Minisforum MS-01 PCIe Card and RAM Compatibility Thread

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

bryan_v

Active Member
Nov 5, 2021
149
77
28
Toronto, Ontario
www.linkedin.com
Yeup, that's a high power XGSPON transceiver. I assume there is also a heatsink on the transceiver as well? The transceiver would probably also be significantly longer than a normal transceiver. (2.5W is the danger zone if there's no heatsink)

Depending on which generational revision the transceiver is from, it might be expecting active airflow over the optics section. Switches and business class routers normally maintain negative pressure in the chassis, causing air to be pulled in through all the ports (thus causing airflow over the optics). If however the original ISP box didn't have active cooling, then it's probably designed to passively dissipate heat and all you'll have to do is make sure the surrounding environment stays around STP and there is headspace for convection to carry the heat away, but in general it will be fine in your MS-01's SFP+ (though the heatsink might block the pcie slot).

If you're still worried, you can grab a Klein or Fluke multimeter with a thermocouple and just double check the temps, they're useful to have around anyways. As long as the SFP+ cage inside the MS-01 <55C, you're fine. If the temps push past 60C you'll need some active airflow somewhere as you probably have a heat pocket forming behind or inside the case. Or alternatively at that point, you can also grab a 10G media converter, which will push that heat to an external box that you can mount on the wall where natural convection will help cool it (also helps keep the APC stuff out of the rack).
 
  • Like
Reactions: Scott Laird

Scott Laird

Well-Known Member
Aug 30, 2014
439
271
63
Also, while 2.5W is a lot for a tiny SFP, it's a trivial amount of heat to cool with any fan at all. I have a few MS-01s with 120mm USB fans just sitting on top of the case blowing air down through the top vent holes, and they all run dramatically cooler than they would otherwise. That's more than enough to keep a ~25W ConnectX-5 happy. Adding pretty much any fan at all would probably keep a 2.5W SFP cool without much trouble or much noise.
 

herbie

New Member
Nov 20, 2025
9
0
1
Yeup, that's a high power XGSPON transceiver. I assume there is also a heatsink on the transceiver as well? The transceiver would probably also be significantly longer than a normal transceiver. (2.5W is the danger zone if there's no heatsink)

Depending on which generational revision the transceiver is from, it might be expecting active airflow over the optics section. Switches and business class routers normally maintain negative pressure in the chassis, causing air to be pulled in through all the ports (thus causing airflow over the optics). If however the original ISP box didn't have active cooling, then it's probably designed to passively dissipate heat and all you'll have to do is make sure the surrounding environment stays around STP and there is headspace for convection to carry the heat away, but in general it will be fine in your MS-01's SFP+ (though the heatsink might block the pcie slot).

If you're still worried, you can grab a Klein or Fluke multimeter with a thermocouple and just double check the temps, they're useful to have around anyways. As long as the SFP+ cage inside the MS-01 <55C, you're fine. If the temps push past 60C you'll need some active airflow somewhere as you probably have a heat pocket forming behind or inside the case. Or alternatively at that point, you can also grab a 10G media converter, which will push that heat to an external box that you can mount on the wall where natural convection will help cool it (also helps keep the APC stuff out of the rack).
To protect the ms-01 ports, is it a good idea to use this kind of switch with added heat dissipators on the xgs-pon module, plug the module inside and use a dac cable to ms-01 ?
1763998161580.png
 

bryan_v

Active Member
Nov 5, 2021
149
77
28
Toronto, Ontario
www.linkedin.com
Don't use a switch, use a media converter. If you use a switch, you need to have management functions so you can set it up as either a virtual wire, or configure the two ports properly to forward PPPoE packets to each other (not hard, but not trivial either), at which point you've wasted so much money you might as well get a dedicated box with proper cooling. This is an example of SFP+ to 10Base-T, but you might want a SFP+ to SFP+ if you want to completely bypass your ISP's hardware.
 

