PDA

View Full Version : LLC (Load Line Calibration) not working... Bios fix required ASAP



morph.
10-14-2017, 05:55 AM
Any eta on when there will be a bios patch/fix for LLC?

Vdrop is just too high at the moment it seems.

https://www.youtube.com/watch?v=hCAQpvFhiUs

Raja@ASUS
10-14-2017, 07:03 AM
https://www.io-tech.fi/uutinen/paivitys-testilabrasta-coffee-laken-kayttojannite-asuksen-strix-z370-emolevyilla/

Use google translate. That will give some of the you the answer to this hoo-haa. Unfortunately, they aren't measuring from the right place, either. To get a more accurate reading, you'd use one of the MLCC caps behind the socket. You want to get as close as you can to the socket pins so that you can see how much resistive loss you get through the power plane. Review sites simply aren't equipped to do this stuff properly. The non-technical guys profess to know things they don't understand, and those that try to be technical mess it up, too.

The requested VID can be affected by the IA/AC and DC load line settings. Has been like that for a while now. I'll see if I can get a board here and check what's going on.

morph.
10-14-2017, 07:40 AM
Been reported that some are getting as much as 0.9V vdroop though... is that inaccurate?

Just from apps such as CPUID / CPU-Z

Raja@ASUS
10-14-2017, 08:25 AM
Been reported that some are getting as much as 0.9V vdroop though... is that inaccurate?

Just from apps such as CPUID / CPU-Z

That's why you should measure properly before coming to a conclusion. Don't rely on software. Plus the IA AC/DC settings are likely involved, so calling this Vdroop may be inaccurate, too.

morph.
10-14-2017, 10:48 AM
Sorry i'm not an electrical engineer, but don't these apps pull the data from the BIOS and if so the bios seems to be reporting incorrect power data how can 'enthusiasts' trust the read outs without multi-meters and what not to overclock?

JustinThyme
10-14-2017, 11:16 AM
I am an electrical engineer and Im afraid Raja has this right. I have to say Im a bit disappointed in Tom to produce such rubbish with zero supporting documentation.
Are you personally experiencing measured Vdroop issues? If so the prudent thing to do is contact support and arrange for an RMA of your boards.

I have 3 software packages that pull whats supposed to be the same data from the smbus and no two read the same. This so called bug could be a result of these vendors not updating their software yet for the Z370 chipset. There's a million things here and nothing worse than troubleshooting conjecture.

Raja@ASUS
10-14-2017, 12:17 PM
Sorry i'm not an electrical engineer, but don't these apps pull the data from the BIOS and if so the bios seems to be reporting incorrect power data how can 'enthusiasts' trust the read outs without multi-meters and what not to overclock?


Power and voltage are not the same thing. They are related, but one shouldn't confuse where and when to use these terms. Software doesn't use bios to report voltages. Even if you were unsure of voltage, you can check cpu power consumption in software such as aida64. It's power that ultimately matters. Looking at voltage without checking current doesn't tell you much. P=VA. Remember that and you won't need to rely on reviewers for info.

As for the rest, like I said, I'll see if I can get a board to check, or ask one of the r&d guys to look into it. Oh, and yes, software often needs an update, too. That part would be up to the software developer to fix.

Korth
10-14-2017, 01:55 PM
Oh, and yes, software often needs an update, too. That part would be up to the software developer to fix.
Slightly off-topic, but I'm curious:

Benchmarking and hardware monitoring softwares (CPU-Z, HWMONITOR, HWiNFO, etc) need to be calibrated and updated for best accuracy on each new platform. Does ASUS provide sample boards or "confidential" technical data to these software devs before product launch?

Raja@ASUS
10-14-2017, 02:02 PM
Sometimes we do. However, I'm not at liberty to discuss who we sample boards to and why.

Korth
10-14-2017, 03:34 PM
A fair answer. Thanks ;)

ClearedIn2Bravo2
10-15-2017, 02:29 AM
So @Raja is Asus fixing this issue or not? I am seeing 50-60mV drop under load even with LLC set at it's highest setting. This is not acceptable as I am running a manual voltage of 1.290v on my Z370-E with 8700K. This results in me needing to use a much higher voltage than necessary for any given clockspeed to achieve stability when overclocked.

Is Asus saying there isn't an issue? If they are I'll proceed with my purchase protection/charge back against the vendor who is unwilling to accept a return/exchange for a different product. I refuse to keep a product that is not fully functional or inferior to other brands. I paid over $200 for this motherboard and expect a product that is both supported fully by Asus for it's product lifespan and a quality product out of the box.

Yes I am running latest BIOS from Asus website, yes I am on a new Window 10 Clean install, yes I am running latest software versions.

Raja@ASUS
10-15-2017, 05:58 AM
So @Raja is Asus fixing this issue or not? I am seeing 50-60mV drop under load even with LLC set at it's highest setting. This is not acceptable as I am running a manual voltage of 1.290v on my Z370-E with 8700K. This results in me needing to use a much higher voltage than necessary for any given clockspeed to achieve stability when overclocked.

Is Asus saying there isn't an issue? If they are I'll proceed with my purchase protection/charge back against the vendor who is unwilling to accept a return/exchange for a different product. I refuse to keep a product that is not fully functional or inferior to other brands. I paid over $200 for this motherboard and expect a product that is both supported fully by Asus for it's product lifespan and a quality product out of the box.

Yes I am running latest BIOS from Asus website, yes I am on a new Window 10 Clean install, yes I am running latest software versions.

Seeing 50-60Mv of droop in software or measured from a behind-socket MLCC cap? If you read the first page properly, you'll see that I said I'll look into it. The posts were made yesterday. Today is one day later and a Sunday. There is no way for me to get a board and test it this quickly. It'll take a week (minimum) for me to get a board and test it. On the flipside, I know R&D is looking into this. Going by the "scanty" evidence that's being presented by all parties (including the 'reviews' and your post), my suspicion is that there may be a combination of issues between some monitoring software, and on the board side, the IA AC/DC load line setting. The former requires the third-party software developers to update. The latter can be changed manually in UEFI:


Try setting 2.10 for the IA AC/DC load line settings in the Internal CPU Power Management page of UEFI.

Here's a sample screenshot of these two settings:

http://edgeup.asus.com/wp-content/uploads/2017/01/UEFI-3.jpg

Just remember that you want to set 2.10 and not 0.01 as shown. This will change how much VID is requested by the CPU. On these newer architectures, Intel recommends these values are changed in relation to the VRM load line setting.



So @Raja is Asus fixing this issue or not? I am seeing 50-60mV drop under load even with LLC set at it's highest setting. This is not acceptable as I am running a manual voltage of 1.290v on my Z370-E with 8700K. This results in me needing to use a much higher voltage than necessary for any given clockspeed to achieve stability when overclocked.

This statement is ambiguous. Needing to use, and needing to set, are two different things. Let's not confuse them. Needing to use implies the voltage needs to be 50-60mV higher on the ASUS boards under full load. Needing to set implies you have to set 50-60mV higher than the target voltage due to droop. The implications of these statements are completely different.

morph.
10-15-2017, 01:52 PM
Just to be clear this is not 2.1 mOhm but actual 2.10 value yeah Raja?

Thanks.

JustinThyme
10-15-2017, 02:21 PM
So @Raja is Asus fixing this issue or not? I am seeing 50-60mV drop under load even with LLC set at it's highest setting. This is not acceptable as I am running a manual voltage of 1.290v on my Z370-E with 8700K. This results in me needing to use a much higher voltage than necessary for any given clockspeed to achieve stability when overclocked.

Is Asus saying there isn't an issue? If they are I'll proceed with my purchase protection/charge back against the vendor who is unwilling to accept a return/exchange for a different product. I refuse to keep a product that is not fully functional or inferior to other brands. I paid over $200 for this motherboard and expect a product that is both supported fully by Asus for it's product lifespan and a quality product out of the box.

Yes I am running latest BIOS from Asus website, yes I am on a new Window 10 Clean install, yes I am running latest software versions.

An old and wise lady once told me

You get more flies with honey than with vinegar, that is is you are into catching flies.

