Motherboard GIGABYTE AB350N-Gaming WIFI

[Acalanata]

Success. Both sticks at CL14 3200mhz and CPU at 3.9 @ 1.35v
Now to run some synthetic benchmarks..........but first.....I need a break.
Ty guys<3

Nevermind, you asked.

 

[FancyKiddo]

Alright, gave up on modifying my Zalman to work with the case, so I worked with the case to fit my Spire.


Had to bring my OC down to 3.3GHz due to restricted airflow and new, non-set thermal paste. Might be able to get it up against after the latter part is sorted, but I found something interesting: I couldn't get the PC to reliably pass stability testing when using a manual undervolt to 1.15V, but I've found that using offset mode (setting the undervolt to "normal" and then modifying the value beneath it) let me get things stable a lot lower. I have no idea why that is, but I'm happy with my 1.1V undervolt!

 

[grumpyrobin]

Ok, sorry for using this thread for troubleshooting but you guys are best source for everything ;p.

Ok, so I was going to run some benchmarks, but wanted to get windows updated to the latest creators update.
....
24hrs later
...
I am stumped.
I cannot update windows to creators update.
It gives an error right at the end.
I tried reinstalling windows but the same issue even if I run the windows upgrade assistant right off the bat.

Is this normal for Ryzen systems?
Or should I make new installation media because the usb stick that came with my retail windows is corrupt?

 

[djcostcosamples]

What kind of error are you getting?

 

[Ricdeau]

Need the error it gives for more details. However, I'd just recommend making new install media with 1703 rather than dealing with 1607 and then upgrading to 1703 as it would just save you time assuming you haven't gotten all of your additional software installed.

At work I've had numerous PCs that produce errors when running the 1703 update so I'd say it's not abnormal. There's been various causes even on fresh installs that were to be updated immediately. In the end it was faster for us to rebuild our MDT image that we use to reload OEM operating systems on PCs we deploy to our clients rather than try to pin down all the various causes. Now if you have all your software loaded up that can be a different story.

 

[FancyKiddo]

I am stumped.
I cannot update windows to creators update.

Okay, here's the solution (had the same problem myself). Download the Media Creation Tool (second option):
https://www.microsoft.com/en-us/software-download/windows10
Choose to download the ISO for a DVD. Then right-click the ISO, choose Mount, and run the install from there.

 

[grumpyrobin]

At this point I don't care for my software as I've already reinstalled multiple times.
The error was 0xc1900208.
Windows Update Troubleshooter "fixed" a couple of issues, but if I run the update, it gives the same error, and troubleshooter "fixes" the same problems over and over...

I'll do an edit once I finish installing.

edit:
says uninstall this app beause it is not compatible with windows 10.
Intel Ethernet driver netwtw040.sys plus 16 cores present on device

Edit
Success

 

[confusis]

3733MHz... https://valid.x86.fr/uda43e

 

[grumpyrobin]

3733MHz... https://valid.x86.fr/uda43e

Daamn,
Is it stable? the guy at LAB501.RO said 3600 was stable in Prime but not 3733.
Did you mess with all the subtimings or just the ones that the link shows?

 

[confusis]

Stable in 3DMark benchmarking! Have yet to Prime it.

Just the ones the link shows - 18-18-18-18-38. 1.4V.

BTW this is in BIOS F2, have yet to update to F3 as well

 

[grumpyrobin]

Hmm. ok some findings from my board.

I tried the same settings as confusis but it did that click sound and reverted to 2133mhz like earlier.
I did some testing with my other 3200hzcl14 which works fine, but any voltage changes. even just changing the voltage from auto, to 1.35v causes the system to not boot and reset to 2133mhz.

I can also confirm I have the same issues you guys reported with cpu voltage. No settings I apply change it.
I CAN live with this fine though.

On a side note:
I am running 3333cl14 now happily.
I tried 3433cl14 and it boots fine....but the GPU does not output video.....whats up with that? (I got around it with cmos clear)

 

[FancyKiddo]

I can also confirm I have the same issues you guys reported with cpu voltage. No settings I apply change it.

I'm curious where we're seeing this: is this only through software-based overclockers? I've noticed Ryzen Master and Gigabyte's utilities only change voltages in some monitoring software, if at all, but I've had no trouble setting voltage in the BIOS.

 

[grumpyrobin]

I'm curious where we're seeing this: is this only through software-based overclockers? I've noticed Ryzen Master and Gigabyte's utilities only change voltages in some monitoring software, if at all, but I've had no trouble setting voltage in the BIOS.

I set a voltage in the in the bios, reboot. and its the same voltage showing from that side panel. am I missing a setting for enabling CPU voltage changes?

 

[FancyKiddo]

Hmmm, have you tried examining voltages in the OS? I've never looked at the side panel in the BIOS for reliable information.

 

[grumpyrobin]

hwinfo, and ryzenmaster just show 1.35 or 1.344. regardless of setting.

 

[FancyKiddo]

Hrmmm. Well, I didn't do anything special to get voltage control working. Have you tried both offset and manual mode?

 

[grumpyrobin]

Oh I was only doing the direct voltage control but I messed with dynamic range a bit just now and I think I figured out the issue with people not being able to change voltage.

