ASRock Rack has created the first AM4 socket server boards, X470D4U, X470D4U2-2T

You should be able to flash the aspeed 2500 chip from within linux. It seems asrock only makes the .exe available via their support faqs and aspeed has locked the socflash tool behind some silly registered developer access portal… The linked zip should contain tools for uefi, linux, etc.

HTTPS wwwDOTaspeedtechDOTcom/support_driver

HTTPS fichiersDOTtouslesdriversDOTcom/60176/v12000.zip

$ less -F v12000.zip 
Archive:  v12000.zip
Zip file size: 932141 bytes, number of entries: 7
drwx---     6.3 fat        0 bx stor 19-Jan-29 14:05 v12000_release/
-rw-a--     6.3 fat    65176 bx stor 19-Jan-09 10:30 v12000_release/bsdflash_v12000.tar.gz
-rw-a--     6.3 fat   135294 bx defN 19-Jan-09 10:27 v12000_release/lxflash_v12000.tar.gz
-rw-a--     6.3 fat      561 bx defN 16-Sep-14 15:04 v12000_release/readme.txt
-rw-a--     6.3 fat   268442 bx defN 19-Jan-09 10:16 v12000_release/socflash_v12000.zip
-rw-a--     6.3 fat    77673 bx defN 19-Jan-09 10:37 v12000_release/uflash_v12000_x64.zip
-rw-a--     6.3 fat   384993 bx defN 19-Jan-09 10:20 v12000_release/winflash_v12000.zip
7 files, 932139 bytes uncompressed, 930901 bytes compressed:  0.1%

Thanks for your efforts. I tried flashing with the uefi version but I’m unable because the flash chip will not be recognized. I even tried to force flash with option=f.

I dont get it.

Edit: I will return the board. I’m done.

1 Like

Does anyone have any tricks for getting a gpu passed through with this board? I’m having really strange issues. I put a nvidia GT1030 on this thing, and regardless of pcie slot i put it in my ipmi video completely breaks which is incredibly annoying. I’ve also been completely unsuccessful passing it through to any VM’s running latest proxmox and following every guide I can find online. I’m running the beta 3.39A bios. Any tips/help is greatly appreciated!

You need to set Onboard VGA to Enabled instead of Auto.

1 Like

It will be interesting to see how well if at all asrock will be supporting the new zen cpus. I think that they said they will be compatible with 470 and 550 but I could be wrong on that. Also wonder if they will need a bios update?

AMD has stated that they will not support the 400 chipsets for the 5000 CPUs mostly due to ROM size limitations on a lot of the boards. Although I’m really wondering how much it can really cost a mainboard manufacturer to use a bigger ROM in the first place… they buy those in the thousands, so it probably a cent a piece if even that…

edit:
Although their wording was a bit odd in that they also said they were going to enable partners to do it on “select beta BIOSes for AMD B450 and X470 motherboards.”… OK I guess?

edit 2:
OK the article I read only had a partial quote of their statement… the “select beta BIOSes” part was in a second statement after feedback about the first…

It’s been a few months since I wrote about the network bonding issue.
I have a little time now to sum up my findings.

So here goes:

I started with a situation where I had a bonding enabled without a way to disable it (no option in the ipmi web UI - see post 1222 )
The main point of disabling the bonding for me was a complete separation of IPMI network. With bonding enabled it would be accessible if I used the first ‘normal’ lan port - this is a security issue in my eyes.

At first I discovered that I can force disable/enable the bonding by modifying the web UI html - the option is there - it’s just disabled. Details are also in post 1222
After force enabling/disabling/etc I somehow gotten the option in the web-ui to stay visible. I am not sure why - maybe because of fixing MAC adresses (described below)

Then I contacted Asrock support about this and got some info and a tool.
The info was that they have some issues with some boards setting incorrect MAC addresses - they expected that one of the MACs on my board was set to zeros.

The tool is here:
mac_tool_from_asrock.zip (1.1 MB)
It is supposed to be able to enable/disable bonding and change/fix MAC adresses.

