Aorus/Gigabyte X570 - Buildzoid Ripple tests with AORUS & ASUS

Associate
Joined
29 Jun 2009
Posts
549
Sorry it took so long for you to get a response, do you have the ticket number they passed you. I can chase why this took so long, sorry you had these issues.
Thanks for the reply. My ticket number is 743100. I drove home from work during my lunch break on Friday to take the screenshots hoping I would get a quick reply, but have had no further reply as of now.

All I wanted to know was where to probe with a multimeter to get an accurate reading (and additionally asked if the sensor over-reporting was expected), any return would be via OCUK. I don't want to send it back if its just a reporting issue, especially as there's no stock anywhere, however it does seem increasingly likely its a fault. I've called OCUK and got the RMA authorised, but haven't sent the board back yet.

Since flashing to F4L and taking the screenshots for support I've noticed it gets increasingly hard to post as you increase vDDR from 1.4v upwards without encountering no post, boot loops then auto cmos clear (this is at stock after loading optimised defaults and with memory at 2133mhz auto timings changing nothing except memory voltage). I apparently mistook this for memory instability while trying to set up my memory overclock, whereas it actually happens irrespective of timings or speed. This has pretty much convinced me its faulty now rather than misreporting, my feeling is this indicates vTT out of spec due to vDDR running high.
 
Associate
Joined
25 Feb 2009
Posts
1,045
Location
GIGABYTE UK
Thanks for the reply. My ticket number is 743100. I drove home from work during my lunch break on Friday to take the screenshots hoping I would get a quick reply, but have had no further reply as of now.

All I wanted to know was where to probe with a multimeter to get an accurate reading (and additionally asked if the sensor over-reporting was expected), any return would be via OCUK. I don't want to send it back if its just a reporting issue, especially as there's no stock anywhere, however it does seem increasingly likely its a fault. I've called OCUK and got the RMA authorised, but haven't sent the board back yet.

Since flashing to F4L and taking the screenshots for support I've noticed it gets increasingly hard to post as you increase vDDR from 1.4v upwards without encountering no post, boot loops then auto cmos clear (this is at stock after loading optimised defaults and with memory at 2133mhz auto timings changing nothing except memory voltage). I apparently mistook this for memory instability while trying to set up my memory overclock, whereas it actually happens irrespective of timings or speed. This has pretty much convinced me its faulty now rather than misreporting, my feeling is this indicates vTT out of spec due to vDDR running high.

Okay, I will look into the ticket number and see why you don't have any reply yet. If you want to RMA the board again I will get this picked up and arranged ASAP at our cost due to the issues you have had. Mail both [email protected] and [email protected] and I will get this looked into.
 
Associate
Joined
29 Jun 2009
Posts
549
Okay, I will look into the ticket number and see why you don't have any reply yet. If you want to RMA the board again I will get this picked up and arranged ASAP at our cost due to the issues you have had. Mail both [email protected] and [email protected] and I will get this looked into.
Thanks for your time, though it really doesn't seem to have helped as I still have no reply from Gigabyte support. Given how slow this has progressed so far I hope you'll understand if I don't take you up on the offer of sending it to Gigabyte, and I need to get it back to OCUK before my RMA expires so that means its back in the box tonight to be sent off.
 
Soldato
Joined
19 Apr 2003
Posts
13,513

Wow, @GIGA-Man and other vendor reps might get a heart attack from this !!!
@tamzzy wondering if you went with Asus Impact what the VRMs would be once you hit 1m population !!!
@Plec Xtreme it is then !
@orbitalwalsh i'm not sure i catch your drift?
the point i took from the video was the lack of temp monitoring on the asus boards, rather bad temps per se
I dare you to play a drinking game and down a drink, every time the guy says "actually"!

And now i've brought it to your attention - see if you can watch the video again without "actually" punching the screen.
 
Last edited:
Soldato
OP
Joined
1 Dec 2015
Posts
18,514
@orbitalwalsh i'm not sure i catch your drift?
the point i took from the video was the lack of temp monitoring on the asus boards, rather bad temps per se

cost of the boards , you'd hope they be like a fridge :D

still think they should all be naked and ran for 2 hours with prime and a heat camera . see how they get on

I know on comments HWinfo mention Gigabyte places there sensors far way, when from their public marketing slides they place them as close as you can get it
 
Associate
Joined
29 Jun 2009
Posts
549
cost of the boards , you'd hope they be like a fridge :D

still think they should all be naked and ran for 2 hours with prime and a heat camera . see how they get on

I know on comments HWinfo mention Gigabyte places there sensors far way, when from their public marketing slides they place them as close as you can get it

The suggestion that associated E-peen alone should be sufficient to cool boards this expensive really amuses me, especially when the cost of going fanless is £700 :p

I've got my board packed up in front of me here ready to take to the DPD drop off, however I had a breakthrough last night tweaking VTT manually. I've managed to get a setting it likes and its suddenly clocking much better (like 3800xmp stable enough to run an hour of HCI @ 1.42vddr), however I still have my doubts over the health of the voltage regulation on my board with regards to there being some minor manufacturing fault or electrical defect somewhere. It is clearly overvolting vDDR and cannot automatically keep vTT within its operating range resulting in occaisional no post when using 1.42v set value in bios and auto vTT, which increases in frequency as you increase vDDR. 1.47v set value will not post ever with vTT at auto. It also clearly undervolts vSOC (and vDDG) by 50mv, though I have an LLC setting for vSOC that could potentially affect both vSOC and vDDG, so that may be correctable by going above Normal LLC (or just manually adding 50mv).

