Minisforum MS-01 PCIe Card and RAM Compatibility Thread

[GreenAvacado]

Few observations for i9, with Proxmox host running, 10G SFP NIC running at 1G, idle power is about ~18W. Consistent with Patrick's.

x710 NIC and iGPU works like a charm.

Question for community, how can I get Windows VM like fluid experience over RDP with video acceleration and audio for Linux VM (Ubuntu or Mint)?

Audio and video (iGPU SR-IOV) on Linux Mint works like a charm over SPICE but RDP or Remmina would be more universal. Anybody?

 

[FingerBlaster]

my earlybird units haven't come, but my amazon unit has... very annoyed by minisforum on mishandling this.

Just installed proxmox, with a single 4tb wd sn850x installed, and 96gb of ram it was drawing about 25 watts. Changing the power governor brought that down to 23. doing a powertop autoconfigure brought my idle down to 12.7 watts. no vm's running yet.

I'm having an issue with the sfp+ ports, none of the transceivers I have are being recognized.

i40e 0000:02:00.1: Rx/Tx is disabled on this device because an unsupported SFP module type was detected.
i40e 0000:02:00.1: Refer to the Intel(R) Ethernet Adapters and Devices User Guide for a list of supported modules.



But @JaxJian says these are unlocked?

Build in Intel NIC has already unlocked for SFP+ modules.

I'm using an original Aquantia aqs-107 10gbe. I also have some dell sfp modules i've tried to no luck.

 

[GreenAvacado]

my earlybird units haven't come, but my amazon unit has... very annoyed by minisforum on mishandling this.

Just installed proxmox, with a single 4tb wd sn850x installed, and 96gb of ram it was drawing about 25 watts. Changing the power governor brought that down to 23. doing a powertop autoconfigure brought my idle down to 12.7 watts. no vm's running yet.

I'm having an issue with the sfp+ ports, none of the transceivers I have are being recognized.

i40e 0000:02:00.1: Rx/Tx is disabled on this device because an unsupported SFP module type was detected.
i40e 0000:02:00.1: Refer to the Intel(R) Ethernet Adapters and Devices User Guide for a list of supported modules.



But @JaxJian says these are unlocked?



I'm using an original Aquantia aqs-107 10gbe. I also have some dell sfp modules i've tried to no luck.

Interesting. So what exactly does powertop do under the hood to lower the idle power? Turn off cores, lower turbo frequency?

Ran powertop --auto-tune on PVE host and did not notice any drop in idle power.

 

[webmaster]

Ran powertop --auto-tune on PVE host and did not notice any drop in idle power.

Maybe the CPU won't reach lower C states, because of the SPF module or because you have not enabled all power saving options in the BIOS? What does "cpupower monitor" show?

 

[FingerBlaster]

So with my aqs-107 10gbe rj-45 transceiver, which sfp+ to rj-45 xceivers are known to be power hungry, my idle power jumps up. So it jumps up 2watts without an rj45 cable plugged in, and 3 watts when it's connected. Now i can't actually get link because it seems to be blacklisted, but thought it was intersting info. I play to use a DAC once i'm in "production", i'm just testing.

 

[FingerBlaster]

Interesting. So what exactly does powertop do under the hood to lower the idle power? Turn off cores, lower turbo frequency?

If you run powertop and go to the "tuneables" tab, it shows you all sorts of kernel parameters/tunables, and if they are configured in such a way to maximize efficiency. --auto-tune sets all these to a value that the developer of powertop says is good. powertop is written and maintained by intel employees, so i generally trust they know what they're doing, esp on an intel cpu

Edit: these do not persist after a reboot, you'd have to install powertop as a system service if you want it to.

Edit2: now that i have some VM's running mostly idle, the system went from about 20W in powersave, to about 33W in performance, this was achieved with the linux tool "cpupower frequency-set -g powersave". In performance i was seeing occasionally jumps to 40W for a second or so, and 150W for an Instant.

 

[GreenAvacado]

Like to get an opinion about U.2 SSD that I'm considering.

7.68TB NVMe PCIe 3.0 SSD MZ-QLB7T6B Samsung U.2 Enterprise NVME Drive PM983a


About $450 for new on eBay,
won't burn a hole in the pocket

Besides being Gen 3 (newer PM993a are Gen 4 but almost x2 price). I don't see any obvious downside. Anybody using this drive like to share their experience?