Have you read the entire thread and measured this with a calibrated DVM and not 3rd party software that has yet to be developed for this platform?
What issues, if any are you experiencing?
OC amounts of any CPU are not guaranteed results and vary from chip to chip...Silicon Lottery

Charge backs are between you and where you bought the board from. There has been zero proof presented other than people watching a you tube jockey that is spewing conjecture and quite honestly walking a treacherous line of liability. I watched that video and not at any point did he do anything but run his yap. Not a single shred of anything to support the atrocity spewing from his mouth. With that being said saying its inferior to other brands is not a valid reason for a charge back. Some like Fords better than Chevys and call a Chevy Truck inferior. That doesn't mean they can take it back to the dealer and get their money back. As for the fully functional, no one has shown anything different, only conjecture.

Raja said they are looking into it so keep checking back.

Tech Hog
10-16-2017, 01:38 AM
Seeing 50-60Mv of droop in software or measured from a behind-socket MLCC cap? If you read the first page properly, you'll see that I said I'll look into it. The posts were made yesterday. Today is one day later and a Sunday. There is no way for me to get a board and test it this quickly. It'll take a week (minimum) for me to get a board and test it. On the flipside, I know R&D is looking into this. Going by the "scanty" evidence that's being presented by all parties (including the 'reviews' and your post), my suspicion is that there may be a combination of issues between some monitoring software, and on the board side, the IA AC/DC load line setting. The former requires the third-party software developers to update. The latter can be changed manually in UEFI:


Try setting 2.10 for the IA AC/DC load line settings in the Internal CPU Power Management page of UEFI.

Here's a sample screenshot of these two settings:

http://edgeup.asus.com/wp-content/uploads/2017/01/UEFI-3.jpg

Just remember that you want to set 2.10 and not 0.01 as shown. This will change how much VID is requested by the CPU. On these newer architectures, Intel recommends these values are changed in relation to the VRM load line setting.




This statement is ambiguous. Needing to use, and needing to set, are two different things. Let's not confuse them. Needing to use implies the voltage needs to be 50-60mV higher on the ASUS boards under full load. Needing to set implies you have to set 50-60mV higher than the target voltage due to droop. The implications of these statements are completely different.

I tried this and the software reported over 1.5V Vcore with 1.355V set in the BIOS. You really released this board without any testing whatsoever, huh? ANd on top of that, you're getting snarky over being called out and you're shifting the blame. You couldn't have delayed the boards another 2-3 weeks to make sure it launched in a usable state? Last Asus product I'll ever even consider buying.

Raja@ASUS
10-16-2017, 02:50 AM
I tried this and the software reported over 1.5V Vcore with 1.355V set in the BIOS. You really released this board without any testing whatsoever, huh? ANd on top of that, you're getting snarky over being called out and you're shifting the blame. You couldn't have delayed the boards another 2-3 weeks to make sure it launched in a usable state? Last Asus product I'll ever even consider buying.

R&D does the testing, not me. Again, measure the voltage properly and see what it's actually feeding, then adjust the ia ac/dc loadline to get the voltage you want. If third party software isn't accurate, you'll need to talk to them about it. 2.10 mOhms is Intel ref. I'm not sure if this affects all voltage modes. Like I said, I need a board to see what's going on. If you don't want to test or can't, then you'll need to wait until I do (or someone more capable tests).

Asking for proper testing before drawing firm conclusions isn't snarky.

JustinThyme
10-16-2017, 04:00 AM
I tried this and the software reported over 1.5V Vcore with 1.355V set in the BIOS. You really released this board without any testing whatsoever, huh? ANd on top of that, you're getting snarky over being called out and you're shifting the blame. You couldn't have delayed the boards another 2-3 weeks to make sure it launched in a usable state? Last Asus product I'll ever even consider buying.

What software? If its not a DVM then its third party that hasn't been updated.
What problems are you experiencing as a result?
Making unfounded claims is not calling out. Its better known as conjecture.
They could have delayed the boards for 10 years and still gotten the same reception. Consumer complain if they take too long to market, then they complain that they got to market too fast then they complain that once everything is ironed out the platform is obsolete and its time to start all over again. Im sure ASUS wont miss your $20.

Raja@ASUS
10-16-2017, 04:42 AM
R&D is working on it.

morph.
10-16-2017, 04:58 AM
2.10 mOhms is Intel ref. I'm not sure if this affects all voltage modes. Like I said, I need a board to see what's going on.

Raja is it 2.10 mOhms or 2.10 value? it seems to be different.

As a value of 100 = 1.0 mOhms.

so in theory to get 2.10 mOhms value needs to be 210?

Raja@ASUS
10-16-2017, 06:18 AM
Raja is it 2.10 mOhms or 2.10 value? it seems to be different.

As a value of 100 = 1.0 mOhms.

so in theory to get 2.10 mOhms value needs to be 210?

The value I wrote is as r&d suggested. I'd stick to what was supplied rather than applying your own reasoning to the unit value (flip the UEFI unit explanation and it makes sense). And I'd suggest measuring the voltage before making changes, too. The required value may not be the same, and will likely need changing according to the external VRM LLC used. This is for adaptive/offset modes only. Plus, I'd wait and see if they patch it via a UEFI update. Especially if you can't make measurements.

illusiveguy
10-16-2017, 11:12 AM
This is only somewhat related, but is there a reason why the E and F boards got a BIOS update and not the H. Aren't these all essentially the same boards, except for WIFI and RGB additions?

Should I be nervous if/when the LLC issue is fixed via a software update?

Raja@ASUS
10-16-2017, 11:25 AM
This is only somewhat related, but is there a reason why the E and F boards got a BIOS update and not the H. Aren't these all essentially the same boards, except for WIFI and RGB additions?

Should I be nervous if/when the LLC issue is fixed via a software update?


This depends on what the updates are for and how busy the teams are. Staggered releases are common.

Tech Hog
10-16-2017, 11:32 AM
R&D does the testing, not me. Again, measure the voltage properly and see what it's actually feeding, then adjust the ia ac/dc loadline to get the voltage you want. If third party software isn't accurate, you'll need to talk to them about it. 2.10 mOhms is Intel ref. I'm not sure if this affects all voltage modes. Like I said, I need a board to see what's going on. If you don't want to test or can't, then you'll need to wait until I do (or someone more capable tests).

Asking for proper testing before drawing firm conclusions isn't snarky.

And the only way yo properly test is to buy a new cooler? Fine, then buy me one and for all of your future boards indicate that noctua's coolers are incompatible with your boards. :)

And you are being snarky. You're putting others down for questioning the greatness of Asus. And even if it's off, it's not .15V off. You people are just shifting responsibility and throwing blame wherever you can. All that you should have said is "We'll look into it. Also keep in mind that third party software could be reading incorrectly." When I put in 0.01 instead the reading is close to what I set in BIOS, and the temps seem in line with what to expect at that voltage, so there's no freaking way that the actual Vcore is far enough away from the true Vcore for it to be safe. I've seen people measure it. The Delta is half of what it would need to be for your advice to be sound. Tell R&D to look into it. I can guarantee that doing that will cause voltage to spike to no less than .1V higher than it should be.

Raja@ASUS
10-16-2017, 11:41 AM
And the only way yo properly test is to buy a new cooler? Fine, then buy me one and for all of your future boards indicate that noctua's coolers are incompatible with your boards. :)

And you are being snarky. You're putting others down for questioning the greatness of Asus. And even if it's off, it's not .15V off. You people are just shifting responsibility and throwing blame wherever you can. All that you should have said is "We'll look into it. Also keep in mind that third party software could be reading incorrectly." When I put in 0.01 instead the reading is close to what I set in BIOS, and the temps seem in line with what to expect at that voltage, so there's no freaking way that the actual Vcore is far enough away from the true Vcore for it to be safe. I've seen people measure it. The Delta is half of what it would need to be for your advice to be sound. Tell R&D to look into it. I can guarantee that doing that will cause voltage to spike to no less than .1V higher than it should be.

No, you wouldn't need a different cooler, because you'd see the initial delta on a DMM and would use values within its confines. Remember, I already stated that I'd look into it or have R&D look into it. Those posts were made on a Saturday. No fewer than 24 hours later, someone came in asking for a definitive answer. Let's digress, though.