I got it in a binary form so instead believing what it is supposed to do and running it, I did some reverse engineering…

I managed to extract some commands that can be used to identify/fix the MAC issues. (they are listed at the bottom of this post)

As it turned out my board had duplicate MACs on IPMI side.
To make it more clear: the ports have different MACs on the IPMI and host side. My board was i this situation:

physical port: | LAN_1             | LAN_2             | IPMI_LAN           |
---------------+-------------------+-------------------+--------------------+
ipmi side MAC: | d0:50:99:e3:44:d9 | N/A               | d0:50:99:e3:44:d9  |
host side MAC: | d0:50:99:d2:d0:9c | d0:50:99:d2:d0:9d | N/A                |

And after manually fixing the duplicate (commands below):

physical port: | LAN_1             | LAN_2             | IPMI_LAN           |
---------------+-------------------+-------------------+--------------------+
ipmi side MAC: | d0:50:99:e3:44:d8 | N/A               | d0:50:99:e3:44:d9  |
host side MAC: | d0:50:99:d2:d0:9c | d0:50:99:d2:d0:9d | N/A                |

The commands I extracted are as follows:

#get MAC 0
ipmitool -I lanplus -H <IP> -U <user> -P <password> raw 0x3a 0xa1 0x00
>d0 50 99 e3 44 d8
#get MAC 1
ipmitool -I lanplus -H <IP> -U <user> -P <password> raw 0x3a 0xa1 0x01
>d0 50 99 e3 44 d9

#set MAC
#ipmitool -I lanplus -H <IP> -U <user> -P <password> raw 0x3a 0xa0 0x00  <NEW MAC>
#ipmitool -I lanplus -H <IP> -U <user> -P <password> raw 0x3a 0xa0 0x00  <NEW MAC>
#for example:
#set MAC 0
ipmitool -I lanplus -H <IP> -U <user> -P <password> raw 0x3a 0xa0 0x00 0xd0 0x50 0x99 0xe3 0x44 0xd8
#set MAC 1
ipmitool -I lanplus -H <IP> -U <user> -P <password> raw 0x3a 0xa0 0x01 0xd0 0x50 0x99 0xe3 0x44 0xd9

#detect bonding
ipmitool -I lanplus -H <IP> -U <user> -P <password> raw 0x32 0x72 0x01 0x00 0x00
>00 00 01 00 00 00 01
# check first byte: (no idea what other bytes mean)
# disabled: 00 00 01 64 00 03 01
# enabled:  01 00 01 64 00 03 01

#get bmc version:
ipmitool -I lanplus -H <IP> -U <user> -P <password> raw 0x06 0x01
> 20 01 01 90 02 bf d6 c1 00 02 02 00 00 00 00
#        1.90                      00 00 00 00
2 Likes

Anyone know how to decrease cpu core voltage on the x470d4u? Running a R9 3900x in a 2U chassis with the Dynatron A24. Currently CPU cycles between 50-70c with a few VMs at “idle.” IPMI showing vcore at 1.44v. The way the fan ramps up and down is super annoying. Also, unfortunately I don’t have an isolated area to run the fans at full blast so I need to keep this system as quiet as possible so I can sacrifice a bit of performance to keep things as quiet as possible. I’ve looked around in BIOS but can’t seem to find what I am looking for in terms of setting a negative offset. Let me know if I’m just missing a setting in BIOS.

Just use ‘AMD Overclocking’

1 Like

EDIT: Solved using an advanced “Try turning it off then on again” variant called “Reseat the DIMMs even if they are latched correctly and detected, you never know”. I guess RAM is really sensitive to the tiniest imperfect contacts.

I have added 2x M391A2K43BB1-CTD to my already present 2x M391A2K43BB1-CTD. It’s visible in Bios but it’s NOT visible by any OS or by Memtest. Tried both BIOS 3.30 and 3.37

Mobo: X470D4U
CPU: 1700

In BIOS: everything is “fine”

index

Linux confirms BIOS sees all ram:

dmidecode --type memory
# dmidecode 3.2
Getting SMBIOS data from sysfs.
SMBIOS 3.2.0 present.

Handle 0x000F, DMI type 16, 23 bytes
Physical Memory Array
        Location: System Board Or Motherboard
        Use: System Memory
        Error Correction Type: Multi-bit ECC
        Maximum Capacity: 128 GB
        Error Information Handle: 0x000E
        Number Of Devices: 4

Handle 0x0016, DMI type 17, 84 bytes
Memory Device
        Array Handle: 0x000F
        Error Information Handle: 0x0015
        Total Width: 128 bits
        Data Width: 64 bits
        Size: 16384 MB
        Form Factor: DIMM
        Set: None
        Locator: DIMM 0
        Bank Locator: P0 CHANNEL A
        Type: DDR4
        Type Detail: Synchronous Unbuffered (Unregistered)
        Speed: 2666 MT/s
        Manufacturer: Samsung
        Serial Number: 1697115E
        Asset Tag: Not Specified
        Part Number: M391A2K43BB1-CTD    
        Rank: 2
        Configured Memory Speed: 1866 MT/s
        Minimum Voltage: 1.2 V
        Maximum Voltage: 1.2 V
        Configured Voltage: 1.2 V
        Memory Technology: DRAM
        Memory Operating Mode Capability: Volatile memory
        Firmware Version: Unknown
        Module Manufacturer ID: Bank 1, Hex 0xCE
        Module Product ID: Unknown
        Memory Subsystem Controller Manufacturer ID: Unknown
        Memory Subsystem Controller Product ID: Unknown
        Non-Volatile Size: None
        Volatile Size: 16 GB
        Cache Size: None
        Logical Size: None

Handle 0x0019, DMI type 17, 84 bytes
Memory Device
        Array Handle: 0x000F
        Error Information Handle: 0x0018
        Total Width: 128 bits
        Data Width: 64 bits
        Size: 16384 MB
        Form Factor: DIMM
        Set: None
        Locator: DIMM 1
        Bank Locator: P0 CHANNEL A
        Type: DDR4
        Type Detail: Synchronous Unbuffered (Unregistered)
        Speed: 2666 MT/s
        Manufacturer: Samsung
        Serial Number: 16969DF9
        Asset Tag: Not Specified
        Part Number: M391A2K43BB1-CTD    
        Rank: 2
        Configured Memory Speed: 1866 MT/s
        Minimum Voltage: 1.2 V
        Maximum Voltage: 1.2 V
        Configured Voltage: 1.2 V
        Memory Technology: DRAM
        Memory Operating Mode Capability: Volatile memory
        Firmware Version: Unknown
        Module Manufacturer ID: Bank 1, Hex 0xCE
        Module Product ID: Unknown
        Memory Subsystem Controller Manufacturer ID: Unknown
        Memory Subsystem Controller Product ID: Unknown
        Non-Volatile Size: None
        Volatile Size: 16 GB
        Cache Size: None
        Logical Size: None

Handle 0x001C, DMI type 17, 84 bytes
Memory Device
        Array Handle: 0x000F
        Error Information Handle: 0x001B
        Total Width: 128 bits
        Data Width: 64 bits
        Size: 16384 MB
        Form Factor: DIMM
        Set: None
        Locator: DIMM 0
        Bank Locator: P0 CHANNEL B
        Type: DDR4
        Type Detail: Synchronous Unbuffered (Unregistered)
        Speed: 2666 MT/s
        Manufacturer: Samsung
        Serial Number: 16DDF1A3
        Asset Tag: Not Specified
        Part Number: M391A2K43BB1-CTD    
        Rank: 2
        Configured Memory Speed: 1866 MT/s
        Minimum Voltage: 1.2 V
        Maximum Voltage: 1.2 V
        Configured Voltage: 1.2 V
        Memory Technology: DRAM
        Memory Operating Mode Capability: Volatile memory
        Firmware Version: Unknown
        Module Manufacturer ID: Bank 1, Hex 0xCE
        Module Product ID: Unknown
        Memory Subsystem Controller Manufacturer ID: Unknown
        Memory Subsystem Controller Product ID: Unknown
        Non-Volatile Size: None
        Volatile Size: 16 GB
        Cache Size: None
        Logical Size: None

Handle 0x001F, DMI type 17, 84 bytes
Memory Device
        Array Handle: 0x000F
        Error Information Handle: 0x001E
        Total Width: 128 bits
        Data Width: 64 bits
        Size: 16384 MB
        Form Factor: DIMM
        Set: None
        Locator: DIMM 1
        Bank Locator: P0 CHANNEL B
        Type: DDR4
        Type Detail: Synchronous Unbuffered (Unregistered)
        Speed: 2666 MT/s
        Manufacturer: Samsung
        Serial Number: 16969ED5
        Asset Tag: Not Specified
        Part Number: M391A2K43BB1-CTD    
        Rank: 2
        Configured Memory Speed: 1866 MT/s
        Minimum Voltage: 1.2 V
        Maximum Voltage: 1.2 V
        Configured Voltage: 1.2 V
        Memory Technology: DRAM
        Memory Operating Mode Capability: Volatile memory
        Firmware Version: Unknown
        Module Manufacturer ID: Bank 1, Hex 0xCE
        Module Product ID: Unknown
        Memory Subsystem Controller Manufacturer ID: Unknown
        Memory Subsystem Controller Product ID: Unknown
        Non-Volatile Size: None
        Volatile Size: 16 GB
        Cache Size: None
        Logical Size: None

But it’s completely missing:

free  -m
total        used        free      shared  buff/cache   available
Mem:          32103       10035       20640         235        1428       21413

But I’m not seeing anything in the logs that helps…

dmesg | grep -i memory
[    0.000000] check: Scanning 1 areas for low memory corruption
[    0.000000] Early memory node ranges
[    0.000000] PM: Registered nosave memory: [mem 0x00000000-0x00000fff]
[    0.000000] PM: Registered nosave memory: [mem 0x000a0000-0x000fffff]
[    0.000000] PM: Registered nosave memory: [mem 0x09d02000-0x09ffffff]
[    0.000000] PM: Registered nosave memory: [mem 0x0a200000-0x0a20afff]
[    0.000000] PM: Registered nosave memory: [mem 0x0b000000-0x0b01ffff]
[    0.000000] PM: Registered nosave memory: [mem 0xd67be000-0xd67befff]
[    0.000000] PM: Registered nosave memory: [mem 0xd67de000-0xd67defff]
[    0.000000] PM: Registered nosave memory: [mem 0xd67fe000-0xd67fefff]
[    0.000000] PM: Registered nosave memory: [mem 0xd6806000-0xd6806fff]
[    0.000000] PM: Registered nosave memory: [mem 0xd7157000-0xd72b6fff]
[    0.000000] PM: Registered nosave memory: [mem 0xd730a000-0xd730afff]
[    0.000000] PM: Registered nosave memory: [mem 0xdab78000-0xdc177fff]
[    0.000000] PM: Registered nosave memory: [mem 0xdc351000-0xdc767fff]
[    0.000000] PM: Registered nosave memory: [mem 0xdc768000-0xdd465fff]
[    0.000000] PM: Registered nosave memory: [mem 0xdf000000-0xdfffffff]
[    0.000000] PM: Registered nosave memory: [mem 0xe0000000-0xf7ffffff]
[    0.000000] PM: Registered nosave memory: [mem 0xf8000000-0xfbffffff]
[    0.000000] PM: Registered nosave memory: [mem 0xfc000000-0xfcffffff]
[    0.000000] PM: Registered nosave memory: [mem 0xfd000000-0xffffffff]
[    0.000000] Memory: 32697396K/33481604K available (14339K kernel code, 2367K rwdata, 4980K rodata, 2696K init, 5040K bss, 784208K reserved, 0K cma-reserved)
[    0.003400] Freeing SMP alternatives memory: 40K
[    0.153844] x86/mm: Memory block size: 128MB
[    0.762239] Freeing initrd memory: 41156K
[    0.770282] check: Scanning for low memory corruption every 60 seconds
[    0.998750] nvme nvme0: allocated 32 MiB host memory buffer.
[    1.068455] nvme nvme1: allocated 32 MiB host memory buffer.
[    1.083660] Freeing unused decrypted memory: 2040K
[    1.087040] Freeing unused kernel image memory: 2696K
[    1.107280] Freeing unused kernel image memory: 2008K
[    1.110255] Freeing unused kernel image memory: 1164K
[    1.416617] mpt2sas_cm0: Allocated physical memory: size(1687 kB)
[   15.965722] [TTM] Zone  kernel: Available graphics memory: 16437134 KiB