 

[MetalPhreak]

hello guys..

received mine from Amazon, how to disable secure boot? on the bios is greyed and it cannot boot anything

thanks in advance!

Make sure you hit esc/del to enter bios otherwise you can’t edit if you just let it boot into bios by itself. Also try resetting tpm.

 

[cableslayer]

has anyone gotten a 40Gb card to work? i didn't see this covered in any of the reviews.

It works with Mellanox ConnectX3 MCX354A-FCBT.
But I'm having issue to PCI passthrough the NIC to PVE VMs. I'm getting following error logs while start the VM:

kvm: -device vfio-pci,host=0000:01:00.0,id=hostpci0,bus=ich9-pcie-port-1,addr=0x0: vfio 0000:01:00.0: Failed to set up TRIGGER eventfd signaling for interrupt INTX-0: VFIO_DEVICE_SET_IRQS failure: Device or resource busy
TASK ERROR: start failed: QEMU exited with code 1

host dmesg reporting:

[ 199.050163] genirq: Flags mismatch irq 16. 00000000 (vfio-intx(0000:01:00.0)) vs. 00000080 (i801_smbus)

 

[OJO360]

Like to get an opinion about U.2 SSD that I'm considering.

7.68TB NVMe PCIe 3.0 SSD MZ-QLB7T6B Samsung U.2 Enterprise NVME Drive PM983a


About $450 for new on eBay,
won't burn a hole in the pocket

Besides being Gen 3 (newer PM993a are Gen 4 but almost x2 price). I don't see any obvious downside. Anybody using this drive like to share their experience?

Just got mine yesterday. I plan on using it as network storage/proxy video editing. I set it up as ZFS in Promox then passed 6tb to a turnkey lxc container. After formatting you get about 6.9Tb of usable space. In a win10 VM using the 2.5GB NIC, at first i saw it hitting over 1,000MB/s then after a min or so it would drop very low then after a while jump back up to about 500MB/s but now seems to be at always around 2-300MB/s after a windows explorer crash. I ran a quick crystaldiskmark and bassically got full SSD speeds(3531.16MB/s read / 3365.88MB/s write). I am running test disk to recover another drive but i want to reboot proxmox/win10 vm to make sure numbers are accurate. but so far works great. I wish the 15TBs were'nt 2.5x the price of the 7.68TB i would have gotten one of them, maybe next year or if i get a bonus somewhere.

 

[GreenAvacado]

Just got mine yesterday. I plan on using it as network storage/proxy video editing. I set it up as ZFS in Promox then passed 6tb to a turnkey lxc container. After formatting you get about 6.9Tb of usable space. In a win10 VM using the 2.5GB NIC, at first i saw it hitting over 1,000MB/s then after a min or so it would drop very low then after a while jump back up to about 500MB/s but now seems to be at always around 2-300MB/s after a windows explorer crash. I ran a quick crystaldiskmark and bassically got full SSD speeds(3531.16MB/s read / 3365.88MB/s write). I am running test disk to recover another drive but i want to reboot proxmox/win10 vm to make sure numbers are accurate. but so far works great. I wish the 15TBs were'nt 2.5x the price of the 7.68TB i would have gotten one of them, maybe next year or if i get a bonus somewhere.

Sweet, ended up ordering mine a while back too.

BTW, isn't 2-300MB/s is about right for transfer over 2.5G NIC? 1000MB/s you mentioned must be burst that fills up the DRAM cache.

 

[LiKenun]

I’ve scoured the Internet for 13th generation Intel Core CPUs with the Optane H10/H20 series of SSDs. Thus far, I’ve only found confirmation of socketed 13th generation CPUs working with them with the Z690 chipset.

But the screenshots @vxspiritxv posted shows that even the Raptor Lake (13th generation) model has an Alder Lake (12th generation) PCH, so it’s gotten my hopes up that my Optane H20 isn’t marooned and can be carried forward one more generation of hardware.

Has anyone checked yet that at least both sides of the H10 or H20 SSD can be accessed by the OS? I don’t mind the lack of support from RST, but I do mind that I can only see one side of the SSD when connected to AMD motherboards or inside a Thunderbolt enclosure.