The sag that is being reported and you've allegedly seen people measure would take more than a value of 0.01 to cure, so something in your 'testing' is off. But if you are happy with 0.01, carry on. The value of 0.01 was for Kaby Lake (as shown in this (http://edgeup.asus.com/2017/kaby-lake-overclocking-guide/3/) guide).


And oddly enough, I'm expecting some UEFI updates later. ;)

Silent Scone@ROG
10-16-2017, 12:03 PM
I tried this and the software reported over 1.5V Vcore with 1.355V set in the BIOS.

You've not mentioned how you measured the voltage?

Tech Hog
10-16-2017, 12:20 PM
No, you wouldn't need a different cooler, because you'd see the initial delta on a DMM and would use values within its confines. Remember I already stated that I'd look into it or have R&D look into it. Those posts were made on a Saturday. No fewer than 24 hours later, someone came in asking for a definitive answer. Let's digress, though.

The sag that is being reported and you've allegedly seen people measure would take more than a value of 0.01 to cure, so something in your 'testing' is off. But if you are happy with 0.01, carry on. The value of 0.01 was for Kaby Lake (as shown in this (http://edgeup.asus.com/2017/kaby-lake-overclocking-guide/3/) guide).


And oddly enough, I'm expecting some UEFI updates later. ;)

So if I don't have a DMM I should just not use my computer until there's an update? Okay. Well if I can't use my computer anyway there's little reason to not just buy a new board, is there?

And I'd still need a different cooler to measure the initial value.

Raja@ASUS
10-16-2017, 12:54 PM
So if I don't have a DMM I should just not use my computer until there's an update? Okay. Well if I can't use my computer anyway there's little reason to not just buy a new board, is there?

And I'd still need a different cooler to measure the initial value.

Vdroop isn't going to harm your computer, so carry on using it.

You would not need a different cooler to make a measurement. You would make the measurement before you'd determine you need to change the value. Making changes before you know you need to make them is akin to putting the cart before the horse.

Tech Hog
10-16-2017, 01:10 PM
Vdroop isn't going to harm your computer, so carry on using it.

You would not need a different cooler to make a measurement. You would make the measurement before you'd determine you need to change the value. Making changes before you know you need to make them, is akin to putting the cart before the horse.

But I can't make the measurement because the backplate is in the way. So yes, I'd need to change the cooler.

Raja@ASUS
10-16-2017, 01:13 PM
But I can't make the measurement because the backplate is in the way. So yes, I'd need to change the cooler.

Then dont change a thing and carry on using your PC. When there's an update for it, use the update. Like I said, Vdroop is not harmful.

Tech Hog
10-16-2017, 01:28 PM
Then dont change a thing and carry on using your PC. When there's an update for it, use the update. Like I said, Vdroop is not harmful.

If I don't change anything from "stock" then MCE will be on due to another UEFI bug which sets "sync all cores" to default. And my concern is more that voltage might be spiking.

I guess this really is my fault, though. I could have avoided this by getting the Hero like originally planned. Well, lesson learned: only buy Asus products if they're high-end. I'll tell my friend who can't afford to spend $300 on a board to look at another company instead. Thank you. I'll wait for the update, and maybe get something new once Z390 rolls around.

Raja@ASUS
10-16-2017, 01:34 PM
If I don't change anything from "stock" then MCE will be on due to another UEFI bug which sets "sync all cores" to default. And my concern is more that voltage might be spiking.

I guess this really is my fault, though. I could have avoided this by getting the Hero like originally planned. Well, lesson learned: only buy Asus products if they're high-end. I'll tell my friend who can't afford to spend $300 on a board to look at another company instead. Thank you. I'll wait for the update, and maybe get something new once Z390 rolls around.


From the test report I received from R&D this morning, it is not spiking over the applied voltage, so you are good.

Tech Hog
10-16-2017, 01:53 PM
From the test report I received from R&D this morning, it is not spiking over the applied voltage, so you are good.

If you say so... I just hope that the wait for an update isn't too long.

Korth
10-16-2017, 02:11 PM
"I am not satisfied with this *rushed* product. I demand you fix it *right now*."

Just makes me shake my big blue head in bemused wonderment.

Tech Hog
10-16-2017, 02:16 PM
"I am not satisfied with this *rushed* product. I demand you fix it *right now*."

Just makes me shake my big blue head in bemused wonderment.

I mean, they could have just not rushed it. The M10H was fine, so if that's all they had actually ready at launch that's all that they should have launched. And I'm not demanding anything right now, but the implication that a fix is over a week away isn't really acceptable to me. Something like this shouldn't take 3 weeks to fix. The Apex and Formula aren't launching before they're ready, so I don't see why the Strix should.

Raja@ASUS
10-16-2017, 02:39 PM
If you say so... I just hope that the wait for an update isn't too long.


Shouldn't be long.

Tech Hog
10-16-2017, 04:30 PM
Shouldn't be long.

I guess I'll just give it a week like I had previously planned...

Raja@ASUS
10-16-2017, 05:14 PM
Yep. Seems wise.

ernest.eedesign
10-17-2017, 05:32 AM
Shouldn't be long.

Hi Raja,

Does this mean you guys agree there seems like there is a FW bug? My comment here is just to point out several other potential customers are monitoring this issue...

I'm not stomping around angrily, I also bought 1 of these boards and although I am waiting for the 8700k, and if there is no formal statement from you guys or acknowledge/debunk in the next 1-2 weeks, I will likely return (within the 30 days) unfortunately. It's got nothing to do with branding; I bought the 8700k for playing with Overclocking (else I would have bought the 8700), and being limited by a mobo FW bug seems disappointing.

It's easy to say "Oh Asus will not care about the 1 unhappy customer," but the Overclock3d video now has 25,000+ views in just 3-4 days... That's real potential revenue loss for you guys, so I'm glad you're taking it seriously... As a customer I am obviously rooting for you guys to get it solved.



...Also as a fellow senior and highly experienced Hardware Electrical Engineer (PCB, mixed signal both Analog and Digital), I tend to disagree that risk of this problem is low. I'll be blunt; I know little about Overclocking; I don't design for MOBO's or Intel processors. You guys know more about LLC than I'll ever know (hence why I'm relying on your expertise)... That said I've done several PCB's with FPGA/Arm CPU's, Analog, RF, ADC, DAC, and dozens of 1k+ component count PCBA's. #1 rule is get your power supply/rails correct. Don't screw that up... Over-Voltage transients can kill any chip depending on the bandwidth, overshoot, and duration (on-chip ESD diodes really are not that good)... And I'm assuming the #'s people are posting are wrong from the SW issues like you said, but having that much ground bounce just an inch away seems pretty sketchy to me... Add more PWR/GND planes, layers are cheap, and plenty of vias :)...

Afterall we are only talking about a little cpu processor (not the 10kW 100A+ RF Generator I'm working with now :)

JustinThyme
10-17-2017, 06:30 AM
Hi Raja,

Does this mean you guys agree there seems like there is a FW bug? My comment here is just to point out several other potential customers are monitoring this issue...

I'm not stomping around angrily, I also bought 1 of these boards and although I am waiting for the 8700k, and if there is no formal statement from you guys or acknowledge/debunk in the next 1-2 weeks, I will likely return (within the 30 days) unfortunately. It's got nothing to do with branding; I bought the 8700k for playing with Overclocking (else I would have bought the 8700), and being limited by a mobo FW bug seems disappointing.

It's easy to say "Oh Asus will not care about the 1 unhappy customer," but the Overclock3d video now has 25,000+ views in just 3-4 days... That's real potential revenue loss for you guys, so I'm glad you're taking it seriously... As a customer I am obviously rooting for you guys to get it solved.



...Also as a fellow senior and highly experienced Hardware Electrical Engineer (PCB, mixed signal both Analog and Digital), I tend to disagree that risk of this problem is low. I'll be blunt; I know little about Overclocking; I don't design for MOBO's or Intel processors. You guys know more about LLC than I'll ever know (hence why I'm relying on your expertise)... That said I've done several PCB's with FPGA/Arm CPU's, Analog, RF, ADC, DAC, and dozens of 1k+ component count PCBA's. #1 rule is get your power supply/rails correct. Don't screw that up... Over-Voltage transients can kill any chip depending on the bandwidth, overshoot, and duration (on-chip ESD diodes really are not that good)... And I'm assuming the #'s people are posting are wrong from the SW issues like you said, but having that much ground bounce just an inch away seems pretty sketchy to me... Add more PWR/GND planes, layers are cheap, and plenty of vias :)...

