Minisforum MS-01 PCIe Card and RAM Compatibility Thread

[bobbysteel]

no issues here with the 96gb ram. have had it since day one of probably one of the v first shipments they made from the preorder period

 

[goodt]

I ordered a kit and it arrives tomorrow the day after tomorrow.

Here is the info:

Crucial 128GB Kit (64GBx2) DDR5-5600 SODIMM | CT2K64G56C46S5 | crucial.com

Buy Crucial 128GB Kit (64GBx2) DDR5-5600 SODIMM CT2K64G56C46S5.

www.crucial.com

Amazon link:

https://www.amazon.com/dp/B0DSQMKYLN

I order this and the first set had an issue machine would not post. I returned it and got a replacement which worked fine!

 

[bruce1066]

There's a working option to add one more M.2 slot to the MS-01 without using the PCIe slot or the Wi-Fi M.2 slot.
I installed the QNAP QDA-U2MP using the M.2 to U.2 adapter that comes with the MS-01. The QNAP enclosure needs to be removed for the card to fit. It's a tight fit, but is easy to do. It works well: you can put the original fan/plate back and close the case normally, no clearance issues. While there’s not enough space for individual SSD heatsinks, the original MS-01 fan/plate acts as one.

With an external fan blowing on the side of the MS-01, the two SSDs in the QNAP adapter idle at around 55°C, which is actually cooler than the SSDs in slot 2 and 3 (those have heatsinks and idle at around 60C).
EDIT : I ended up removing the heatsink on the 3rd SSD (the one closest to the WiFi card), to prevent it from obstructing the fan. I hadn't noticed it at first, but realized the 2nd and 3rd SSDs were getting unusually hot. Doing this lowered the idle SSD temp by 10°C vs before. Thank you @randman for the tip.

Downsides:

• The controller in the QNAP downgrades the link to PCIe 3.0 x4 (instead of PCIe 4.0 x4).
• It’s still relatively expensive compared to using an M.2 riser on the Wi-Fi slot or a PCIe M.2 expansion card.

But if you want to keep the PCIe slot free and avoid being limited to PCIe 3.0 x1 via the Wi-Fi slot, this is a solid hack option.



View attachment 42646

View attachment 42643
View attachment 42645

I'm thinking that the chassis hardware must have changed. I purchased the QNAP adapter, delivered last week (April 2025) and fitted it in to the slot. The nvme connectors are flush with the top surface of the chassis and the mounting screws for the nvme's are a couple of mm proud. So no way to refit the fan plate. Even if the 2xnvme slots are empty, the fan plate can't be fitted. My machine, delivered last week is an 19-13900.

 

[inbox24]

Has anyone tested the Crucial DDR5 128gb (2x64) RAM kit on anything OTHER than the 13900H model of the Minisforum MS-01? Looking to purchase the i5 12600H model with the 128gb of RAM

There seems to be a large price differential between that model and some of the i5 12600H models and even the i9 12900H models and seeing that most homelab self hosted apps don't use that much in the way of CPU cores, a cheaper CPU model with more RAM might be the best bang for buck in terms of a Proxmox hypervisor

 

[caplam]

I understand but...

Even for a home lab, my criterion n°1 is stability.
Plex, Home Assistant, a nightmare for the family when the box crashes. I'll be public enemy n°1 if it doesn't work.

I have been running 1 ms01 since feb 2024. The first weeks were tough with overheating problems which caused stabilities issues.
But after sitting the unit on top of 120mm fan and with another 80mm above it has been running really well.
64Gb ram, 3 ssd and 1 9300 hba.
Running unraid with 3 vm (one of which is a hungry security onion vm with 24G ram and a 10Gbps monitoring port) and almost 80 running docker containers.
Since having solved heating problems i have not had hardware issue (only issues are related to unraid oom impacting gui)
In fact i'm even looking at the new ms a2

 

[freegate]

That's the problem. I don't want to have to tinker with the box to make it work.
It has to be plug and play.

