I’ve never heard of metacomputing before, but as far as I can tell this is the first ARM mainboard for framework so thought I’d share it here.
Neat seeing more 3rd party boards coming out, I think this is the 3rd now? ‘Cause we got 2 deepcomputing mainboards.
Please excuse the incorrect tag on this post, it forced me to use a tag but there isn’t one for ARM yet so I chose the only 3rd party tag available.
16W at idle, I’ll pass
I ordered one about a week ago
I did order something like that from a near-identical company called deepcomputing, I am still working on the project of porting fedora to it with all the features working
the issue with this (metacomputing ARM) mainboard appears to be high power consumption at idle, I think this specific CPU at design should have high latency between core2core access (because of the layout), but I am waiting to get it to make a report
weirdly enough I didn’t get any shipping notification for a week 
maybe I was tricked, but assuming that deepcomputing did send me a board, I don’t think it’s likely, I checked them up and they share a CEO
Sadly it looks like the memory shortages bumped the price substantially; it’s showing as $906 for the mainboard + 16GB LPDDR5 + 1TB NVMe SSD now (up from $549), and between $1400-1600 in a full build with the Framework 13 chassis.
I’m hoping to receive a test unit next month, and will be posting my benchmarks and usage info here, prior to writing up a more in depth review on my blog/YouTube: MetaComputing AI PC Framework 13 Mainboard with Cix CP8180 · Issue #103 · geerlingguy/sbc-reviews · GitHub
Has anyone else received a unit for testing yet, or one that they’ve ordered?
It looks like they’re running a version of Debian in their main image, off this Linux 6.6.89 fork.
I’ve started my testing (full results will be posted here)—one quick note, the idle power consumption (with battery fully charged, after 2 hours in that state, and display off) is around 11-12W. Maximum power consumption under load (without GPU engaged, and screen still off) is around 32W.
Geekbench 6 scores are in line with Radxa Orion O6 and Minisforum MS-R1, and the default install MetaComputing provides is Ubuntu 25.04 on the Linux 6.6 kernel. (Default kernel for vanilla Ubuntu 25.04 is 6.14).
I haven’t explored the BIOS, but it looks like it’s TianoCore EDKII, and I’m currently on a beta version. I have the model with 16GB RAM.
Also of note, the prices shown on the MetaComputing website have gone down since my last post:
See related discussion, too: ARM-based CPUs
One note in that thread:
@yuning_liang - I was wondering, is there a newer BIOS that might have further idle power fixes? Right now I’m seeing 11W idle, with the display off, after letting the battery charge overnight.
I’ve updated the BIOS, and now idle is settling in around 7.9W (a 30% reduction, nice!), and maximum power consumption settles in around 26.5W (which is lower than the previous 31.7W), though performance results are consistent.
I am seeing some benchmarks perform a lot worse than on the Orion O6 and MS-R1 though, so maybe some further improvements could be made. This is the first time I’ve seen idle power on this chip go below 10W, which is nice. But it’s still a lot more than quasi-comparable chips like the A18 Pro in the Neo (2.8W) and the AI 340 in the base model AMD Framework 13 board (2.7W)
Have you managed to get the Immortalis G720 MC10 GPU to work on Linux?
Yes, at least Vulkan was working on the iGPU when I was testing with GravityMark, and it is recognized in GLMark2 as well. I have only tested so far with the official Ubuntu 25.04 install provided by MetaComputing.
I have posted a full video and blog post with my experiences with this board:
Towards the beginning of the video, you wonder if it has some of the problems you found with the Minisforum’s MS-R1.
But i don’t think that question is answered in the video.
What were the problems with the MSR1 and does this board have the same problem or not?
That being said, i am guessing the problem is the difficulty in running more up to date kernels on it.
Two major ones: kernel support (still an issue), and idle power consumption (still an issue, but at least it’s 7.9W instead of 12-14W). This board adds a third, though, and that is odd performance on the HPL (FP64) benchmark that is memory/thermal-constrained. No matter what I tried, I could only get a result half that of the MS-R1 and Orion O6. See more here: Benchmark MetaComputing AI PC Framework 13 Mainboard with Cix CP8180 · Issue #94 · geerlingguy/top500-benchmark · GitHub
Their kernel is based on 6.6.89.
Most of the commits they have done since then are fairly minor, except for this one:
That one is huge, it would take a while to get that commit so that it can be applied to a more modern Linux kernel. I.e. the mainline kernel. But at least the source code is there, so the task is easier than it being a binary blob kernel.
I did a similar exercise on another board, so it could use modern Linux kernels:
check upstreaming status…here.
@yuning_liang is there any chance that CoreBoot support is on your roadmap for this? I would love if this mainboard was fully open source.
Definitely if you guys want to buy some and help on coreboot implemtation
I was considering this, but in order to do that, I would need to know information about the initialization process for that CIX p1 chipset which Im not seeing available online. Is the silicon vendor willing to publish that information?
you should be able to see how edk2/kernel upstreaming code
It is really only x86 cpus that have complex binary blobs to start them up.
Arm cpus are far simpler, and their equivalent of bios, UBOOT, has been open source for a long time.
In summary, i don’t think the bios aspect of a ARM mainboard is a problem at all.
Wishing to add/use ACPI/edk2 on ARM is some added complexity that some people are doing. I don’t really see the point, as device-tree has worked fine up to now.