Afterall we are only talking about a little cpu processor (not the 10kW 100A+ RF Generator I'm working with now :)

What he agreed to was he would pass it along to R&D and they would look into it. Thing is once rumor mill gets started it goes out of control whether there is any substance to it or not. This rumor was started by Tom at OC3D TV and a lot of people posting as if they are a recording of what he said. Thing is its all talk. Not a single person anywhere has posted anything to back it up. If there is any substance to it whoever is genuinely having issues should document it and present it, I see Tom didn't do that either. All hot air. Taking 5 minutes to do whatever you need to do to replicate a problem so R&D can spend more time looking at actual issues instead of conjecture would go a long way into proving or disproving anything.

10kW? Running a little weak there arent we? The linear amp on my Ham radio pumps twice that and why are you running it on 100 Volts, isnt that a bit odd? Spend more time generating that voltage than anything else. Even in a household you would want to run that on 240V so it would only be 40 amps.

ernest.eedesign
10-17-2017, 06:59 AM
10kW? Running a little weak there arent we? The linear amp on my Ham radio pumps twice that and why are you running it on 100 Volts, isnt that a bit odd? Spend more time generating that voltage than anything else. Even in a household you would want to run that on 240V so it would only be 40 amps.

Watercooled, 240V / 480V, and not 100V... 100A of RF current, continuous. Non-50 ohm environment. This does not run in a household or standard outlet (hint RF plasma).

Unfortunately this would melt your Ham radio, literally... (granted I'm sure you could pick it up a few rooms away if you're ham radio has sufficient dynamic range and able to tune to low MHz).

bloot
10-17-2017, 08:22 AM
It's not a rumor anymore, it's a real issue http://www.overclock.net/t/1638955/z370-z390-vrm-discussion-thread/190#post_26394902

And it affects other boards too, Asrock just released a beta bios fixing it https://www.asrock.com/MB/Intel/Z370%20Taichi/index.asp#BIOS

Cheers :)

elmor
10-17-2017, 09:03 AM
https://www.io-tech.fi/uutinen/paivitys-testilabrasta-coffee-laken-kayttojannite-asuksen-strix-z370-emolevyilla/

Use google translate. That will give some of the you the answer to this hoo-haa. Unfortunately, they aren't measuring from the right place, either. To get a more accurate reading, you'd use one of the MLCC caps behind the socket. You want to get as close as you can to the socket pins so that you can see how much resistive loss you get through the power plane. Review sites simply aren't equipped to do this stuff properly. The non-technical guys profess to know things they don't understand, and those that try to be technical mess it up, too.

The requested VID can be affected by the IA/AC and DC load line settings. Has been like that for a while now. I'll see if I can get a board here and check what's going on.

IO-tech got told by Asus to measure this way, which was a mistake. It will not yield accurate results. It would be best to measure at the MLCCs at the back of the socket, or even better using the on-die sense pins which are a bit trickier to find. Basically the closer to the CPU the better.

As a comparison on latest test BIOS with 1.360V manual mode. 4.7G loaded with Prime95 29.1 SmallFFTs. LLC levels are not fully tuned yet so can't reference which one yet. Probably around 4-5. CPU-Z has been relatively accurate and at most ~15mV off.

CPU-Z 1.264V
Measuring at IO-Tech indicated location 1.345V
Measuring at socket MLCC 1.253V

Raja@ASUS
10-17-2017, 09:03 AM
Hi Raja,

Does this mean you guys agree there seems like there is a FW bug? My comment here is just to point out several other potential customers are monitoring this issue...

I'm not stomping around angrily, I also bought 1 of these boards and although I am waiting for the 8700k, and if there is no formal statement from you guys or acknowledge/debunk in the next 1-2 weeks, I will likely return (within the 30 days) unfortunately. It's got nothing to do with branding; I bought the 8700k for playing with Overclocking (else I would have bought the 8700), and being limited by a mobo FW bug seems disappointing.

It's easy to say "Oh Asus will not care about the 1 unhappy customer," but the Overclock3d video now has 25,000+ views in just 3-4 days... That's real potential revenue loss for you guys, so I'm glad you're taking it seriously... As a customer I am obviously rooting for you guys to get it solved.



...Also as a fellow senior and highly experienced Hardware Electrical Engineer (PCB, mixed signal both Analog and Digital), I tend to disagree that risk of this problem is low. I'll be blunt; I know little about Overclocking; I don't design for MOBO's or Intel processors. You guys know more about LLC than I'll ever know (hence why I'm relying on your expertise)... That said I've done several PCB's with FPGA/Arm CPU's, Analog, RF, ADC, DAC, and dozens of 1k+ component count PCBA's. #1 rule is get your power supply/rails correct. Don't screw that up... Over-Voltage transients can kill any chip depending on the bandwidth, overshoot, and duration (on-chip ESD diodes really are not that good)... And I'm assuming the #'s people are posting are wrong from the SW issues like you said, but having that much ground bounce just an inch away seems pretty sketchy to me... Add more PWR/GND planes, layers are cheap, and plenty of vias :)...

Afterall we are only talking about a little cpu processor (not the 10kW 100A+ RF Generator I'm working with now :)


R&D has found the issue. Awaiting updates.

Raja@ASUS
10-17-2017, 09:04 AM
IO-tech got told by Asus to measure this way, which was a mistake. It will not yield accurate results. It would be best to measure at the MLCCs at the back of the socket, or even better using the on-die sense pins which are a bit trickier to find. Basically the closer to the CPU the better.

As a comparison on latest test BIOS with 1.360V manual mode. 4.7G loaded with Prime95 29.1 SmallFFTs. LLC levels are not fully tuned yet so can't reference which one yet. Probably around 4-5. CPU-Z has been relatively accurate and at most ~15mV off.

CPU-Z 1.264V
Measuring at IO-Tech indicated location 1.345V
Measuring at socket MLCC 1.253V

Thanks, Jon. Someone needs to learn where to measure.

Tech Hog
10-17-2017, 11:21 AM
Thanks, Jon. Someone needs to learn where to measure.

You mean whoever at Asus told them to measure there? :P

Raja@ASUS
10-17-2017, 11:56 AM
You mean whoever at Asus told them to measure there? :P

More so the person that told them, but also for everyone that measures. If you need to ask... :D

Tech Hog
10-17-2017, 12:26 PM
More so the person that told them, but also for everyone that measures. If you need to ask... :D

True. I think I'll order a new DMM just in case. You never know after all.

Fingers crossed for a fix this week!

Raja@ASUS
10-17-2017, 12:29 PM
True. I think I'll order a new DMM just in case. You never know after all.

Fingers crossed for a fix this week!


Good choice. Expecting updates this week.

Tech Hog
10-17-2017, 01:41 PM
Good choice. Expecting updates this week.

Is this only affecting Strix boards btw? I've heard people saying that they think Prime-A and even M10H have the issue as well.

Raja@ASUS
10-17-2017, 01:47 PM
Is this only affecting Strix boards btw? I've heard people saying that they think Prime-A and even M10H have the issue as well.

The Prime, yes. Elmor would know about the Hero. Either way, all boards that require an UEFI update will get one.

aznguyen316
10-17-2017, 01:48 PM
Is this only affecting Strix boards btw? I've heard people saying that they think Prime-A and even M10H have the issue as well.

I really hope Asus R&D has recognized this does affect the Prime-A line as well and not only the Strix line.

I have already spoken with NewEgg and set up an RMA where they "made a one time exception" to allow for store credit refund. I have 14-days to return, but I am hoping I can just fix via BIOS update this week before I tear down the system. As long as the Prime-A has been recognized as also being affected, but I'm afraid it will be glossed over since many reports are just mentioning the Strix Z370 line.

aznguyen316
10-17-2017, 01:50 PM
The Prime, yes. Elmor would know about the Hero. Either way, all boards that require an UEFI update will get one.

Wow quick reply. I guess my fears in the previous post should hopefully be put to rest this week in regards to the Prime-A. Thanks for the acknowledgement.

Korth
10-17-2017, 01:56 PM
True. I think I'll order a new DMM just in case. You never know after all.