The i9-13900 runs extremely hot.
I hope the AMD Ryzen 9 9955HX is more reasonable in this regard.

 

[masewo]

After reading most of the pages, I still have some doubts, if I can realize my plan to use the MS-01 as my new NAS because I have laying around 5 unused NVMe drives for RAID5.

Since Bifurcation is not supported by the mainboard, I need an PCIe to M.2 adapter with a chip on it, or? The cheapest I found is the following one https://de.aliexpress.com/item/1005005838511246.html

This plus some additional fans should make me happy, or? Can we expect 1 GB/s write speed over the network?

 

[wadup]

New x700 nvm update released:

NVM Update Utility for Intel® Ethernet Network Adapter 700 Series

NVM Update Utility for Intel® Ethernet Network Adapter 700 Series (All Supported Operating Systems)

www.intel.com

EFI flashed with no issues:

 

[timot]

Hello, just to inform you all BIOS v1.27 was uploaded 04/29/2025.
Available here : minisforum

Release Date: 2025/04/03
BIOS Version: 1.27
BIOS Checksum: 4CE9

BIOS Description:
    1.Add Turbo Ratio Limit Options form
    2.BIOS disables the execution of the UEFI storage Option ROM  by default
    3.The PCIE SLOT supports speed control
    4.Set PL1 to 55W and PL2 to 80W for the 12800H
    5.Set PL1 to 60W and PL2 to 80W for the 270H
    6.Update GOP version to 21.0.1066

Hope it fixes some of the bugs seen around here.

 

[Stanthewizzard]

3 big changes for me to share:
- Bios 1.27 (no issues with proxmox) (turbo boost already disabled)
- change the past on the rad (4° lost) for Thermal Grizzly - PhaseSheet PTM
- change fan settings in BIOS (https://forums.servethehome.com/ind...ompatibility-thread.42785/page-32#post-415381)

At the end 10° lost before and after

 

[dokterdok]

Since Bifurcation is not supported by the mainboard, I need an PCIe to M.2 adapter with a chip on it, or? The cheapest I found is the following one https://de.aliexpress.com/item/1005005838511246.html

2 things to check before ordering:
* Card dimensions : Needs to be 170mm or shorter without the bracket, or it won't fit.
* Chip : You'll need a chip if you need more than one drive on the expansion card. Minisforum recommends using a card with an ASM chip over PLX, since PLX ones can have compatibility issues with the MS-01.

 

[Fazio]

Anyone testing different values for Turbo Ratio Limits on V1.27 BIOS?
I'm considering to decrease the values to improve the thermal efficiency while keeping Turbo Boost enabled. That would be really interesting.

 

[Stanthewizzard]

Disabled it.
Better for the life of ms-01

 

[Fazio]

Disabled it.
Better for the life of ms-01

Unfortunately I can not disable it. I use it as main router for my 10G PPPoE line and I’m unable to reach maximum throughput with turbo boost off. I’m using Mikrotik CHR as VM, in Proxmox.

 

[Stanthewizzard]

Using proxmox with opnsense. 8gb sym ISP line.
Other vm and no pb
8gb down / 7.5 upload

 

[Fazio]

Using proxmox with opnsense. 8gb sym ISP line.
Other vm and no pb
8gb down / 7.5 upload

Are you passing through the interfaces or using vmbr with multiqueues? Turbo boost?
Hope you’re not using PPPoE like my ISP..

 

[Stanthewizzard]

VMBR
Multiqueues 4
No PPOE (sadly double nat)

 

[Miesepies]

I just found this on AliExpress: US $5.80 | Compact M.2 Key E/A/E to NVME SSD Adapter PCIE GEN3 X4 For 2230 2242 2260 2260 2280 SSD Riser Card

https://a.aliexpress.com/_oFl3hDj

Thanks for the tip. I ordered one aswell, and it works like a charm.

