@aBav.Normie-Pleb, is you debug LED lighting up when the PSU connected to power, my seller said it should be lighting up when PSU is connect to the power
No, the Debug LED is not lighting up by just connecting power to the PSU. However two green LEDs on the motherboard should light up and after a while one starts blinking.
The Debug LED only lights up when the system is properly powered on.
Note: I have only the non-2T version of the motherboard but I donāt see any reason why this behavior should change.
Important note: Be aware that CPU fan control is broken in the public BIOS 3.04 (compare the postings in this thread), I would contact ASRock Rack support and get version 3.09, NOT 3.11)
Regarding troubleshooting:
After the BIOS flash process Iād use the IPMI to reset everything to factory default settings and after that re-flash BIOS and the BMC firmware - maybe another factory default settings reset after that.
Over many years Iāve seen weird BIOS issues and sometimes it helped to re-flash the same BIOS version twice in a row with a factory/CMOS reset between the flash processes, so now I always do it this way to avoid potential trouble from even surfacing.
This helped me once when for unkown reason the motherboard would not boot properly.
thanks , then it might not my PSU issue, I do have 2 green LED light up as you said, I will try to use IPMI to reset everything to factory default setting tomorrow to see if it can help, again, thanks for your help
Please report back how the motherboard is working then for you, what exact PSU model you are using and if you have CPU_PROCHOT IPMI log entries.
So far we do not know if the X470D4U2-2T is also affected by the issues detailed in this thread but unfortunately since these models base on the same hardware design I think itās likely.
I also tested an Intel X550 T2 card in PCIe 5 of my X470D4U models (essentially making them into X470D4U2-2T) and the issues Iāve experienced did not change.
Edit: Just saw that the X470D4U2-2T has different/earlier BIOS/BMC firmware versions so I donāt know if all the bugs mentioned in context of the X470D4U and BMC 1.50/BIOS 3.04 et. al also applies to your X470D4U2-2T.
Noctua are usually very good about sending brackets and such, should just have to message them.
Help will always be given to those at Hogwarts, whom ask for it.
Have Asrock provided any further information on when a newer BIOS might be available that addresses some of the issues? I would be okay with purchase of a new PSU just for this project but am concerned with comments that ESXi 6.7 canāt even be loaded with newer non-public BIOSes.
I realize that this thread is for discussion of the Asrock board but Iām at a decision making point here as for not a whole lot more than I would spend on this board and associated bits and pieces I can get a refurbished server with a warranty, enterprise grade hardware, etc. I do know that I would have more flexibility for tweaking and lower power consumption with a DIY solution like this but itās tough.
Latest bone Dell has thrown me is R640 w 48Gb RAM, 4116 CPU, 2x240GB SSDs, PERC H330 controller, dual 495w hot swap PSUs, 2x10Gb + 2X1Gb NIC and assorted bits and bobs for $2100 which seems like a pretty fair deal. The only issue is Iām not sure if I would run into insurmountable challenges getting a PCIE GPU to work in pass through mode under ESXi in such a server.
Yeah Iāll email them and see if I can get 2 sets, gonna get another cooler once I rebuild my NAS.
Iām also going to test a graphite thermal pad, seeing my rig runs 24/7 and it gets damn hot here thermal paste has a tendency to dry out quick. Iāll report back on how it does when I do it .
Joining in this discussionā¦
I bought the X470D4U (non 2T) and am running a 3900x as a home lab. Iām also running 3.04 bios (borrowed a first gen Ryzen 1700 to flash the board from 1.50 that it shipped with) This board seems to be finicky from day one. Iām running ESXi 6.7 u2. Every day or so (sometimes less), ESXi and all the VMās completely freeze. ESXi is stuck at its grayed out screen and keystrokes donāt register even sitting in front of the machine (not via IPMI remote KVM). IPMI itself is still responding fine and I can reboot the host from there. I also am getting āno asserationā with CPU temperatures. (Though I was with the old 1.5 bios with the 1700x) Not sure whatās causing ESXi to hard lock. Iāve tried disabling c-states. The log doesnāt seem to be super descriptive. I am passing a Quadro P2000 to my plex VM that that seems to function fine (until it freezes). I swapped my SSD from a Samsung 950 pro that I had lying around to a 860 EVO, and that seemed to last longer anyway between freezes anyway. (The 860 Evo is fresh out of the box). I see that most of the issues are with CPU_PROCHOT IPMI in the IPMI logs. Iāll double check my own logs to see if I can find something like that. It sounds like an updated bios is imminent? 3.09? 3.11? I sure hope so.