I prize my tried and true old DMMs. And I already have too many.

And yet I still find myself lusting at silly specs on the newest models.

Tech Hog
10-17-2017, 02:45 PM
I prize my tried and true old DMMs. And I already have too many.

And yet I still find myself lusting at silly specs on the newest models.

If I get one it'll be something cheap and basic lol

Menthol
10-17-2017, 05:39 PM
A new bios today that may deal with this, I am not sure

Memory go cheap if leaving at default's, otherwise get a Samsung B-die kit

Tech Hog
10-17-2017, 06:29 PM
A new bios today that may deal with this, I am not sure

Memory go cheap if leaving at default's, otherwise get a Samsung B-die kit

I wish it would release today lol

Tech Hog
10-18-2017, 04:37 PM
Well, looks like there's another issue with the board to note. It seems that, after sleep when running at stock settings, the all-core multiplier somehow increases to 4.4GHz and voltage increases by as much as 60mV (HWiNFO64 reading) in spikes. Temps increase as well so I doubt that it's just an error.

Nate152
10-18-2017, 04:42 PM
Hi Tech Hog

At default settings and the cpu core/cache voltage on auto your cpu will get overvolted some, you can set the core ratio and voltage manually or try disabling Multicore Enhancement.

Korth
10-18-2017, 05:10 PM
At default settings and the cpu core/cache voltage on auto your cpu will get overvolted some, you can set the core ratio and voltage manually or try disabling Multicore Enhancement.
I've always found this a little puzzling.

Intel specs are very comprehensively and rigidly defined, engineered and documented to death for consumption by thousands of IT departments and millions of consumers, almost every word and every number has it's own pdf paper filled with tables of revision notes and addenda and errata, lol.

And yet these days any motherboard by ASUS (or any other OEM) is preconfigured out of the box to deliver a little bump to Intel's exhaustively rated specs. Factory defaults, stock values, all the overclocking settings disabled and overclocking toys uninstalled. Still +50mV (or whatever) above Intel's parameters. You have to actually manually reconfigure things to operate within pure Intel spec, lol.

aznguyen316
10-18-2017, 05:39 PM
Looks like Asrock's top 4 boards (Fatility i7/Taichi/k6/Extreme4) got BOIS fix/updates now for LLC vdroop. Awaiting ASUS's update.

*edit* Strix LLC updates here:

https://rog.asus.com/forum/showthread.php?96800-Strix-Z370-UEFI-updates-UEFI-0419-0426

Ah cool, but where is the Z370-A Prime UEFI?? I guess because this is a ROG forum. Any links would be great to the Prime.

Tech Hog
10-18-2017, 05:54 PM
Hi Tech Hog

At default settings and the cpu core/cache voltage on auto your cpu will get overvolted some, you can set the core ratio and voltage manually or try disabling Multicore Enhancement.

I did disable MCE, but I hadn't set the cache multiplier. I was about to set it to 37, but then the update released so I'm just gonna install that and start testing.

Tech Hog
10-18-2017, 06:23 PM
I did disable MCE, but I hadn't set the cache multiplier. I was about to set it to 37, but then the update released so I'm just gonna install that and start testing.

So, I have my voltage set to 1.29V, and LLC at level 5. CPU-Z is showing 1.232V... an improvement I guess?

aznguyen316
10-18-2017, 06:27 PM
So, I have my voltage set to 1.29V, and LLC at level 5. CPU-Z is showing 1.232V... an improvement I guess?

How's LLC7?

Tech Hog
10-18-2017, 06:33 PM
How's LLC7?

I'm trying 6 now and there's a massive difference. It's working, though now I see the harsh limits of my cooler. :(

But yeah, this is much better. I'm at 4.9GHz @1.3V now and could probably push it to 5GHz if I use an AVX offset.


EDIT: Also, you can't leave power limits on auto with this bios if you're using adaptive or offset voltage. You have to set a limit manually or it defaults to the 95W TDP and you'll power throttle.

meekrophone
10-18-2017, 08:22 PM
How's LLC7?

Level 7 is overkill on my Maximus X Hero WiFi. Sensors show +50mV over desired setting.
Level 6 is around +10mV for me, nearly perfect spot

bloot
10-18-2017, 08:44 PM
Vdroop issue corrected in 0419 BIOS. However I need much more voltage now for the same frequency, so I reverted back to 0410.

illusiveguy
10-18-2017, 11:07 PM
So no BIOS update for the Strix-H?

I thought the only difference between the F and H was RGB lighting.

bloot
10-18-2017, 11:08 PM
OK nevermind, it seems this bios has modified package power limits, adjusting Long Duration Package Power Limit has resolved my vcore problems on 0419.

elmor
10-19-2017, 04:27 AM
Looks like Asrock's top 4 boards (Fatility i7/Taichi/k6/Extreme4) got BOIS fix/updates now for LLC vdroop. Awaiting ASUS's update.

*edit* Strix LLC updates here:

https://rog.asus.com/forum/showthread.php?96800-Strix-Z370-UEFI-updates-UEFI-0419-0426

Ah cool, but where is the Z370-A Prime UEFI?? I guess because this is a ROG forum. Any links would be great to the Prime.

Prime Z370-A here http://www.mediafire.com/file/by8182ybkl99oc9/PRIME-Z370-A-ASUS-0419.7z



So no BIOS update for the Strix-H?

I thought the only difference between the F and H was RGB lighting.

Strix-H has a different VRM.

illusiveguy
10-19-2017, 05:00 AM
Prime Z370-A here http://www.mediafire.com/file/by8182ybkl99oc9/PRIME-Z370-A-ASUS-0419.7z




Strix-H has a different VRM.

Interesting. Only getting my 8700k in a few days, so I'm going to assume that the Strix-H is unaffected by the LLC issue then and sleep better at night.

AntonioL
10-19-2017, 05:51 AM
OK nevermind, it seems this bios has modified package power limits, adjusting Long Duration Package Power Limit has resolved my vcore problems on 0419.

That's good news. Does it mean, please, that overvoltage behavior has also be solved ? I mean if you disable multicore enhancement and revert to intel stock frequencies, does vcore comes back to its default value or does it remain higher (about 0,1V compared to other brand motherboards according to some reviews), causing overheating ? And what about vtt (or whatever it's called, maybe vccio) ?

elmor
10-19-2017, 06:00 AM
Interesting. Only getting my 8700k in a few days, so I'm going to assume that the Strix-H is unaffected by the LLC issue then and sleep better at night.

AFAIK this problem exists on -H as well and will not be fixed, it's a 4-phase board and there's only so much it can handle.

Berdugo
10-19-2017, 07:44 AM
OK nevermind, it seems this bios has modified package power limits, adjusting Long Duration Package Power Limit has resolved my vcore problems on 0419.

Hi, how did you adjust it exactly ? I mean with what value ? Sorry it's the first time I have to adjust it...

When I sync all cores to 4.7, under OCCT it will revert back to 4.3 a few minutes after the start of the test, is that the reason ?

bloot
10-19-2017, 08:15 AM
That's good news. Does it mean, please, that overvoltage behavior has also be solved ? I mean if you disable multicore enhancement and revert to intel stock frequencies, does vcore comes back to its default value or does it remain higher (about 0,1V compared to other brand motherboards according to some reviews), causing overheating ? And what about vtt (or whatever it's called, maybe vccio) ?
Mmm haven't tested it by default sorry, can't tell you how it behaves. vccio has a bit overvoltage, but I have setted it manually to 1.2375 (1.272 readed on hwinfo)


Hi, how did you adjust it exactly ? I mean with what value ? Sorry it's the first time I have to adjust it...

When I sync all cores to 4.7, under OCCT it will revert back to 4.3 a few minutes after the start of the test, is that the reason ?

Most probably the same issue, just take a look at the lines from hwinfo64 "Core #X Power Limit Exceeded" if you see yes, then it's the same that happened to me.

Here's a picture of my settings (you can ignore the ia ac/dc, those values are needed for my offset vcore)

https://i.imgur.com/vHVVrQF.png

ernest.eedesign
10-19-2017, 09:06 AM
So confirmed the new UEFI has solved this issue?

Should I flash Bios update before setting up my OS, or OS before Bios? Never had to update Bios before; I assume it's easy. CPU arrives next week.