What’s the next step ?

1 Like

Hey Guys, I installed Proxmox on my X470d4u2-2t and everything was fine until testing network speeds. The ports seem to give me 100 mbit speeds [11,5mb/s] !!!
Also using ethtool enp1s0f0| grep Speed returns Speed: Unknown!.
Also passing one of the ports through to a TrueNAS VM also gave me similar speeds. What is going on here ? Can someone help ?

if you read a few post above, you will find a solution, which is turn off the “keep_ncsi_link_up” in IPMI setting

Recently picked up this board and am planning a build home server around it. Here is my parts list:

  • 3950X
  • 4xCrucial 32GB ECC UDIMM MTA18ASF4G72AZ-3G2B1 - I know these aren’t on the QVL, but there aren’t very many large ECC UDIMM choices out there. I tried for the Samsung modules on the QVL, but I couldn’t find them in stock anywhere reputable.
  • 6xSeagate Exos X16 SATA drives, to be used in ZFS raidz2
  • EVGA 550 G3 PSU
  • Fractal Node 804 (from a previous build)
  • Thermalright Ultra Extreme with AM4 mounting kit (surprisingly, fits in the node 804 and should clear the RAM)
  • OS choice is currently Ubuntu Server 20.04

My original plan was to put two NVMe SSDs into RAID1 via the two onboard NVMe slots to act as the OS drive, but looking more closely I see that that may not be the best idea, with the x2 and previous-version x4 PCIe configurations for those slots. Am I better off using an M.2 card like the one in post 1237? Any problems if I wanted to RAID1 two NVMe drives using this card? Finally, if I am only using two drives on this card, is it save to use a video card in the x8 slot?

I saw in the GamersNexus server build video using this board they opted to use a controller card to connect the storage drives - is that really necessary? Are there any reliability problems with the onboard SATA ports?

I’ll report back to this thread if my RAM ever gets here (currently in FedEx SmartPost limbo between FedEx and the USPS) as to whether it works with this board. I’m hopeful, since the board was able to POST with a G.Skill module from my desktop.

Thanks in advance for any help. And thank you to everyone else in this thread who’s contributed valuable information - it was very helpful in getting me this far!

I wouldn’t expect any significant problems unless you actually needed all the speed of 4 lanes per ssd.

Not that I’ve noticed.

Hi All,

  1. Can someone post power consumption results of their build ? How is X470 holding in terms of idle power ?

  2. Does the recent BIOS support Package Power Tracking (“PPT”) or any CPU tweaking capabilities ? I am mostly interested in undervolting / underclocking the CPU and the PPT is a good start

Thanks in advance.

