Framework is a relatively new laptop manufacturer that's building very user-serviceable laptops. In particular, those laptops are very modular. The main board itself can be purchased separately and there are already third party cases for them (notably they've worked with Cooler Master). I started thinking that these board might be small enough, and low power enough, that maybe you could make a cluster of them in a server. Of course, I wasn't the only one who thought of that. This is so cool ; a high density x86 blade server out of laptop main boards ! The mess of cables is very reminiscent of the OG raspberry pi clusters. I would LOVE to see someone work on a high density board for these framework main boards to deal with this mess of cables.

My TP1 has two CM3+ at the moment. I've since ordered a used CM4S off eBay and I'm curious about availability on CM3, CM3E, and CM4S otherwise. Has anyone used CM3E or CM4S on their Turing Pi 1?

I ordered one of these from Amazon: Cablecc Mini PCI-E PCI Express to 6Gbps Four Ports SATA 3.0 Adapter Converter Hard Drive Extension Card for SSD https://a.co/d/2y9RUdi. This is a full-length, 4-port SATA mPCIe card based on the ASM1064 chip. We know that 2-port SATA, half-length cards, based on the ASM1061 chip, work fine on the TP2. The ASM1061 supports one PCIe Gen 2 lane with a maximum throughput of 500MB/s. The ASM1064 supports one PCIe Gen 3 lane with a maximum throughput of 1GB/s. The ASM1064 won't yield scalable performance for CM4s, but it will take full advantage of the PCIe Gen 3 lane available to some Jetson SoMs and the RK1.

The picture shows around a 1” gap in using the Nvidia Fan. I can’t reroute the cable out a different slot as the fan is fixed in position (no screws). Anyone got any ideas?

Two related questions. For the "1x Case FAN header - 2 Pin small Molex Fan Power Connector" on the motherboard, is it expecting a 5V or 12V fan to be connected? And related - what is the pinout for the 2 pins? Assuming one is ground, is it the pin closest to the "back" of the board (direction of the network ports), or the pin closest to the "front" of the board (direction of the SATA ports)? Thank you.

While waiting for further updates to the BMC firmware, I thought I'd share two scripts that help me keep the home network chugging along. The first is an init script that lives on the BMC at /etc/init.d/S99timstart.sh. It automatically starts a Raspberry Pi (in Node 4) which runs core network services and my Zigbee lights across the house. The script waits until the BMC's API is up and responding, then assigns the USB port to Node 1 briefly to clear some kind of error, reassigns USB to Node 4 and then powers up Node 4. (It took*** a lot*** of troubleshooting to discover that my Zigbee controller issues weren't a bad power cable or my Pi 🤨 ). This script is pretty barebones, but deliberately designed to only use tools already installed on the BMC. The second is a bash script that lives on the Raspberry Pi and allows for API-style shutdown requests via https by the BMC. I'm serving it via nginx's built in fastcgi feature. The BMC's init script makes a call as part of the shutdown routine. There's some pretty rudimentary anti-replay security in the script (mainly to stop me accidentally nuking the Pi with a bookmarked link) and I also restrict access to local IPs via nginx rules. This isn't intended to provide Fort Knox-levels of security, but I wanted a solution that didn't require me to keep SSH keys on the BMC and also minimised replay attack risks on the API. Sharing as these features are frustratingly absent from the default firmware, and I can't be the only person that has this use case. It'd be very simple to extend this skeleton to multiple host power on scenarios. I'm also keen for any feedback on the script, and any tips on if there's a programmatic way for the BMC to determine if the Pi is fully powered down. I suspect not, but would be keen to do something more precise than my "guessing 15 seconds is approximately right" approach. Keen for your thoughts and feedback

Just out of curiosity... https://sipeed.com/licheepi4a/

So I noticed yesterday that the Jetson Nano actually has the capability to remap some of the I/O pins - it's recommended to do it for each carrier board. I wonder if it's possible to unlock some of the missing Jetson functionality this way

Are there plans to release datasheets for these? I'm interested in using the RK1 in a custom carrier board as a more cost effective alternative to the Firefly Core-3588J.

I have a Turing Pi 2 sitting in its box on my desk, but haven't found the CM4 modules with 8GB of RAM that I want for a not absurd price. Is there some way you all are finding these or are you waiting for the RK1's to be developed? Really want to put this thing to work, but it's been difficult to fill it. Will backers of the TP2 get dibs on the RK1's?

