Again I need to apologize for not giving enough updates. We were quite busy this year and also careful that Covid doesn't messes up our business. Update here: we are doing well and there is no risk that GRiSP-2 is not delivered or further delays due to business side problems. We were in a good starting point since we are a fully distributed company from the start.
The partner company who is helping us with hardware design and manufacturing however have been slowed down by Covid a bit since hardware development can't be as easy moved to home office than software due to high need of specialised equipment.
So, where are we at the moment: we needed to move to a more beefy SOM (System on Module) because the one used in the first prototype has only one USB interface of the CPU exposed. On GRiSP 2 need two USB, one for Wifi like on GRiSP 1 and one for the newly added USB host connector. We have one board with the SOM re-soldered to the final larger one that has passed all the hardware-guys tests and is now with us to test if Erlang still works and can boot from the SD-card and the built in eMMC (= basically a faster soldered on SD-card) which now resides on the SOM instead of the external one on the board.
After we finished this test successfully we are ready to push the button for producing the first batch. What does that mean exactly: the hardware devs have a few finalising steps to do which won't take too long to do (the layout is in the final state I hear). Then the production run starts. The expected duration for this is 8-12 weeks, depending on capacity and also part availability on the market. That also means the holidays are in the interval which might push the delivery date a bit to the back. We will get some feedback about the ETA from the manufacturer some time into the process (when they sourced all the parts).
Something good is also coming out of the delay caused by the change of SOM: the final hardware will have more CPU (792MHz) power and RAM (512 MB) because of the larger SOM. At the same time the power usage is only going up a little bit when this extra compute power is also saturated. For most real world applications there will be no change since the aim when making low power consumption systems is to race as fast as possible to the next sleep and then sleep as long as possible. So more CPU MIPS gives faster reaction time and because then the system sleeps earlier without more power consumption (we can also reduce the clock speed programmatically but that often does not save as much power).
Another good news I'd like to share is that booting is already much faster than on GRiSP 1 and that is without any optimisations and only from the SD card. eMMC should add more to that since it has more data lines than the SD card and therefore supports a faster transfer speed.