Well, I am sure that the interest for doing so are there, and I am probably not the only person who has an idle framework motherboard (soon to be 2 once the AMD Phoenix one shows up) laying around.
So, repurposing the framework board like a NUC/Not-so-TMM - good idea, or bad idea?
Generally speaking? Not great if you depend on it, but for messing around in a lab environment? Not too bad. Take for instance "Thunderscreech", which is my Framework Gen 1 board on a Coolermaster plastic case. What's in it? Tiger Lake i5-1135G7, which is around the ballpark of an AMD Renoir based Ryzen 3 4000U series machine, with 64GB of DDR4 RAM and a 512GB PCIe 3x4 NVMe SSD.
So, what's the issue holding it back from being a decent server?
a) The M.2 A+E port is PCIe 3x1, placed in an awkward place and only breaks out to a 2230 slot on the side, so putting a decent ethernet card in it and still fit in the case...is out. I will have to figure out a way to shove a NIC in there...probably an Intel i210. Not sure if I can bring the M.2 A+E port out to a breakout and then go MiniPCIe using a mezzanine board.
b) The eDP breakout is custom to Framework, and Framework did not ship a breakout board that will allow you to drive an external display. So no easy way to have a dedicated display output for it, even though its laptop brethren has one built-in.
c) The Framework 13 board features 4 USB4/TB3 based expansion ports, and you'll need one to be USB-C/PD to inject power, one to drive a display (HDMI/Displayport), at least one for USB input (mouse/keyboard) and then your ethernet would need to be USB based.
Oh yeah, the entire USB based setup will not play well with "enterprise" stuff. Like for example, getting it to play nice with Citrix Hypervisor/xcp-ng. Take for instance this bag of fun where I had to use a USB-PD power injecting Ethernet adapter (RTL8152 based), and it's doing some weird stuff when the Xen PV vNIC is being used on the guest VMs, or due to Xen Hypervisor's kinda-RedHat device rename udev rules, every hypervisor reboot will come up with a different NIC. That took a while to deal with, and I am still dealing with the NIC acting up on me, like shutting down every hour or so.
So yeah, can you use the Framework as a ghetto NUC-like device? Probably, but it takes much more work then previously assumed, and the design is fairly limiting in a way...
So, repurposing the framework board like a NUC/Not-so-TMM - good idea, or bad idea?
Generally speaking? Not great if you depend on it, but for messing around in a lab environment? Not too bad. Take for instance "Thunderscreech", which is my Framework Gen 1 board on a Coolermaster plastic case. What's in it? Tiger Lake i5-1135G7, which is around the ballpark of an AMD Renoir based Ryzen 3 4000U series machine, with 64GB of DDR4 RAM and a 512GB PCIe 3x4 NVMe SSD.
So, what's the issue holding it back from being a decent server?
a) The M.2 A+E port is PCIe 3x1, placed in an awkward place and only breaks out to a 2230 slot on the side, so putting a decent ethernet card in it and still fit in the case...is out. I will have to figure out a way to shove a NIC in there...probably an Intel i210. Not sure if I can bring the M.2 A+E port out to a breakout and then go MiniPCIe using a mezzanine board.
b) The eDP breakout is custom to Framework, and Framework did not ship a breakout board that will allow you to drive an external display. So no easy way to have a dedicated display output for it, even though its laptop brethren has one built-in.
c) The Framework 13 board features 4 USB4/TB3 based expansion ports, and you'll need one to be USB-C/PD to inject power, one to drive a display (HDMI/Displayport), at least one for USB input (mouse/keyboard) and then your ethernet would need to be USB based.
Oh yeah, the entire USB based setup will not play well with "enterprise" stuff. Like for example, getting it to play nice with Citrix Hypervisor/xcp-ng. Take for instance this bag of fun where I had to use a USB-PD power injecting Ethernet adapter (RTL8152 based), and it's doing some weird stuff when the Xen PV vNIC is being used on the guest VMs, or due to Xen Hypervisor's kinda-RedHat device rename udev rules, every hypervisor reboot will come up with a different NIC. That took a while to deal with, and I am still dealing with the NIC acting up on me, like shutting down every hour or so.
So yeah, can you use the Framework as a ghetto NUC-like device? Probably, but it takes much more work then previously assumed, and the design is fairly limiting in a way...
Last edited:
