycling center.

I couldn't say whether people who buy a $600 board are less disciplined. There are many budget lines crossing over the space of indifference curves. On the other hand, why pay for something -- some feature -- that you have no desire to use?

My Z170 HD3 board has a six phase for the CPU. 1 phase for IGPU, 1 phase for VCCIO, 1 phase for VCCSA, 3 phases for the cores= 6 phases. In the past 30 years I always buy budget board for 24/7 overclocking.
When folks Know how a VRM(voltage regulator module) works like me they usually don't over pay on a motherboard for a daily use. It has been marketing that drives highend motherboards, not 24/7 overclocking ability. I have been around overclocking before the word Enthusiasts and you are correct they like to spend money. If folks only knew what I know about VRMs they would not waste there money for 24/7 overclocking. Also motherboard VRMs are tested for the maximum power delivery and 3 year warranty, running 24/7.

For over clocking I use all Intel power saving features and I don't disable or set a level to Intel's LLC (load line Calibration) specification, I just leave it on Auto. Once you know how most everything works on the motherboard and CPU a person will come to the final conclusion for a 24/7 overclock, just change the multiplier and Adaptive Dynamic VID offset for Vcore, then after testing I'm done overclocking.

For me I have a Bios setting of +20mV Vcore buffer, testing with prime95. I don't know what Intel uses for core voltage buffer, however I can overclock from stock turbo 3.9GHz VID 1.271v to 4.2GHz on stock VID of 1.271v with Prime95.

If you wan't to test what Intel's Vcore buffer is, just lower the Vcore at default Bios settings and test with a stress test.

 

[BonzaiDuck]

My Z170 HD3 board has a six phase for the CPU. 1 phase for IGPU, 1 phase for VCCIO, 1 phase for VCCSA, 3 phases for the cores= 6 phases. In the past 30 years I always buy budget board for 24/7 overclocking.
When folks Know how a VRM(voltage regulator module) works like me they usually don't over pay on a motherboard for a daily use. It has been marketing that drives highend motherboards, not 24/7 overclocking ability. I have been around overclocking before the word Enthusiasts and you are correct they like to spend money. If folks only knew what I know about VRMs they would not waste there money for 24/7 overclocking. Also motherboard VRMs are tested for the maximum power delivery and 3 year warranty, running 24/7.

For over clocking I use all Intel power saving features and I don't disable or set a level to Intel's LLC (load line Calibration) specification, I just leave it on Auto. Once you know how most everything works on the motherboard and CPU a person will come to the final conclusion for a 24/7 overclock, just change the multiplier and Adaptive Dynamic VID offset for Vcore, then after testing I'm done overclocking.

For me I have a Bios setting of +20mV Vcore buffer, testing with prime95. I don't know what Intel uses for core voltage buffer, however I can overclock from stock turbo 3.9GHz VID 1.271v to 4.2GHz on stock VID of 1.271v with Prime95.

If you wan't to test what Intel's Vcore buffer is, just lower the Vcore at default Bios settings and test with a stress test.

that sounds about right . . . the prevailing folklore says to turn those features off, but once you feel sure about the key settings you mention, turning them on shouldn't change a stress-test result, although earlier boards required turning off the c3 and c6 reporting or idle instability would result. Spread Spectrum can be re-enabled, as well as EIST and C1E. As for the LLC, I wanted to know more or less what "Auto" provides. once you found it, you could just set it back to Auto or leave it where you think it belongs. I was just surprised that some new boards such as mine have a default matching what you'd give it anyway for overclocking if you just wanted to minimize droop.

 

[wingman04]

that sounds about right . . . the prevailing folklore says to turn those features off, but once you feel sure about the key settings you mention, turning them on shouldn't change a stress-test result, although earlier boards required turning off the c3 and c6 reporting or idle instability would result. Spread Spectrum can be re-enabled, as well as EIST and C1E. As for the LLC, I wanted to know more or less what "Auto" provides. once you found it, you could just set it back to Auto or leave it where you think it belongs. I was just surprised that some new boards such as mine have a default matching what you'd give it anyway for overclocking if you just wanted to minimize droop.