I'm left on the one hand suddenly elated that I've got it clocking better (and knowing there is no stock to replace my board), but on the other I have confirmed 100% that my build has an issue with memory voltage control as I've identified the fault and how to work around it. My concern is that I'm letting this breakthrough cloud my judgement...should I now keep my faulty board in the hope that it doesn't get any worse and isn't doing anything that will cause damage to other parts if used long term (such as "fuzzy" regulation of the line that the software sensors cant pick up due to polling rates)? Sensible me says no, lazy just use it me says yes. Opinions anyone?
 
Soldato
OP
Joined
1 Dec 2015
Posts
18,514
@MrPhils

i run 1.6v through ddr4 and 1.5 volts through 2700 non x with 45mm 360 rad so prob worst person to talk about voltages and i just run them to max .

I know from samsung staff to keep it under 1.75 volts if i want it to last longer then a year when gaming

as the rule with anything in life, not 100% or sure of a product, return or RMA it if you can . See how new one works out, if same product and same fault, replace with another brand- still same issue then its out of your control .
 
Associate
Joined
16 Jun 2019
Posts
1,143
Location
Leek staffordshire
I just bought the x570 aorus pro with its 12 x 40a = 480 amps. It is by far the most expensive mobo that I have ever bought. I am sure that it will be OK for the 16 core when it comes out and I should be able to overclock that quite heavily correct?
 
Associate
Joined
16 Jun 2019
Posts
1,143
Location
Leek staffordshire
thinking about it - it has 480 amps - if this is supplied to the cpu at 1.45 volts = 696 w. Is this correct?

I have seen overclocked systems that pull 400 w at load.. I imagine that a 12 core 3900x may do this - can anyone confirm?

If we scale this up to 16 cores then that is 400 x 16/12 = 533w. - does this sound correct? if my gpu then takes another 300w then I need a 850w psu (at least). Currently I have gold 750w and a kolink 850w ungraded (less than a bronze quality) supply which is 4 yrs old. Does this sound correct?

In the meantime I can Oc my 3600 but in the future I intend to get the 16 core - perhaps I should not OC it too much?
 
Soldato
OP
Joined
1 Dec 2015
Posts
18,514
thinking about it - it has 480 amps - if this is supplied to the cpu at 1.45 volts = 696 w. Is this correct?

I have seen overclocked systems that pull 400 w at load.. I imagine that a 12 core 3900x may do this - can anyone confirm?

If we scale this up to 16 cores then that is 400 x 16/12 = 533w. - does this sound correct? if my gpu then takes another 300w then I need a 850w psu (at least). Currently I have gold 750w and a kolink 850w ungraded (less than a bronze quality) supply which is 4 yrs old. Does this sound correct?

In the meantime I can Oc my 3600 but in the future I intend to get the 16 core - perhaps I should not OC it too much?

will be fine for 1 cores :)

16 core Zen2 shouldn' pull that much, think Thread ripper is 350w at 4ghz

@Plec @tamzzy - now i want one !

 
Soldato
Joined
19 Apr 2003
Posts
13,513
right... need to get a dremel ! my cheap rotary tool won't cut fully through the steel!
Borrow one - once you've mentally absorbed ~£50 for dremel plus ~£10 for quality cutting disc you'll start justifying the x570i...

Or have you made some 'no return' cuts already in the plastic? (If so pics.)
 
Associate
Joined
25 Feb 2009
Posts
1,045
Location
GIGABYTE UK
Thanks for your time, though it really doesn't seem to have helped as I still have no reply from Gigabyte support. Given how slow this has progressed so far I hope you'll understand if I don't take you up on the offer of sending it to Gigabyte, and I need to get it back to OCUK before my RMA expires so that means its back in the box tonight to be sent off.

I haven't forgotten about this and am chasing the department in HQ to see why you didn't get any reply.
 
Associate
Joined
29 Jun 2009
Posts
549
I haven't forgotten about this and am chasing the department in HQ to see why you didn't get any reply.

I had a reply on Friday, Gigabyte support have confirmed that their board overvolts vDDR too (which I presume is the cause of the delay...a bit of an "oh crap" moment). I have also been contacted via another forum by another two users of this board who also have the exact same vDDR overvoltage issue. This positive drift on the vDDR output is causing a vTT issue as you increase vDDR. The formula for Auto vTT in bios is using 1/2 REQUESTED value, as a result it is setting vTT too low. Depending on the amount of positive drift and the memory in use this has varying negative effects on stability. You can confirm this effect by loading optimised defaults in bios, manually setting vDDR to 1.45v and touching nothing else - result is a no post until you manually configure vTT to 1/2 REPORTED value instead of Auto.

Another factor is that module density also has an effect on the amount of positive vDDR drift. 4gb sticks give little to no increase, 8gb sticks give around 25mv idle and 40mv load and 16gb sticks give around 50mv idle and 65mv load increases over requested value.

I have suggested to support in my reply that if possible they should update the vTT algorithm to use 1/2 of the REPORTED value instead of 1/2 the REQUESTED value, however that was only yesterday so I don't expect a reply until the end of the week if not next week.

TLDR: This ITX board should not be used with "Auto" selected for vTT as this value should be fairly strictly tied down to 1/2 vDDR. Auto will only be useful to get you started and will severely limit your memory oc and stability. If you have a particularly bad case of voltage drift or sensitive modules to vTT you may have issues posting even at 1.35v/XMP. I hesitate to call this a design fault, but its pretty close.
 
Back
Top Bottom