Is there a main stream version of Linux (other than Android) that has been compiled for the this yet? I know that there are for the processor, but maybe not for the rest of the hardware? I know that you can run it containerized inside of the Android install anyway, but that slows things down and takes more memory. I am looking at the possibility of getting multiple sets of the 4 board and a cluster box arrays if an initial proof of concept test works. The idea would be to perform distributed computing in a business’s own private cloud in order to have better control over sensitive data or lower latency. So optimizing both performance and security with a real version of Linux running directly is important for the uses I see, as is the ability to connect at least one SSD somehow to each Blade, handle the TRIM on the drive properly (ideally, probably a boot and software drive on the intended connector, and a couple mirrored SATA SSDs on the USB3). Plus I would still need the ability to connect and pool the Ethernet connections, and use at least one video port on each for set up and debug. I know that this is all probably a while off, but I can really see the use for things like I talked about, plus more.
Hi @JRRT
Yes, we listened to the feedbacks and have changed the pre-loaded system from Android to Debian 11. We will open a remote web console (TTY) next week, you can try it by yourself.
For your requirements, I have two solutions:
- Connect one m.2/u.2 SSD to the board via PCIe3 Port A and leave the Port B for high-speed low-latency clustering. Connect several SATA SSDs to the native SATA3 port via a SATA multiplexer.
- Connect two m.2/u.2 SSDs to the board via PCIe3 Port A and Port B. Using Ethernet bond (2.5Gb + 2.5Gb) to get a high bandwidth link.
1 Like
This should actually meet my needs well. This would be my first use of the ARM architecture for a performance computing situation, and I still want to ask some friends some questions and I expect to initially only get one cluster as a proof of concept, but the energy use to listed performance to cost is appealing. Thank you.