If LLC issue is indeed solved, thanks Asus for taking care of this quickly.

Raja@ASUS
10-19-2017, 09:07 AM
If you're having to set the long duration power limit, which CPUs are you guys using, please?

bloot
10-19-2017, 09:10 AM
If you're having to set the long duration power limit, which CPUs are you guys using, please?

8700K@4.8GHz here :)

I didn't have to set it on previous bioses.

Raja@ASUS
10-19-2017, 09:22 AM
reported it back.

elmor
10-19-2017, 10:24 AM
8700K@4.8GHz here :)

I didn't have to set it on previous bioses.

Can you detail which board and exact settings when your power limit is kept at default? Saving a profile and settings would help as well. Could it be that you're selecting XMP then choosing to not enable "Sync all cores" when asked? This setting would be kept until you restart or re-load defaults (F5).

bloot
10-19-2017, 11:04 AM
Can you detail which board and exact settings when your power limit is kept at default? Saving a profile and settings would help as well. Could it be that you're selecting XMP then choosing to not enable "Sync all cores" when asked? This setting would be kept until you restart or re-load defaults (F5).

Z370-F

Yes, XMP on and MCE disabled (answered 'no' when asked). But I adjusted it the same way on previous bioses.

Settings with power limits defined https://www.dropbox.com/s/s2i5ujkkubad9cu/powerlimitno_setting.txt?dl=0
And this is the bios config if you want to load and try it https://www.dropbox.com/s/l5uchy9law7hlhb/powerlimitno.CMO?dl=0

No power limit exceeded on hwfinfo

https://i.imgur.com/gPPN5TO.jpg

Settings without power limits defined https://www.dropbox.com/s/dg8ly3eww7xpjg0/powerlimityes_setting.txt?dl=0
And this is the bios profile if you want to load and try it https://www.dropbox.com/s/d3z9njzufx9ib0j/powerlimityes.CMO?dl=0

Power limit exceeded can be seen as soon as the stress test is started (and it fails because of this)

https://i.imgur.com/I1S81sQ.jpg

Hope this helps :)

Tech Hog
10-19-2017, 12:55 PM
reported it back.

I had the issue too, though for some reason I didn't have it when testing initially at manual voltage in Aida64 with the FPU test. The CPU package was sitting around 150W (8700K @4.9GHz @ 1.3V in BIOS and 1.264-1.28V in CPU-Z with LLC at 6 on the Z370-E) for that whole test. Once I switched to adaptive voltage though, my clockspeeds could no longer hold up. That I found pretty weird. Granted, I wasn't looking for it, so I guess it's possible that I missed it and Aida64 didn't detect it as throttling.

Edit: Wait, I didn't have XMP on when I was testing the overclock. Maybe that's what the trigger is. I didn't turn on XMP until after getting a stable overclock.

Raja@ASUS
10-19-2017, 01:03 PM
I had the issue too, though for some reason I didn't have it when testing initially at manual voltage in Aida64 with the FPU test. The CPU package was sitting around 150W (8700K @4.9GHz @ 1.3V in BIOS and 1.264-1.28V in CPU-Z with LLC at 6 on the Z370-E) for that whole test. Once I switched to adaptive voltage though, my clockspeeds could no longer hold up. That I found pretty weird. Granted, I wasn't looking for it, so I guess it's possible that I missed it and Aida64 didn't detect it as throttling.

Edit: Wait, I didn't have XMP on when I was testing the overclock. Maybe that's what the trigger is. I didn't turn on XMP until after getting a stable overclock.


Yep, there's a gap in the rules for certain configs. R&D is looking into it.

owca6666
10-19-2017, 01:54 PM
Are the other series of z370 mobos will be getting LLC bios fix ?

Raja@ASUS
10-19-2017, 02:45 PM
Are the other series of z370 mobos will be getting LLC bios fix ?

Certain Strix and Prime models only. ROG Maximus are models are okay. Anything outside that has limitations, so that's it.

owca6666
10-19-2017, 03:12 PM
Certain Strix and Prime models only. ROG Maximus are models are okay. Anything outside that has limitations, so that's it.

So you're saying people like me who bought anything other than strix range, will be stuck with broken/not fully functional product ?

Tech Hog
10-19-2017, 04:09 PM
So you're saying people like me who bought anything other than strix range, will be stuck with broken/not fully functional product ?

You have a TUF or... Wait, TUF is the only other series, right? Well, does your board have the issue in the first place? That's the first thing that needs to be known before worrying about if something needs to be fixed. I think what he's saying is that TUF has limitations which mean that LLC won't be an issue either way.

owca6666
10-19-2017, 04:13 PM
You have a TUF or... Wait, TUF is the only other series, right? Well, does your board have the issue in the first place? That's the first thing that needs to be known before worrying about if something needs to be fixed. I think what he's saying is that TUF has limitations which mean that LLC won't be an issue either way.

Yes my board does have an issue, no matter LLC setting i always get the same vdroop.

vcore set to 1.31 LLC does literally nothing, under load voltage always drops to 1.24

Raja@ASUS
10-19-2017, 04:15 PM
So you're saying people like me who bought anything other than strix range, will be stuck with broken/not fully functional product ?


No. The capable Strix models and Prime models have been updated. The ROG Maximus boards didn't have this level of vdroop so don't need the 'fix'. Models outside that are budget class/low-end, so need a given level of Vdroop to compliment the capabilities of their power solution. There is nothing surprising about this and it has nothing to do with broken or not fully functional.

owca6666
10-19-2017, 04:59 PM
No. The capable Strix models and Prime models have been updated. The ROG Maximus boards didn't have this level of vdroop so don't need the 'fix'. Models outside that are budget class/low-end, so need a given level of Vdroop to compliment the capabilities of their power solution. There is nothing surprising about this and it has nothing to do with broken or not fully functional.

This is literally taken from this boards overview, on your website.

"TUF CHOKES

Certified military-grade TUF chokes deliver rock-steady power to the CPU, helping to improve system stability."

I fail to see rock steady power delivery to cpu on my board.

Raja@ASUS
10-19-2017, 05:43 PM
This is literally taken from this boards overview, on your website.

"TUF CHOKES

Certified military-grade TUF chokes deliver rock-steady power to the CPU, helping to improve system stability."

I fail to see rock steady power delivery to cpu on my board.

The chokes store energy, and their are numerous factors about them that can help. However, they are not the only component in the power supply.

Berdugo
10-19-2017, 06:19 PM
Mmm haven't tested it by default sorry, can't tell you how it behaves. vccio has a bit overvoltage, but I have setted it manually to 1.2375 (1.272 readed on hwinfo)



Most probably the same issue, just take a look at the lines from hwinfo64 "Core #X Power Limit Exceeded" if you see yes, then it's the same that happened to me.

Here's a picture of my settings (you can ignore the ia ac/dc, those values are needed for my offset vcore)



Thanks a lot, will try this !

illusiveguy
10-19-2017, 10:14 PM
AFAIK this problem exists on -H as well and will not be fixed, it's a 4-phase board and there's only so much it can handle.

Well that's unfortunate and would've been good to know a week ago. I thought I did a pretty decent job of comparing the boards, but I guess not.

This will be the last time I buy an Asus product. Definitely no longer trust the Strix brand.

Tech Hog
10-19-2017, 11:14 PM
Well that's unfortunate and would've been good to know a week ago. I thought I did a pretty decent job of comparing the boards, but I guess not.

This will be the last time I buy an Asus product. Definitely no longer trust the Strix brand.

Yeah, I have to be honest... I'm happy that it didn't take too long to fix it on the relevant boards, but this means that I can't recommend low-end Asus boards to anyone. I had a friend looking at TUF series, but I think I'll have to recommend him another brand. If LLC is an option in the BIOS, it needs to be fixed. Either fix it or remove the option along with any claims that the board is usable for overclocking.

owca6666
10-19-2017, 11:38 PM
Yeah, I have to be honest... I'm happy that it didn't take too long to fix it on the relevant boards, but this means that I can't recommend low-end Asus boards to anyone. I had a friend looking at TUF series, but I think I'll have to recommend him another brand. If LLC is an option in the BIOS, it needs to be fixed. Either fix it or remove the option along with any claims that the board is usable for overclocking.