Also: many thanks for the lspci -vvv output, @vxspiritxv. Could you do lspci -tv as well? I’m especially curious how many of the built-in devices are funneled through the narrow PCIe 2.0 x8 chipset link.

 

[JaxJiang]

hello guys..

received mine from Amazon, how to disable secure boot? on the bios is greyed and it cannot boot anything

thanks in advance!

In lastest UEFI BIOS standard.
You need enter BIOS before you try to enter OS.
Just reboot your machine and press del to enter BIOS, you will get secure boot setting can be modified.
If you try to enter OS, such as boot failed because of secure boot reasion. After that, you cannot change any secure setting.

 

[BlueChris]

Like to get an opinion about U.2 SSD that I'm considering.

7.68TB NVMe PCIe 3.0 SSD MZ-QLB7T6B Samsung U.2 Enterprise NVME Drive PM983a


About $450 for new on eBay,
won't burn a hole in the pocket

Besides being Gen 3 (newer PM993a are Gen 4 but almost x2 price). I don't see any obvious downside. Anybody using this drive like to share their experience?

I had ordered also one with almost the same price and i am waiting it to arrive.
The joke part is that i have received everything that i ordered the last 3-4 weeks from amazon, aliexpress, greek shops and i still dont have a single clue when the MS-01 will arrive here. I did sent a second email today to minisforum asking WTH is huppening really.

 

[Tsatsar]

I had ordered also one with almost the same price and i am waiting it to arrive.
The joke part is that i have received everything that i ordered the last 3-4 weeks from amazon, aliexpress, greek shops and i still dont have a single clue when the MS-01 will arrive here. I did sent a second email today to minisforum asking WTH is huppening really.

Just received mail that they are back from "Spring Break".
Shame on them
Patrida, hold on..

 

[nicoska]

After a couple of testing ( i9-13900H - 96 GB RAM - 1 NVME SSD - SFP+ Ehternet module - Proxmox 8.1.4) I'm keep getting those errors and the system itself reboot:

Feb 19 12:25:46 nicoska2 kernel: RAS: Soft-offlining pfn: 0xb2c66e
Feb 19 12:25:46 nicoska2 kernel: RAS: Soft-offlining pfn: 0x10e2a5b
Feb 19 12:25:46 nicoska2 kernel: Memory failure: 0x10e2a5b: unhandlable page.
...
...
Feb 19 12:26:51 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:26:51 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:26:51 nicoska2 kernel: Memory failure: 0x10e1bdf: unhandlable page.
Feb 19 12:26:59 nicoska2 kernel: mce: CMCI storm detected: switching to poll mode
Feb 19 12:26:59 nicoska2 kernel: RAS: Soft-offlining pfn: 0x10e1b32
Feb 19 12:26:59 nicoska2 kernel: RAS: Soft-offlining pfn: 0x10209d
Feb 19 12:26:59 nicoska2 kernel: RAS: Soft-offlining pfn: 0x942368
Feb 19 12:26:59 nicoska2 kernel: Memory failure: 0x10e1b32: unhandlable page.
Feb 19 12:26:59 nicoska2 kernel: Memory failure: 0x10209d: unhandlable page.
Feb 19 12:28:12 nicoska2 kernel: mce_notify_irq: 27 callbacks suppressed
Feb 19 12:28:12 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:28:20 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:29:57 nicoska2 kernel: mce_notify_irq: 2 callbacks suppressed
Feb 19 12:29:57 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:29:57 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:31:03 nicoska2 kernel: mce_notify_irq: 1 callbacks suppressed
Feb 19 12:31:03 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:31:32 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:31:42 nicoska2 kernel: RAS: Soft-offlining pfn: 0x10e1b37
Feb 19 12:31:42 nicoska2 kernel: Memory failure: 0x10e1b37: unhandlable page.
Feb 19 12:32:49 nicoska2 kernel: mce_notify_irq: 1 callbacks suppressed
Feb 19 12:32:49 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:32:51 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
-- Reboot -

Those are the additional hardware I put inside:

• Lexar NQ790 2 TB PCIe 4.0 SSD Interno, M.2 2280 PCIe Gen4x4 NVMe 1.4
• Crucial RAM 96GB Kit (2x48GB) DDR5 5600MHz
• 1/2.5/5/10Gb SFP+ RJ45 Transceiver