Anandtech
If you've ever overclocked a system, chances are that at some point or another you've had opportunity to become upset with your Vdroop "problem." Some users, confused as to why their system refuses to exactly match actual processor supply voltage to the value specified in BIOS, are quick to blame the quality their motherboard; still others find fault with the difference noted between their board's idle and full-load processor supply voltages. Actually, load line droop (Vdroop) is an inherent part of any Intel power delivery design specification and serves an important role in maintaining system stability. In most cases, comments regarding unacceptable power delivery performance are completely unfounded. To make matters worse, unjustified negative consumer perception surrounding this often misunderstood design feature eventually forced a few motherboard manufacturers to respond to enthusiasts' demands for action by adding an option in their BIOS that effectively disables this important function.http://www.anandtech.com/show/2404/5

I If you leave everything stock and just change the multiplier and Adaptive VID dynamic offset Vcore for overclocking, the clock will go up and down with voltage and it won't lock up at idle.

Voltage identificationThe correct supply voltage is communicated by the microprocessor to the VRM at startup via a number of bits called VID (voltage identification). In particular, the VRM initially provides a standard supply voltage to the VID logic, which is the part of the processor whose only aim is to then send the VID to the VRM. When the VRM has received the VID identifying the required supply voltage, it starts acting as a voltage regulator, providing the required constant voltage supply to the processor.

Instead of having a power supply unit generate some fixed voltage, the CPU uses a small set of digital signals, the VID lines, to instruct an on-board power converter of the desired voltage level. The switch-mode buck converter then adjusts its output accordingly. The flexibility so obtained makes it possible to use the same power supply unit for CPUs with somewhat different nominal supply voltages and to reduce power consumption during idle periods by lowering the supply voltage.[5]

For example, a unit with 5-bit VID would output one of at most 32 (25) distinct output voltages. These voltages are usually (but not always) evenly spaced within a given range. Some of the code words may be reserved for special functions such as shutting down the unit, hence a 5-bit VID unit may have fewer than 32 output voltage levels. How the numerical codes map to supply voltages is typically specified in tables provided by component manufacturers. As of 2008 VID comes in 5-, 6- and 8-bit varieties and is mostly applied to power modules outputting between 0.5 V and 3.5 V.https://en.wikipedia.org/wiki/Voltage_regulator_module

 

[BonzaiDuck]

Yup. And another Anand article around 2007/2008 had cautions about twiddling with LLC -- something about pushing transition voltage spikes above VID. But it seems that the "auto" on my board gives the equivalent of a medium setting for LLC.

The low-load or idle lockups had an explanation related to VCore offset, when people were able to choose a negative value, or simply increase it as a means of increasing VCORE for the overclock. It would have an effect on the effective idle voltage with EIST enabled, which might drop too low -- therefore instability.

on the old Z68 boards, there was this intermittent idle instability that could occur as infrequently as once in 10 days if you left the machine running that long. Hard to troubleshoot. The info I was able to gather from forums, guides and other sources pointed to the C3/C6 Reporting items in the BIOS. For my particular problem, that fixed it.

Oh! I just realized -- we're talking about the same article of December, 2007.

 

[VirtualLarry]

Motherboard resources is mostly marking hype. PCs can only do one task at a time it is all just smoke and mirrors when it comes to sharing tasks. The PCI-E buss actually runs data in order from one device to the next.

My understanding is that this couldn't be more wrong.

PCI-E supports peer-to-peer, as well as device-host and host-device transfers, but it doesn't use a token-ring, that I know of.

As I understood it, a big advantage of PCI-E is that slots / root ports are concurrent, whereas the prior PCI standard, the bus was shared.

 

[wingman04]

Yup. And another Anand article around 2007/2008 had cautions about twiddling with LLC -- something about pushing transition voltage spikes above VID. But it seems that the "auto" on my board gives the equivalent of a medium setting for LLC.

The low-load or idle lockups had an explanation related to VCore offset, when people were able to choose a negative value, or simply increase it as a means of increasing VCORE for the overclock. It would have an effect on the effective idle voltage with EIST enabled, which might drop too low -- therefore instability.

on the old Z68 boards, there was this intermittent idle instability that could occur as infrequently as once in 10 days if you left the machine running that long. Hard to troubleshoot. The info I was able to gather from forums, guides and other sources pointed to the C3/C6 Reporting items in the BIOS. For my particular problem, that fixed it.