So I was messing with the bios and was only using the side panel to check my voltages and clocks.

So as soon as you make a clock change to the cpu it defaults to a voltage of 1.416 ish in the side panel. And if you actually set the manual voltage to 1.35 and save and exit. The side panel would still show 1.416. So I tried 1.30 and rebooted and surprise the side panel now shows 1.356 ish.


So at this point you are thinking. Hmmmm either the bios has a slight gap of +o.050 to what you tell it to do or that it is simply displaying it higher.

So therefore I left the voltage to 1.30 and took the clock to 3.9 ghz and rebooted and I could get in the bios and check the 1.356 voltage in the side panel just fine.

So let's go to windows. Everything goes smoothly but 10 seconds in the desktop everything just goes black. System still running.

So I thought.......hey. What if the side panel IS correct but just that it keeps that higher voltage for the bios and boot time only and then goes to your actual preset.

I go back to bios and set the clock to 2.9ghz. Side panel still showing 1.356 even and the preset still being 1.30 and boot and go to windows fine.

And I go check hwinfo. And surprise surprise . For the first time ever it actually reports something different from the 1.35 I am used to and actually shows 1.294 . Up until his point I thought my board was only displaying 1.35 and I was really scared it was wrong but this cleared it.

Now, back to my theory that the boot has a higher voltage.
With all the stuff I explained above it still leaves quite a bit of room to make it just a possibility.

But then I compared the VRM mos temps shown in hwinfo and the bios( you can check it from the system fan settings in mit)
And compared them and they match. At 1.30 it's 43 ish, at 1.356 it's 52-56 ish and at 1.416 it's 69-75 ish.
So if the VRM temps displayed by both bios and hwinfo match then they must be both reporting the correct voltage right?

So I guess this backs my theory a bit.

What do you guys think/can check?

 

[grumpyrobin]

Also I have one more question guys. Right now I'm testing prime 95 blend mode stability for the first time in my life.
And I am looking at hwinfo cpu and VRM temps. But I noticed something weird. As soon as I run it it reduces cpu voltage by 0.05 is this normal or sim default setting of it?

Edit: hold on I think my noobness got all out of proportion. So I am using cpu core voltage in hwinfo as my reference. That one shows 1.325. But if I go down to the ab 350 part vcore shows 1.416-1.448......which one should I trust?

When not in prime they match more or less

 

[djcostcosamples]

Oh I was only doing the direct voltage control but I messed with dynamic range a bit just now and I think I figured out the issue with people not being able to change voltage.

So I was messing with the bios and was only using the side panel to check my voltages and clocks.

So as soon as you make a clock change to the cpu it defaults to a voltage of 1.416 ish in the side panel. And if you actually set the manual voltage to 1.35 and save and exit. The side panel would still show 1.416. So I tried 1.30 and rebooted and surprise the side panel now shows 1.356 ish.


So at this point you are thinking. Hmmmm either the bios has a slight gap of +o.050 to what you tell it to do or that it is simply displaying it higher.

So therefore I left the voltage to 1.30 and took the clock to 3.9 ghz and rebooted and I could get in the bios and check the 1.356 voltage in the side panel just fine.

So let's go to windows. Everything goes smoothly but 10 seconds in the desktop everything just goes black. System still running.

So I thought.......hey. What if the side panel IS correct but just that it keeps that higher voltage for the bios and boot time only and then goes to your actual preset.

I go back to bios and set the clock to 2.9ghz. Side panel still showing 1.356 even and the preset still being 1.30 and boot and go to windows fine.

And I go check hwinfo. And surprise surprise . For the first time ever it actually reports something different from the 1.35 I am used to and actually shows 1.294 . Up until his point I thought my board was only displaying 1.35 and I was really scared it was wrong but this cleared it.

Now, back to my theory that the boot has a higher voltage.
With all the stuff I explained above it still leaves quite a bit of room to make it just a possibility.

But then I compared the VRM mos temps shown in hwinfo and the bios( you can check it from the system fan settings in mit)
And compared them and they match. At 1.30 it's 43 ish, at 1.356 it's 52-56 ish and at 1.416 it's 69-75 ish.
So if the VRM temps displayed by both bios and hwinfo match then they must be both reporting the correct voltage right?

So I guess this backs my theory a bit.

What do you guys think/can check?

Just so we are on the same page, when you are looking at CPU voltage in HWinfo, are you reading the motherboard's sensor or the CPU's sensor?

Can you see if you are able to go below 1.30v at idle when manually setting the voltage?

And to answer your second question above, yes, that is normal. People refer to it as vdroop. Load line calibration options are usually implemented in motherboards to reduce/stabilize the drop in voltage at load.

 

[grumpyrobin]

Just so we are on the same page, when you are looking at CPU voltage in HWinfo, are you reading the motherboard's sensor or the CPU's sensor?

Can you see if you are able to go below 1.30v at idle when manually setting the voltage?

And to answer your second question above, yes, that is normal. People refer to it as vdroop. Load line calibration options are usually implemented in motherboards to reduce/stabilize the drop in voltage at load.

For the first part I was looking at cpu sensor. For the second part at both.
I can now reconfirm however, regarding the first part That the even the motherboard sensor is lower than the vcore in the bios side panel. Which I now assume use the same sensor?