I know it's been mooted a few times but I don't know if there's a formal topic on this. I understand that the factory CM4 carrier boards route PCIe to the node specific PCIe controller (raw PCIe, SATA, USB, etc) for good reasons - cost, design history, usefulness etc. As I understand it, there are two alternative designs: routing it to the NVMe instead, or putting in a PCIe switch to allow both to be used (sharing bandwidth). What would it take to get such an alternative carrier board? Are the designs for the existing board available somewhere as a starting point? I am aware that there are careful considerations like keeping the PCIe traces the same length - what else would one have to consider if they were to try to design an alternative? Is it possible for a motivated technical amateur or do you really really need specialised design knowledge?

Has anyone tried to allocate IP addresses based on which slot the node is inserted into?

Does the turning pi 2 support lacp bonding of the two ethernet cards? If yes, how is it configured? I couldn't find anything If not, any reason why? Is it a software o hardware issue, if software, are the plans to add support to it in the (near) future? Thank you

Hi everyone! 👋 My name is Abe Hoffman, and I'll be developing the install process for Ubuntu 22.04 LTS (Jammy Jellyfish) for the RK1 compute module. I have been into hardware and software since I was eight and now produce a "discovery" operating system for PayPal's entire compute stack. I'm also a Computer Scientist and have several patents related to operating system installation and software compilation. I have long-term hardware experience with five major manufacturers at the enterprise level and love ARM! I am most familiar with the ASPEED family of BMCs. I also contribute to Amazon's Firecracker for aarch64. I am delighted to be working with the TuringPi team and family (you all)! I want to open the floor to the community and hear what you all would like through the process. My goal is to create a dead simple and enjoyable OS install process for the RK1. Please tell me your thoughts!

Anyone modify the kernel flags on the Orin NX? I need the following flags set in order to use cilium for my Kubernetes cluster. https://docs.cilium.io/en/stable/operations/system_requirements/#base-requirements

CONFIG_NET_CLS_BPF=y
CONFIG_NET_SCH_INGRESS=y
CONFIG_CRYPTO_USER_API_HASH=y
CONFIG_CGROUP_BPF=y

CONFIG_NETFILTER_XT_TARGET_CT=m

CONFIG_NET_SCH_FQ=m

I'm fairly new to this level of tinkering with Linux. I asked chatGPT, and got a pretty straight forward response:

cd /usr/src/linux-headers-5.10.104-tegra-ubuntu20.04_aarch64/kernel-5.10/
cp .config .config.bak
make menuconfig
make -j$(nproc)
sudo make install
sudo make modules_install
sudo update-grub
sudo reboot
uname -r

I'm a bit confused on the GRUB part, as it seems the Jetson NX is using EFI. Checking

which update-grub

returned nothing, so I'm leary to continue. Anyone done this already? Could these flags be added during the preliminary flashing process?

Welcome!

Ryan Berry

joined #│forum.

Given that I am a complete Noob with all this, I am struggling a bit to know what all I am looking at now that it is running. I have played with Linux some (I built a Fedora based file server nearly 20 years ago as an experiment; used the GUI and was not headless - not really familiar with command line or SSH.) I am using an older Macbook running Catalina for communication with the TPi2 for now, though I guess I will have to borrow the wife's Windows machine for updating the BMC (or am I better off waiting for an easier update method??) I did manage to get UART working to watch the BMC boot, but when it is done, I cannot seem to scroll back through all of the information. Screen grab of what I can see is attached. Where does the rest of the information go? Also, it says "network is unreachable" so how do I determine IP address to access the web interface? I did install ADB, but not sure how to use it yet. My setup so far: I am using a Thermaltake Core V1 case with a 450 watt power supply (I know, overkill... I messed up when finalizing my TPi2 order and ended up with a power brick but no PicoPSU, so using what was easier and cheaper to get at the time.) The TPi2 board is raised 12mm on standoffs, in order to provide clearance for the M.2 slots when I can use them, so I modified Dan's printable IO shield to match. Nodes 2 and 3 are set up with CM4 002016, with heatsink/fan combo from Waveshare. To control the fans, I am using a 2 channel PWM controller from Amazon, with a self designed 3D printed mounting plate attached to existing holes in the case. I have one RK1 on order for now, will hopefully eventually get at least 1 more to fill out the board. I have enjoyed the process of getting the TPi2 set up, now to learn how to use it.

Hi does anybody have any experience with an extension mini pci-e card that provides hardware RAID 1 SATA support? If so what card can you recommend? My goal is to create a data drive that is highly reliable (by having a copy of the data on the second drive). Want to manage the risk of drive failure.