Oh! I just realized -- we're talking about the same article of December, 2007.

All boards are the same for Intel's Load line calibration, the board manufactures back in 2007 made a option to disable it. Auto lets the BIOS automatically configure this setting and sets the voltage following Intel®'s specifications.

If folks use the dynamic CPU VID for Vcore, it uses the CPU dynamic VID to communicated dynamic voltage requirement to the VRM on every motherboard. MSI and ASUS calls VID Adaptive. Gigabyte calls VID Dynamic DVID.

Intels VID (Voltage identification) and LLC (load line calibration) are two different things. I don't know how you acquired your voltage readings? The VID voltage varies with load. however with LLC enabled it should have a voltage drop with fixed Vcore under load. With VID on and LLC on you can set what you wan't for your peak voltage under stress testing the the voltage will lower when the load is lowered.

Intel engineered VID and LLC to run together at the same time so the voltage will not over shoot all the time with varying load.

My understanding is that this couldn't be more wrong.

PCI-E supports peer-to-peer, as well as device-host and host-device transfers, but it doesn't use a token-ring, that I know of.

As I understood it, a big advantage of PCI-E is that slots / root ports are concurrent, whereas the prior PCI standard, the bus was shared.

Let me explain this simply. The CPU can only do one task at a time with software. First Bios instructions loads into the CPU then into memory, then windows load into memory, while doing all this the CPU only works with on task at a time with all things linked to it, the CPU goes round robin. So when it is the PCI E turn being called from by the software, the slots run in order of IRQs within the windows software and only one device communicates with the CPU at a time, then the data is sent to memory and read from the memory, this is all done in ns, then the CPU takes the next instruction from the software.

The only main difference from PCI to PCI-E, is the PCI lanes runs a parallel data set sent to and from the CPU and PCI-E lanes is Serial data set sent to and from the CPU, like SATA Drive communication.

The only thing I did not mention was the south bridge that controls the traffic one data set at a time.

So now you can see that the CPU only handles one task at a time, otherwise there would be a collision in calculations by the CPU.

 

[BonzaiDuck]

All boards are the same for Intel's Load line calibration, the board manufactures back in 2007 made a option to disable it. Auto lets the BIOS automatically configure this setting and sets the voltage following Intel®'s specifications.

If folks use the dynamic CPU VID for Vcore, it uses the CPU dynamic VID to communicated dynamic voltage requirement to the VRM on every motherboard. MSI and ASUS calls VID Adaptive. Gigabyte calls VID Dynamic DVID.

Intels VID (Voltage identification) and LLC (load line calibration) are two different things. I don't know how you acquired your voltage readings? The VID voltage varies with load. however with LLC enabled it should have a voltage drop with fixed Vcore under load. With VID on and LLC on you can set what you wan't for your peak voltage under stress testing the the voltage will lower when the load is lowered.

Intel engineered VID and LLC to run together at the same time so the voltage will not over shoot all the time with varying load.

How I acquired voltage readings? I'd seen some people show a need to use a multi-meter. I just examine the software readings. I understand that VID changes with load.

I had everything on stock settings, and then experimented with lowering LLC by taking it off "Auto." By trial and error, watching the peak and drooped voltage during stress, I came to a good idea that these boards give LLC=5 on Auto. That-- by watching the VCore reading -- the VID readout is higher by an increment. Ultimately, if you overclock this and maybe other Z170 boards, and reviewing what others did, you need LLC=5 anyway above 4.5 Ghz -- if all the boards have the same scale. On the Sabertooth, there are 8 LLC levels. My best judgment of it and from what I'd read: you WANT to have some vDroop in the mix, even if it's just 10 mV. I think I'd seen pronouncements either about this particular board or Z170 overclocking in general suggesting you'd want Vcore to be close to the VID. But howsoever that may be so, I could see how going from LLC=1 to 2 . . . etc. would narrow the droop value at load against the unloaded peak.