herbie

New Member
Nov 20, 2025
9
0
1
Don't use a switch, use a media converter. If you use a switch, you need to have management functions so you can set it up as either a virtual wire, or configure the two ports properly to forward PPPoE packets to each other (not hard, but not trivial either), at which point you've wasted so much money you might as well get a dedicated box with proper cooling. This is an example of SFP+ to 10Base-T, but you might want a SFP+ to SFP+ if you want to completely bypass your ISP's hardware.
In fact, there is no need for pppoe. It is juste vlan tagging that can be done on xgs module or opensense and it's pure dhcp
 

bryan_v

Active Member
Nov 5, 2021
149
77
28
Toronto, Ontario
www.linkedin.com
If they're not using PPPoE then yea, you're fine; though pure DHCP is pretty unusual because it has no protection for rouge endpoints just spicing into the network at a pole and spoofing a mac because it's just a passive network. Have you done a dry run with an homebrew box?
 

herbie

New Member
Nov 20, 2025
9
0
1
@bryan_v I already tested xgs-pon directly in sfp router and it works fine, but never tested with switch or media converter in between. For the xgs-pon to work, we have to put serial number of original provider box and imei number, and the module appears as legitimate to receive public ip.
 

nicoska

Member
Feb 8, 2023
67
11
8
@bryan_v I already tested xgs-pon directly in sfp router and it works fine, but never tested with switch or media converter in between. For the xgs-pon to work, we have to put serial number of original provider box and imei number, and the module appears as legitimate to receive public ip.
Hello,
out of curiosity, which provider are you talking about? I'm asking because I'm in Italy and I'm interested about this option.
Currently I'm using a 10G fritzbox to connect to my 10Gbs provider, but I'd love to bypass it and use Opnsense in my MS-A2 box.
thanks
 

herbie

New Member
Nov 20, 2025
9
0
1
Hello,
out of curiosity, which provider are you talking about? I'm asking because I'm in Italy and I'm interested about this option.
Currently I'm using a 10G fritzbox to connect to my 10Gbs provider, but I'd love to bypass it and use Opnsense in my MS-A2 box.
thanks
Using Bouygues french provider but you may find your ISP on pon.wiki
 

bryan_v

Active Member
Nov 5, 2021
149
77
28
Toronto, Ontario
www.linkedin.com
If you've tested it in a home-built box, then you should be fine. If you're running 8311 on the transceiver itself, then the layer 2-ish auth is S/N and RegID. You should still keep an eye on Layer 3, especially if Bouygues issues IPv6 addresses, but in general, a dumb switch or a media converter will both work.

Also just saw the transceiver on pon.wiki, and yes that heatsink looks really small for XGS and is probably designed for active airflow, or at least headspace above for convection to cool the optics. Since it runs linux, once you have it set up, you should see if you can get temperature data from the PRX126. If you can access the die temp, just set up a batch script to poll it every 1-5 min and feed that to a timescale db (Prometheus probably, but anything will really work). If it peaks over 60C, just point a fan at it (5V USB powered arctic/delta/sunon/noctua would do the trick), try pushing cold air into the router room.
 
  • Like
Reactions: Scott Laird

dak64

New Member
Jun 11, 2024
10
3
3
There is a very active community on discord discussing XGSPON, link is at the pon.wiki site. I am tempted to bypass my Frontier FRX523 but 1) I'm not clear on the advantage, I already get 2.5 Gbps up/down and 5 ms pings to my closest DNS (9.9.9.9) on my 2 Gbps plan. How much better can it get? And 2) with tariffs the XGSPON's have become quite expensive. There is a seller on alliexpress that ships from the US and it includes a tiny heatsink and fan, but still $125+
 

routenull0

New Member
Dec 28, 2016
22
4
3
44
Confirming for anyone else searching. I got the above Glotrends card along with this m.2 adapter working with the DEG1:
https://www.aliexpress.us/item/3256807419018300.html
I'm looking to use my MS-01 and the above items + DEG1 as a temporary solution to play Arc Raiders when I go on travel, so I'm hoping it works well.

What GPU did you end up using the DEG1? I've read some posts elsewhere about issues with Nvidia GPUs.
 

JaxJiang

Member
Jan 10, 2023
94
82
18
I'm looking to use my MS-01 and the above items + DEG1 as a temporary solution to play Arc Raiders when I go on travel, so I'm hoping it works well.