Thanks for the first hand experience @JackOfAllTech thatās exactly what I needed to hear. Not what I wanted to hear but what I needed to hear. Do you think the GPU pass through is what is killing things? You might take the GPU out for a week and see if the stability improves.
As neat as this project is I donāt know that I am ready to take the plunge since I need some stability under ESXi.
I suppose that I could look into running a different hypervisor but I have an ESXi essentials license with a host available to use on it so Iād have the enterprise capability of ESXi and could still do things like Veeam backups of my VMs if I needed to migrate things.
That Dell R640 is looking more tempting right now. The only thing holding me back on that one is some kind of assurance that a P400 would work with passthrough which of course Dell canāt provide.
@jmpage2 Yeah thatās one of the options Iām going to try next, just doing CPU transcoding instead of passing the GPU through. Iām thinking its something else though as the Plex VM works anytime I access it. ESXi āfreezesā randomly. If you need ESXi stability, then yeah probably get the Dell. Just prior to this machine, I tried a Dell Powerdge R730 and it was just too loud and hot and power hungry for my needs. Additionally even dual 2680 v3ās youāre talking a total passmark for both CPUās of 37,130 whereas the 3900x alone puts up almost 32,000. Obviously thereās a huge generational difference, but with power consumption and head and noise a factor in my home, I went this route. Though I definitely have some bugs to work out, i feel long term, this will be the better fit.
Iām still considering going this route for all of the reasons you stated. I could spend less on consumer grade gear and get something that is more powerful and uses less power, quieter, etc.
Doesnāt do me a lot of good though if I canāt get the needed uptime.
after I taken out the battery and clear the CMOS, the system finally got posted, now running memtest86 now to tes the memory
Iām use 2 of KSM26ED8/16ME, it is ECC memory, but I just notice it shows ECC enable: N/A , I donāt remember there is any ECC related setup in bios, will use dmidecode to check later
Most likely the issue is the current BIOS isnāt up to AGESA 1.0.0.3 for the Ryzen 3000 series processors. 1.0.0.1 was a NIGHTMARE and basically just didnt work, so it may be that the BIOS you need to bug AsRock to send would aleviate that issue
Thanks for that update with the BIOS battery - only had to do that once in the past with an Intel C236 motherboard (S1200SPLR) to get it to boot again.
BIOS/BMC UPDATE for X470D4U
Who is feeling lllllucky?
BIOS 3.10
BMC: 1.60
https://www.asrockrack.com/general/productdetail.asp?Model=X470D4U#Download
newegg done lost their mind againā¦
These things are either selling like hot cakes or are just rare.
X470D4U - Bug report for BIOS 3.10/BMC 1.60
-
No DIMM temperature readings from booted OS (the IPMI DIMM sensor listings seem working)
-
No DIMM SPD details readable in booted OS
-
No CPU fan speed RPM sensor readable from booted OS but more importantly it is changing depending on actual CPU temperatur which as you learned unfortunately isnāt something you can safely assume.
-
CPU_PROCHOT-affected users: After about 30 min of system power on time CPU_PROCHOT is back
X470D4U_SN-Unit_1-2_2019-08-07_BIOS_3.10_BMC_1.60_CPU_PROCHOT_SELLog.txt (5.3 KB)
Can you guys confirm that or have different experiences?
whats the perferred update order ?
BMC > BIOS
BIOS > BMC
Since I also have firmware PTSD I always do BMC first and flash everything twice, with a restore to factory defaults between the flash processes.
But so far, CPU_PROCHOT has not been triggered with the āproblematicā Seasonic G-450 PSU. (knock on hard wood)
Asrock send me a link for 3.11 bios, but as the feedback 3.11 has issues, Iām not sure If update to it or not