Any idea? Doing a google search resulted in two different discussions where the poster said it was a faulty CPU. I'm e little bit scared :/

Thank you for your help
Nico

 

[the-last-englishman]

After a couple of testing ( i9-13900H - 96 GB RAM - 1 NVME SSD - SFP+ Ehternet module - Proxmox 8.1.4) I'm keep getting those errors and the system itself reboot:

Feb 19 12:25:46 nicoska2 kernel: RAS: Soft-offlining pfn: 0xb2c66e
Feb 19 12:25:46 nicoska2 kernel: RAS: Soft-offlining pfn: 0x10e2a5b
Feb 19 12:25:46 nicoska2 kernel: Memory failure: 0x10e2a5b: unhandlable page.
...
...
Feb 19 12:26:51 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:26:51 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:26:51 nicoska2 kernel: Memory failure: 0x10e1bdf: unhandlable page.
Feb 19 12:26:59 nicoska2 kernel: mce: CMCI storm detected: switching to poll mode
Feb 19 12:26:59 nicoska2 kernel: RAS: Soft-offlining pfn: 0x10e1b32
Feb 19 12:26:59 nicoska2 kernel: RAS: Soft-offlining pfn: 0x10209d
Feb 19 12:26:59 nicoska2 kernel: RAS: Soft-offlining pfn: 0x942368
Feb 19 12:26:59 nicoska2 kernel: Memory failure: 0x10e1b32: unhandlable page.
Feb 19 12:26:59 nicoska2 kernel: Memory failure: 0x10209d: unhandlable page.
Feb 19 12:28:12 nicoska2 kernel: mce_notify_irq: 27 callbacks suppressed
Feb 19 12:28:12 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:28:20 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:29:57 nicoska2 kernel: mce_notify_irq: 2 callbacks suppressed
Feb 19 12:29:57 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:29:57 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:31:03 nicoska2 kernel: mce_notify_irq: 1 callbacks suppressed
Feb 19 12:31:03 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:31:32 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:31:42 nicoska2 kernel: RAS: Soft-offlining pfn: 0x10e1b37
Feb 19 12:31:42 nicoska2 kernel: Memory failure: 0x10e1b37: unhandlable page.
Feb 19 12:32:49 nicoska2 kernel: mce_notify_irq: 1 callbacks suppressed
Feb 19 12:32:49 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
Feb 19 12:32:51 nicoska2 kernel: mce: [Hardware Error]: Machine check events logged
-- Reboot -

Those are the additional hardware I put inside:

• Lexar NQ790 2 TB PCIe 4.0 SSD Interno, M.2 2280 PCIe Gen4x4 NVMe 1.4
• Crucial RAM 96GB Kit (2x48GB) DDR5 5600MHz
• 1/2.5/5/10Gb SFP+ RJ45 Transceiver

Any idea? Doing a google search resulted in two different discussions where the poster said it was a faulty CPU. I'm e little bit scared :/

Thank you for your help
Nico

Definately memory error, what speed is your memory running at ?
Should be no more than 4800 MT/s for 96Gb

dmidecode --type 17

 

[nicoska]

Definately memory error, what speed is your memory running at ?
Should be no more than 4800 MT/s for 96Gb

dmidecode --type 17

Mine is running at 5200 MT/s


Do I need to change at BIOS level or do I need to complete replace the modules?
Thank you

 

[the-last-englishman]

I would do a bios reset and see if it trains to 4800 MT/s after the reset.

And update microcode:

Microcode Install Instructions:

1) Install Proxmox 8.1
2) Add non-free-firmware to debian repo in sources.list - Edit the /etc/apt/sources.list file.
-- Add non-free-firmware to the 1st line so it looks like this <deb http://ftp.debian.org/debian bookworm main contrib non-free-firmware>
3) Save Changes
4) #apt clean && apt update
5) #apt install intel-microcode - The current version Debian has in the repo is 3.2023114.1~deb12u1
6) Reboot, and the microcode patch should apply automatically.
7) You can check what microcode you are running after reboot by grep 'stepping\|model\|microcode' /proc/cpuinfo

 

[nicoska]

I would do a bios reset and see if it trains to 4800 MT/s after the reset.

You mean flashing a stock BIOS or a specific "reset" menu in the BIOS?