I remember something like that: Got a X570 Taichi/3700X combo, all OK with BIOS version 2.x. Later I got a 3200G for the testing of the HDMI out of the board, and after placing it in the Taichi the BIOS stated its version to be 1.x (including reverted menu options (SATA Hotplug options) that had been changed in version 2).
I still haven’t comprehended how that’s possible, the Taichi does not have a backup BIOS to use for a restore…
I understood the option as “Disable the remote management completely”. So might there be a bug when having “Wait for BMC” enabled and the board is waiting for the BMC when starting, but the BMC isn’t started which will result in an endless loop?
The BMC is a full linux system (you can extract the system using binwalk from the binary). The BMC is writing it’s settings to a different config “partition” in plaintext files which are symlinked to /etc in the main linux system. Maybe this partition isn’t cleared when removing the CMOS battery?
Do you have a way to reflash the BMC? But I think you’ve done this by replacing the BMC and BIOS flash chips, have you?
Is it possible to get a signal through the serial port of the board? Maybe they’ve implemented it?
What does the debug LED show? Is the board booted successfully? I currently don’t rember the code when the BIOS is waiting for the BMC, maybe some one else is remembering it?
Did ASRock officially remove the ECC support for non Pro CPUs?
I just saw in the datasheet that they’re writing:
“For Picasso Ridge and Raven Ridge CPUs, ECC is only supported with PRO CPUs.”. This hint isn’t written in the datasheet of the 2T board.
Maybe this hint exists since the release and I’ve simply read over it?
ASRock doesn’t officially support ECC for any non Pro Ryzen CPU. This is the same across all manufacturers as well. It should still work but there’s no “official” support.
You have to differentiate between APUs (CPUs with integrated GPU) and the pure CPUs.
For some unkown reason AMD decided to disable ECC on the non-Pro APUs (Raven Ridge/Picasso) but ECC isn’t actively disabled (cp. Intel) on the CPU-only SKUs (currently Zen 2/Matisse) if the motherboard manufacturer implements it in its design (additional traces for DIMM slots, BIOS etc.).
You can find some 1st and 2nd gen Pro Ryzen CPU’s on eBay. They’re usually used and a decent more expensive (not a lot but still noteworthy). I’ve yet to see any Zen 2 Pro CPU’s for sale.
Replaced the CPU with a 2600X, no change in behavior.
The LED debug code seems to go through its motions normally (with the countdown clock period), ending with “Ab”.
At least I got an express option at the retailer with the purchase of the motherboard meaning a new one is shipping out as soon as they receive the defective unit so it might not take that long to get a replacement.
When the board displays “Ab” it seems to be in the UEFI setup. You mentioned your keyboard is working. Have you tried to reset the “Out of Band access” blindly?
My board is in production, but maybe somebody else can give you the key combination to change the value?
(Although it should already be set to default, but explicitly set it to enable might work?)
I’ve been using an active external VGA->HDMI adapter that worked in the past with multiple X470D4Us and still works (just tested it) on a different GPU with a DVI-I output (GTX 950).
The native resolution of the display used was 1920 x 1080.
As a last straw I tried an 11-year old display with a native analog VGA input and 1280 x 800 native screen resolution…
…and with that the ASPEED VGA could output a signal again and you could reset the BIOS and BMC settings via direct BIOS access and after another reboot the IPMI is again appearing in the network environment.
Can confirm that swapping the BMC flash chip after a functioning BMC image has been loaded completely actually works for flashing the substitute chip
Open the socket, use non-conductive tweezers and carefully replace the chip.
The BMC stays operational with all settings loaded from the chip that had been present when the PSU got connected to power.
Go to Maintenance -> Firmware Update (make sure not to keep/save previous settings). After uploading the local image from your client it checks the existing/older version on the new chip showing that it recognizes the replacement chip.
Also found out that (at least some) BMC settings are actually permanently stored on the flash chip so it made sense that a CMOS reset didn’t help a bit when the BMC settings were messed up.
hi
ty for the fan script.
It should work for freenas to if i install python in a jail und run the script from there or?
sorry if it is a stupid question, i dont use freenas yet but i wanted to change my existing server to this board, but i think i will wait a bit longer till the x570 version of the x470d4u comes at the end of this year.
if the x570 version uses the same ast2500 it should work on that board too or?
i am just a bit confused that asrock wrote me that the x570 will be mini-ITX and not a m-atx as the 470.
in the memory qvl there is a samsung 32 gb ecc modul named M391A4G43MB1-CTDQ.
i can only find the same without the q at the end.
are these the same modules or is there a differens between M391A4G43MB1-CTDQ and M391A4G43MB1-CTD ?
I don’t think you’ll be able to access /dev/ipmi0 (or the FreeBSD equivalent) from within a jail. I don’t think the other fan FreeNAS control scripts are designed to operate within jails though.
The script should work with any AST2500 BMC.
Those memory modules are correct. I think the suppliers just don’t have the Q on their part numbers for some reason. I’ve ordered modules from 2 different suppliers showing M391A4G43MB1-CTD on their site and the sticker on the modules shows M391A4G43MB1-CTDQ.