Specifically the specs for the Jetson Orin Nano Compute Modules. They have NO eMMC storage as stated here: https://docs.turingpi.com/docs/nvidia-jetson-orin-nx-intro-specs Please see the attached data sheet if you have questions. https://cdn.discordapp.com/attachments/1140800114928144424/1140800115163009165/Jetson_Orin_Nano_Series_DS-11105-001_v1.1.pdf

Will the RK1’s have access to the gpio and if so, can it be configured to operate as CAN BUS? Thanks

Mostly because i thinkt hese would be great to play with ESXi on them. 🙂

Hello I have been working on a number of RK3588 boards and would like to inquire about the RK1 to gather more information. Such as where development resources / firmware are or how I might be able to access it, from my understanding it's open source? I have attempted to reach out to Turing multiple times via email and their help page but have not received any response. Does anyone know of any proper contact information so I may inquire further?

Do ya'll plan to expand the processor offering to include riscv rv64g cores? Thanks in advance! PS - If ya'll are looking into riscv; as a word of warning, the current set of riscv boards/chips boasting an implementation of the vector spec have implemented unratified (pre-release) and outdated versions of the official ratified vector spec. Strongly encourage sticking with rv64g - the base set of extensions - for now.

I attempted to suggest an edit but it wouldn't let me, but on this page https://docs.turingpi.com/docs/nvidia-jteson-orin-nxnano-cooling would be cool to mention the torx t6 is what you need for the screws. save people some time hunting for the right thing 🙂 Also should be a document tag or wiki tag to tag posts for suggested edits if we're not able to suggest them directly on the documentation

In its default configuration the Jetson Nano module is perpetually deprived of sufficient storage. The full default install with all SDK components and software packages clocks in at just under 15 GB, leavig little to no space for maneouvering, let alone running arbitrary docker images. I have 2 Jetson Nano Modules: One 4 GB devkit module with SD card storage (P3448-0000), and then the non-devkit version with 16 GB embedded MMC storage (P3448-0002). I purchased two 1 TB NVME SSD cards early on, as I assumed from [this overview](https://docs.turingpi.com/docs/turing-pi2-specs-and-io-ports#mpcie-m2-nvme-sata-usb-ports-mapping) on the TP 2 docs that it is supported and would be straightforward to set up. But unless one is comfortable either just mounting 1 TB in random locations and using it only for personal storage or cleanly migrating an entire directory tree to the NVME device, it was inherently unsatisfying that the rest of the system was still squeezing itself onto the embedded limited storage. It took me almost 2 weeks to figure this one out, but I finally managed to get the Jetson Nano modules to mount the root filesystem onto the NVME drives during boot and use it as the main storage device. They still boot from the eMMC/SD card, but I am 100% fine with that. The caveat in this setup is that boot and kernel files are duplicated on the MMC storage and the NVME drive, but people cleverer than me can figure this one out.

Welcome!

turingpi

joined #│forum.

Can I just install the OS/JetPack on the NVME drive using the dev kit carrier board, then toss it into the Turning Pi 2 with the Orin Nano? Seems that it'd be much easier and won't require any special cable/ adapters using the instructions.

Has anyone used the turing pi 2 and the RK1 to set up a web server for hosting at home? I am currently building a website for my business and would like to use the turing pi 2 and the RK1 chips to host the site. Also, is it feasible to use Blender with the RK1 to add dynamic features to the site? And then use another slot on the board as the actual server for hosting? In theory: - Chip 1 would host the site - Chip 2 is set up for site traffic (creating a duplicate image of the site) - Chip 3 and Chip 4 would be used to manage background functionality and content of the site (Blender, images, etc..). I am very new to this. Any feedback is appreciated. Thank you in advance!

Hi, I recently came into some CM3+ modules and the Turing Pi v1 seems like the best use of them. Problem is that I can't find one. Is anybody looking to sell a carrier, sans modules? Thank you

I may make an unconventional request. Does anybody here have some experience with creating virtual hardware? What I mean by that is. Would it be possible to have the Turing PI board as a virtual machine? I'm not referencing making 5 machines; but rather an accurate representation of the Turing PI hardware. The reason is that this would maybe very helpful in developing the firmware. If it's possible to spin up a virtual machine of the TPI hardware, then contributing towards the firmware development might become easier without you having to brick one. Also people who have one can keep that one as their production machine while using the virtual one to help with development. I have no experience in creating virtual hardware, and I do not know what is required. but maybe someone here does know how to do this. This is just an interesting thought.