I have TUF board and LLC does absolutely nothing vdroop is the same on every level.

If overclocking on low end boards isnt viable because poor quality vrm, overclocking options should be disabled not to confuse people who are on a budget like me, now i have to return the board and go with competitor as they offer much better value.

infinityloop
10-20-2017, 03:17 AM
Very disappointed to hear about the inability to fix the problem that Strix-H users face, as I bought one in anticipation of using it with a 8700k. I was not aware that a $200 CAD board is considered "Low end" by ASUS and I find it frustrating that it was advertised as an overclocking capable board on the marketing site. I will be returning the board and while I have had bought ASUS motherboards and components as far back as Wolfdale, I will no longer be considering ASUS products for my future use as I have no confidence that they will provide me with the value I'm looking for and the negative surprises that they fail to mention.

Korth
10-20-2017, 03:32 AM
Actually I'm surprised ASUS sells any Z370 motherboards for "only" $200 CDN. The "low-end" usually starts ~$300+, the "top-end" usually costs ~$600+. It looks like "low-end" and "mid-end" motherboard prices have actually dropped a little on this launch.

Silent Scone@ROG
10-20-2017, 07:51 AM
Very disappointed to hear about the inability to fix the problem that Strix-H users face, as I bought one in anticipation of using it with a 8700k. I was not aware that a $200 CAD board is considered "Low end" by ASUS and I find it frustrating that it was advertised as an overclocking capable board on the marketing site. I will be returning the board and while I have had bought ASUS motherboards and components as far back as Wolfdale, I will no longer be considering ASUS products for my future use as I have no confidence that they will provide me with the value I'm looking for and the negative surprises that they fail to mention.

Be sure to do your research into other boards in that price bracket before commiting, in that case.

AntonioL
10-20-2017, 08:29 AM
So it seems that on STRIX E & F, it is now possible to manage vdrop more accurately, and consequently to use lower vcore basis. For that purpose, what option appears, please, to be the best one between offset and adaptative ? In the settings kindly provided by "bloot", offset is prefered, but does'nt it cause BSOD at idle ?

Also, what eventually are the right values for IA AC/DC load line ? "Raja" recommends 2.10, but some people say they get better results with 0.01 like with Kaby Lake, and "bloot" uses 0.20. I don't know if a bad setting can be harmful to CPU or MB.

Thanks for your help.

Raja@ASUS
10-20-2017, 09:31 AM
Use adaptive, and then dial in the ac/dc loadlines only if needed.

infinityloop
10-20-2017, 10:12 AM
Actually I'm surprised ASUS sells any Z370 motherboards for "only" $200 CDN. The "low-end" usually starts ~$300+, the "top-end" usually costs ~$600+. It looks like "low-end" and "mid-end" motherboard prices have actually dropped a little on this launch.

Are you saying ASUS has priced a high end board to be cheaper but is unable to provide even the most basic functionality of a reliable overclock or they priced a low end board like they would a high end board?

AntonioL
10-20-2017, 10:30 AM
Use adaptive, and then dial in the ac/dc loadlines only if needed.

Thank you for your answer. I have just received the 8700k that I had ordered, and I planned to buy the Strix-f, but decided to postpone my purchase until troubles pointed by reviewers and firsts users were fixed.
The only residual concern I have is about default voltage. With previous BIOS, even with stock CPU frequencies (MCE disabled), vcore was up to 0.1V higher than with other manufacturers, and so were temperatures.
The fact that you advize adaptative rather than offset does it mean that voltages for non OC frequencies have been brought back to Intel specs ? The last review I read (guru3d) indicates that the CPU does not heat more than on competitors MB any longer, but maybe you could tell us more about what was fixed in that new BIOS, please ?

Raja@ASUS
10-20-2017, 11:13 AM
Thank you for your answer. I have just received the 8700k that I had ordered, and I planned to buy the Strix-f, but decided to postpone my purchase until troubles pointed by reviewers and firsts users were fixed.
The only residual concern I have is about default voltage. With previous BIOS, even with stock CPU frequencies (MCE disabled), vcore was up to 0.1V higher than with other manufacturers, and so were temperatures.
The fact that you advize adaptative rather than offset does it mean that voltages for non OC frequencies have been brought back to Intel specs ? The last review I read (guru3d) indicates that the CPU does not heat more than on competitors MB any longer, but maybe you could tell us more about what was fixed in that new BIOS, please ?


All I know that has been patched is the LLC and MCE being disabled by default in the 0419 releases. And my advice has nothing to do with the stock voltages. Adaptive mode is preferred because you set the max turbo voltage. With offset, the CPU requests a given level of VID that varies according the the applied ratio. Unlike offset, adaptive allows you to stipulate the max VID directly (both are affected by IA AC/DC load lines. However, adaptive is more predictable than offset). The one caveat of adaptive is that you can't set it lower than the default VID for a given ratio. You can usually work around that with a small negative offset, though.

AntonioL
10-20-2017, 12:09 PM
All I know that has been patched is the LLC and MCE being disabled by default in the 0419 releases. And my advice has nothing to do with the stock voltages. Adaptive mode is preferred because you set the max turbo voltage. With offset, the CPU requests a given level of VID that varies according the the applied ratio. Unlike offset, adaptive allows you to stipulate the max VID directly (both are affected by IA AC/DC load lines. However, adaptive is more predictable than offset). The one caveat of adaptive is that you can't set it lower than the default VID for a given ratio. You can usually work around that with a small negative offset, though.

Yes, I know all that. Without OC, offset and adaptative with offset should work the same if the purpose is to lower voltage. The problem is that negative offset is absolute rather than relative, so it can lead to too low voltage at idle. Furthermore, it is often ignored, and voltage is overridded and set by CPU anyway.
I understand you don't know if the default setting still overvolt CPU above VID at stock ratios, nor if R&D is working on it. I suppose I will have to wait until someone tests it and reports.
Thank you anyway.*

Praz
10-20-2017, 12:45 PM
Yes, I know all that. Without OC, offset and adaptative with offset should work the same if the purpose is to lower voltage. The problem is that negative offset is absolute rather than relative, so it can lead to too low voltage at idle. Furthermore, it is often ignored, and voltage is overridded and set by CPU anyway.
I understand you don't know if the default setting still overvolt CPU above VID at stock ratios, nor if R&D is working on it. I suppose I will have to wait until someone tests it and reports.
Thank you anyway.*
Hello

Without an overclock the adaptive voltage mode is non-existent. Applying an offset in that mode is identical to using the offset mode. A manually set offset voltage is never ignored and is applied throughout the VID table range. If the result is different than expected it is the because of not fully understanding the VID table parameters or the loading effects on the processor. At stock speeds the applied VCORE will fall within Intel's programmed levels if all BIOS settings are correctly set, particularly MCE.

Raja@ASUS
10-20-2017, 01:28 PM
Yes, I know all that. Without OC, offset and adaptative with offset should work the same if the purpose is to lower voltage. The problem is that negative offset is absolute rather than relative, so it can lead to too low voltage at idle. Furthermore, it is often ignored, and voltage is overridded and set by CPU anyway.
I understand you don't know if the default setting still overvolt CPU above VID at stock ratios, nor if R&D is working on it. I suppose I will have to wait until someone tests it and reports.
Thank you anyway.*


I didn't comment on the stock voltages as I have no idea what the site was seeing or thought they were seeing. And as Praz said, one needs to understand how offsets work. Small negative offsets are okay for the most part.

AntonioL
10-20-2017, 01:42 PM
I didn't comment on the stock voltages as I have no idea what the site was seeing or thought they were seeing. And as Praz said, one needs to understand how offsets work. Small negative offsets are okay for the most part.