My idle power draw is around 57 Watts as measured by UPS
(Watts are equal to Volt-Amps in my case, as i use 80+ power supply w/active PFC)

Here is the set-up:

  • X470D4U (BIOS v3.39, BMC v2.05)
  • 350W 80+ active PFC by Seasonic
  • motherboard SATA disabled
  • Ryzen 7 2700 (limited to 45 TDP)
  • 2 x 32GB 1.2V ECC DIMMS @ 2933MHz (KSM32ED8/32ME)
  • Add-in-Cards:
  • PCIex4 Intel i350T4 four gigabit network interfaces ( ~ 5 Watt max. )
  • PCIex8 RAID Controller Adaptec ASR-7805Q RAID (~ 18.5 Watt max.)
    – 3 x HDD 1TB WD GOLD 7200 RPM, RAID 1E ARRAY ( idle at ~ 18 Watts, 24W max. )
    – 2 x SATA SSD 128G each, RAID cache
  • 5 x PWM Case fans ~ 50% duty cycle (arctic silver)
  • USB Keyboard
  • USB Mice

P.S.
There is no need to under-volt, as the standard BIOS options already provide you with the tools to achieve what you want.

In my case case I run a 65W TDP Ryzen 7 2700 at 45W TDP (similar to Ryzen 7 2700E):
I multiply the desired TDP of 45W by 1.35 = 60.75 (~ 61 Watts)

The resulting calculated TDP (cTDP) is 61 Watts

So change the following in UEFI / BIOS:

Advanced > AMD CBS > NBIO Common Options:
Determinism Slider = [Power]
TPP Control = [Manual]
cTDP = 61

There is also ‘Processor Temperature Control’ BIOS In BIOS, which I left on [Auto]. If changed to [Manual] you can set the custom temperature limit. I don’t use it as the CPU temperature is in the low 40C range. I think this could be useful with a passive heat sink provided there is a good airflow.

Before the above changes the above system’s idle power draw was in the 70@ range, but more important I often observed the spikes in power consumption of up to 121 Watts. After making the above changes, I no longer see the spikes in power consumption at normal loads.

Trying to squeeze any more energy savings from my setup would be equivalent to splitting an atom. The achieved result is good enough to stop at this point.
I like the fact that there is available processing power if i ever need it.

Hope this helps,

2 Likes

Hi,

I just bought X470D4U with Ryzen 5 3600X and Predator HyperX DDR4 3200 ram.
The stickers on motherboard indicate bios version 3.30 and BMC version 1.90.

Unfortunately the ram I ordered is incompatible (the board is showing 3B error code, tried pushing it firmy etc). Before I order RAM replacement I need to be sure the board itself is fine.

The thing that bothers me is that BMC never comes up - when the RAM is onboard dr debug shows the error code. When the RAM is completely absent no dr. debug code. But in both scenarios BMC hearbeat led never pulses or lights up and bmc lan interface is blinking it’s led, but never gets any DHCP address. Connecting VGA monitor also does not show any signs of life.

Will BMC boot up fine when I install proper RAM or should I RMA the board itself?

Things I tried:

  • changing ram slots
  • disconnect board from power, remove cmos battery, shorten cmos reset pins

Thanks in advance :frowning:

Will BMC boot up fine when I install proper RAM or should I RMA the board itself?

The BMC should come up regardless of whether RAM is present or not. How are you verifying that it is not booting up? It should show up in your DHCP server after connecting to the IPMI lan port and switching the PSU on.

  • If you didn’t try this configuration already: Put one stick in the A1 slot only and check.
  • re-seat the CPU (although the BMC should work without it too)

Do you have different sticks to check if the board can boot at all?

If you manage to boot either the BIOS or the IPMI then you can try fixing another by cross-flashing. (It could save you potentially time-consuming RMA)

Thanks for reply!
I tried all stick configurations. During each configuration I have to take off the heatsink from stock AMD cpu, because the fan obstructs RAM slot :smiley:
Unfortunately I have no other DDR4 ram here.

I verified lack of boot in two ways:

  • no VGA output
  • IPMI LAN is connected and it does not get any DHCP lease from my pfsense. I can see that the management lan blinks once ~1-2s, but even after a quarter no DHCP lease is present.

Can I try booting the board completely without CPU?

IPMI should boot up without the CPU after just connecting power. (no need to switch the host on)

One more thing: remove >all< add-in cards / M.2.

Other than that I don’t think I can be of much more help.

1 Like