What GPU did you end up using the DEG1? I've read some posts elsewhere about issues with Nvidia GPUs.
Buy minisforum eop4a
 

Arjestin

Member
Feb 26, 2024
45
8
8
Since we're talking about MS-01 without its metal case, I have a question on this topic —

I have two MS-01 units side-by-side in a 3D-printed 2U rack mount, and here are two 12cm fans below them, to cool the NVMe SSDs. This seemed like a good idea at the time, but now I think blowing air at max power on the MS-01 metal case is not the most efficient way to cool those SSDs. I consider getting rid of the metal case, attaching some heatsinks on the SSDs, and connecting the bottom fans to MS-01's internal PWM connector.

Apart from dust and the risk of foreign objects accidentally getting into the MS-01 motherboard, do you see other potential problems with this solution?
 

FingerBlaster

Active Member
Feb 27, 2019
102
47
28
I'm having a strange issue tied to a specific MS-01. I'm running a hyperconverged proxmox cluster with ceph. 3x ms-01 nodes. Call them Node1, Node2, Node3. The problem i'm having is only with Node1.

Node1 went down the other day. I restarted it and it crashed as it was loading VMs. Disable autostart on the VM's. Proxmox starts fine. okay. Start up each VM one at a time, discovered a trigger. Linux Vm's don't seem to cause any issue (none of my linux VM's have a gui).

Every time i start up a Windows VM on it, Node1 itself powers down. It happens after windows is started the console display on Node1 goes black, then a couple seconds later the VM stops responding, then Node1 powers down. Okay, maybe it's related to something weird with this specific windows VM? nope, tried to start a different windows VM same behavior. It's not even a crash, the node just shut's off. This config has been stable for over a 6 months now running the same workload. My previous issue back in May was resolved by swapping one of the MS01's out.

Trying to watch the system logs on the console and i don't see any suspicious output, console just goes black.

Basic bios settings, no overclocking or changed powerlimits. Anyone have any ideas? I'm not doing anything weird, no pcie passthrough, or anything. the only "weird" thing i'm doing is something similar to what the poster above me is doing, and that's i have a 120mm case fan blowing on my nvme drives which if anything should not be causing issues.

EDIT: running a cpu benchmark also caused a crash. Been running the 1.27 bios for over 6 months now.
 
Last edited:

JaxJiang

Member
Jan 10, 2023
94
82
18
I'm having a strange issue tied to a specific MS-01. I'm running a hyperconverged proxmox cluster with ceph. 3x ms-01 nodes. Call them Node1, Node2, Node3. The problem i'm having is only with Node1.

Node1 went down the other day. I restarted it and it crashed as it was loading VMs. Disable autostart on the VM's. Proxmox starts fine. okay. Start up each VM one at a time, discovered a trigger. Linux Vm's don't seem to cause any issue (none of my linux VM's have a gui).

Every time i start up a Windows VM on it, Node1 itself powers down. It happens after windows is started the console display on Node1 goes black, then a couple seconds later the VM stops responding, then Node1 powers down. Okay, maybe it's related to something weird with this specific windows VM? nope, tried to start a different windows VM same behavior. It's not even a crash, the node just shut's off. This config has been stable for over a 6 months now running the same workload. My previous issue back in May was resolved by swapping one of the MS01's out.

Trying to watch the system logs on the console and i don't see any suspicious output, console just goes black.

Basic bios settings, no overclocking or changed powerlimits. Anyone have any ideas? I'm not doing anything weird, no pcie passthrough, or anything. the only "weird" thing i'm doing is something similar to what the poster above me is doing, and that's i have a 120mm case fan blowing on my nvme drives which if anything should not be causing issues.

EDIT: running a cpu benchmark also caused a crash. Been running the 1.27 bios for over 6 months now.
Upgrade to lastest V1.27 BIOS IN BIOS menu find “Turbo frequency adjustment” make your big core and small cores max Lower the frequency of your big cores and little cores. Maybe CPU design issues, it may no longer be able to stably maintain the highest turbo frequency.