We all three agree on how offset should work in principle !
The review I was refering is http://www.hardware.fr/articles/970-3/overclocking-pratique.html (if you can understand french) but I remember other sites observing the same.
You can read that at stock 4.3 GHz on six cores, the Asus card feeds the CPU with 0.1V more than the GB one, and as a result, the température of the CPU is reported 10°C hotter.*

Raja@ASUS
10-20-2017, 02:09 PM
As Praz said, the voltage should not be higher with no MCE, so not sure what they were seeing and why. And if you understand offset and adaptive, you should have no issue setting things up.

bloot
10-20-2017, 07:21 PM
There's a 5419 BIOS on the french website, is the power limit fixed? Changelog just says 'Sync all core'

https://www.asus.com/fr/Motherboards/ROG-STRIX-Z370-F-GAMING/HelpDesk_BIOS/

Raja@ASUS
10-21-2017, 06:24 AM
There's a 5419 BIOS on the french website, is the power limit fixed? Changelog just says 'Sync all core'

https://www.asus.com/fr/Motherboards/ROG-STRIX-Z370-F-GAMING/HelpDesk_BIOS/


That's just 0419 with ASUS MCE enabled by default.

And as for the power limit, in a future build the XMP page will state that the power limit will be fixed if you select "No" to MCE - until you load defaults or clear CMOS. In other words, if you're going to overclock the CPU cores, don't select "No" if you don't want to set the power limit yourself.

Raja@ASUS
10-21-2017, 11:55 AM
Test board arrived.

https://farm5.staticflickr.com/4496/23976766838_f25656816a_o.png (https://flic.kr/p/CwKcrG)Supreme1 (https://flic.kr/p/CwKcrG) by Rajinder Singh Gill (https://www.flickr.com/photos/152488850@N07/), on Flickr
https://photos.app.goo.gl/pRkC8kNOWtmBVwiJ3

https://farm5.staticflickr.com/4501/37780662706_6929f5e30b_o.png (https://flic.kr/p/ZyxLLj)strix (https://flic.kr/p/ZyxLLj) by Rajinder Singh Gill (https://www.flickr.com/photos/152488850@N07/), on Flickr

https://farm5.staticflickr.com/4478/23976767688_0247f584f0_o.png (https://flic.kr/p/CwKcGm)rogeye (https://flic.kr/p/CwKcGm) by Rajinder Singh Gill (https://www.flickr.com/photos/152488850@N07/), on Flickr



Testing commences next week.

illusiveguy
10-21-2017, 01:12 PM
Is 1.36v VCORE with XMP turned on okay? No other settings have been changed.

I now know I shouldn't have got the Strix H (thought the only difference between that and the F was lack of RGB lights).

Kale007
10-21-2017, 07:37 PM
Does it seem valid that the Z370-I board doesn't like LLC6 and LLC7? I get better results with LLC5 (lower voltage for stability in e.g. cinebench etc.) than those two. I would like for them to work "properly", as now i have to set the voltage up to 1.42V to reach stability. Is this normal behavior?

Tech Hog
10-22-2017, 12:42 AM
That's just 0419 with ASUS MCE enabled by default.

And as for the power limit, in a future build the XMP page will state that the power limit will be fixed if you select "No" to MCE - until you load defaults or clear CMOS. In other words, if you're going to overclock the CPU cores, don't select "No" if you don't want to set the power limit yourself.

Ah, so then I was right in my suspicion that this issue only came up after enabling XMP. Interesting.

Raja@ASUS
10-22-2017, 07:57 AM
Ah, so then I was right in my suspicion that this issue only came up after enabling XMP. Interesting.

Well, it's caused by enabling XMP and selecting no to the multi-core enhance when one intends to OC the CPU. Selecting no locks the power limit, which means you'll then have to set it manually if you decide to OC the CPU. Simple message change should clue people up as to which mode they should select.

Raja@ASUS
10-22-2017, 07:59 AM
Does it seem valid that the Z370-I board doesn't like LLC6 and LLC7? I get better results with LLC5 (lower voltage for stability in e.g. cinebench etc.) than those two. I would like for them to work "properly", as now i have to set the voltage up to 1.42V to reach stability. Is this normal behavior?


Use the mode that best suits the CPU and the OC, or reduce the OC so you can set a load voltage that you are happy with.

Kale007
10-22-2017, 08:08 AM
Use the mode that best suits the CPU and the OC, or reduce the OC so you can set a load voltage that you are happy with.

So the Vdroop is supposed to be optimal at like 0.05V? That's rather unfortunate.

Raja@ASUS
10-22-2017, 10:38 AM
Would need to test and measure thoroughly to know.

Kale007
10-22-2017, 11:14 AM
Would need to test and measure thoroughly to know.

Well I certainly hope you look into it.

Raja@ASUS
10-22-2017, 01:43 PM
At this stage, it is unlikely that things are not going to change, so use whatever works best for you.

Praz
10-22-2017, 02:56 PM
Hello

ASUS Z370-A with no LLC control. Adaptive voltage mode used with additional turbo voltage set to 1.340V and LLC left at default (auto). CPU-Z and HWINFO used for software reporting and 5.5 digit DMM with current calibration connected at CPU for physical measurements. Tuning the board for minimum VDROOP results in a average loaded steady state value of approximately 1.322V for 0.018V - 0.020V VDROOP.

While 0.020V VDROOP may not be considered ideal is is definitely something that can be worked with. As can be seen below the discrepancy between software reporting and actual voltage can be 0.015V - 0.050V or more. This margin of error makes relying on software based voltage reporting for anything more than a generalization pretty much useless.



68267

JustinThyme
10-22-2017, 03:55 PM
@Kale Praz is with ASUS and showing you that the performance of what you get with an earlier version is well within range and margin of error. Its also is state that while Vdroop is not optimal its workable and what you get from the budget minded boards.

Thank you Praz for ending this nonsense started by a youtube buffoon who obviously has read and applies excerpts from the book...If you cant dazzle them with brilliance then baffle them with BS.

What most are failing to accept here is you bought an economy low end motherboard and are expecting the same results as the high end Maximus boards. If that were the case there would be no sales of high end boards. It equates to buying a Hyundai Accent and expecting to run it against a Bugatti Veyron. This level of board is aimed at the limited budget/frugal consumers who do not expect to get extreme overclocking and are just running at or close to stock speeds. Are there boards that are even cheaper? Sure but that doesn't make them any better. The old adage you get what you pay for certainly applies here. Take the entire line up of the Z-370 ASUS offerings and arrange them by price point. The low end starts around $189 USD and at this point the high end is the M10 Apex at $349 USD. The Extreme hasnt been launched as its always late to the dance and will come in surely in the $550 USD price range.

So just taking whats at hand am I correct to assume people are expecting to get the same performance out of the low end boards as those who pony up and pay for a board that has better components, more R&D, stricter guidelines etc? Im not knocking those on a budget, just saying don't expect the best experience of the game from the cheap seats and stop taking what what OC3D says as gospel. He presents a point of view, nothing more and honestly a 20 minute video of talking with nothing to back up it even rationalize the rant is demeaning and voids any credibility.

Praz
10-22-2017, 04:22 PM
Update to the latest BIOS version before coming here complaining, some have real problems here.
Hello

I'm not complaining about anything. And the BIOS version is irrelevant to the content I posted. Without tuning I was seeing 0.050V - 0.060V VDROOP. With proper tuning under load the DROOP does not exceed 0.020V and there is no voltage excursions above the value set in the BIOS. This will apply to any BIOS version having issues with LLC. What is also obvious is whining about voltages when relying on nothing more than reported software values is a waste of everyone's time reading such posts and does nothing to lend to the credibility of those making the posts. This thread is now 13 pages long and the only thing relevant posted thus far by users is the fact that LLC was broke on some boards. Most everything else is noise based on erroneous assumptions and faulty data.

Kale007
10-22-2017, 04:40 PM
Hello

I'm not complaining about anything. And the BIOS version is irrelevant to the content I posted. Without tuning I was seeing 0.050V - 0.060V VDROOP. With proper tuning under load the DROOP does not exceed 0.020V and there is no voltage excursions above the value set in the BIOS. This will apply to any BIOS version having issues with LLC. What is also obvious is whining about voltages when relying on nothing more than reported software values is a waste of everyone's time reading such posts and does nothing to lend to the credibility of those making the posts. This thread is now 13 pages long and the only thing relevant posted thus far by users is the fact that LLC was broke on some boards. Most everything else is noise based on erroneous assumptions and faulty data.

Sorry, I'm completely in the wrong here. I'm just frustrated in confusion as to why my board's LLC6 and LLC7 don't work. Pardon my ignorance, please.

Raja@ASUS
10-23-2017, 07:19 AM
Thread closed. If you have questions regarding setup, please open a new thread and detail the issue.