 

[Miesepies]

Thanks. I did a factory reset, and at first it all seemed to work fine. Boot and then a restart with working HDMI out. But unfortunately third time was not a charm. I was not able to get the MS-01 to boot with the 4K display connected after that. I did a factory reset again, I switched the monitor off, I unplugged the power of the MS-01. Nothing.

Only way to get past the POST, is by either using an USB-C to HDMI kabel or no HDMI cable plugged in at all. Once the machine is starting up, I can insert the HDMI cable and I have a working display.

This machine will run as a server, so won't be connected to this monitor after setup. But, considering my warranty, I wonder if this a bug or an actual defect.

Minisforum support suggested to load optimized defaults to fix this. I'm sure I have done that before, but after doing it again the MS-01 boots normally (most of the time).

 

[FingerBlaster]

Hopefully someone here has some thoughts that will help me not have to read 99 pages of form posts, i know there have been some issues with mysterious instability.

I have a 3 node 13900h ms-01 Proxmox cluster with 96gb ram each node. Each node has dual 4tb samsung p983 m.2 nvme drives and a single 8tb samsung p983A u.2 drive. I am running a thunderbolt mesh network for ceph storage. The cluster was running flawlessly from roughly July 2024 until February 2025. No crashes, no issues. Late February 2025 i patched proxmox. Ever since then i've been having stability issues. I run 5 LXC containers, and 2 windows 11 VMs, 2 Windows Server 2022 Vms, and 6 *nix Vm's. What's peculiar is whichever node has one specific windows server vm on it, keeps crashing. The only thing installed is windows server 2022 and ms exchange server 2019. Right now for isolation purpose I keep exchange server on PVE2 and no other VMs/LXC on it, so when it crashes and hard locks the host, it only affects one VM.

I've found no reason for the exchange vm to keep crashing, to rule things out i've tried different virtual storage controllers, various different settings in proxmox, all have made no effect. I patched proxmox again to the newest version. I flashed the newest 1.27 firmware from minisforum. patched the microcode. i reset factory defaults in the bios, removing all the tweaks i applied (mostly aspm and power settings). I also disabled some os boot scripts i was previously running. stopped "powertop --autotune" and stopped running cpupower setting the powersave powergovernor, so it's on performance mode now, which is default. None of which has any impact in these lockups.

I'm not seeing any obvious errors in the journal log, and i'm getting fed up that this keeps happening. I've tried running some synthetic benchmarks to "cause" a crash, at best i've only locked up the guest, but i have not been able to replicate locking the host up at all.

EDIT 05/18/2025: My initial assumption was incorrect, or the problem was multi-factorial and i resolved one of the issues. When i initially ran into this issue PVE2 crashed a few times, i decided to move exchange to pve3 and pve3 started crashing, so i've been isolating it on PVE2. I've since been doing a lot of trouble shooting, and after making this post I wanted to retest my initial premise, and I moved exchange over to PVE3, no crashes in a week. I've also been trying to track down some ceph messages on all nodes on all osd's that seem to be related to packet loss.
May 18 10:01:06 pve3 ceph-osd[2441]: 2025-05-18T10:01:06.703-0400 78d895cdc6c0 -1 bdev(0x5a706f3e6400 /var/lib/ceph/osd/ceph-2/block) aio_submit retries 6
I've had these ceph warning messages the entire time even when the system was stable, but i figured they were worth trying to solve. By setting up traffic shaping to limit bandwidth I've greatly improved the packet loss, so last night I moved exchange over to PVE2 and the node crashed over night. So the issue is this specific node and something do with the type of load exchange places on it. I'm going to run a few stress tests on the system to see if i can replicate a crash. I also have a spare ms-01 that i had bought last year when i bought my inita 3 for another project that i just haven't gotten to yet, so i'm going to try swapping out the server to see if it's a system issue, possibly a degraded 13th gen cpu from running original bios for well over a year (if it's not broke why upgrade it?, woops).