Project update 65 of 75
This is so under-dramatic. You’d think that 1,000 PCBs would take up more space. The reality is each PCB is only 46 mm x 78 mm x 1.2 mm. If stacked in a box 50 cm x 40 cm they would come to a height of only around 30 cm. That’s only 20" x 16" x 12". The hilarious thing is: even when fully assembled, the cards are only 5 mm high, so that will be only around four boxes of the same size. That’s a filing cabinet’s worth of computing power, totaling two million MHz worth of CPU cycles (and around 1,000 GFLOPs) and only drawing around 3 kilowatts if they were all powered up. For a brief period of time, I will have that supercomputer I always dreamed about. Ah, well, easy come, easy go. (Wait: it wasn’t easy!)
1,000 EOMA68-A20 PCBs
So, the next phase is to get all the Micro Desktops made up, and 100 of the v2.7.5 EOMA68-A20 boards made up and tested. Then, if they’re good, we do the remaining 900. My friend in Hong Kong is going to lend me a laptop - we will install Debian GNU/Linux on it, along with an ssh tunnel script, so I can get access to CPU cards plugged into it and use FEL-USB uploads to test that they work, remotely, from outside of China.
Also, Joshua kindly pointed out that the RYF Certification will need
an OS. Unfortunately, with the Parabola master microSD Card having
been corrupted (completely destroyed), the work done on getting it up
and running will need to be completely redone. That includes getting
sysvinit onto it. I have a stock root filesystem: I can’t
use it (via an upgrade), as there were too many changes made a year
ago, so I’ll have to do a complete rebuild. That involved a
pivot_root onto an NFS root mountpoint within qemu running qemu: qemu
parabola-x86 running qemu parabola-armhf. Funfunfun.
In the meantime, the new pre-launch page for the Libre RISC-V SoC is up and starting with updates, too. If you’d like to follow that adventure, please subscribe to project updates there.