@ Zefir
are there any news about the rigs. You had a working pcb of the rig, right? Then they produced a 3rd version. Is the production plan stil valid?
Do you know anything about the mass production of the desks?
Thanks for caring.
I am not part of the management and as such not involved in the daily business. When I knew more and posted it here, people took my feedback and misused it for legal actions against Bitmine, which in the end rendered transparency very expensive. Since I am open by nature and dislike keeping track of what I can disclose and what not, I asked Bitmine to withheld any information from me that is not meant to be shared publicly. With that, effectively my knowledge about delivery and progress status is the same as everyone's else here.
Nevertheless, the SW part I am working on is something I can provide - hoping it is informative only and not suited to harm Bitmine even more.
The 2nd revision of the CCR board was build around a STM32 uC who acted as a proxy to distribute the SPI traffic from RPi to two 6-chip chains. This was basically done to cascade the processing and shift load from RPi to the uC. The core issue addressed was to ensure RPi can handle the large amount of chips in a CCR system. From operating the CCD we learned that a RPi is sufficient to operate a fully populated CCR; from testing rev2 boards we also learned that the STM32 has its own limitations (like issues when running i2c in parallel with 3 SPI buses) and over-complicates system design for a limited gain. Therefore, the rev3 board abandons the uC approach and instead uses an SPI multiplexer to map the up to 16 chip chains to RPi's SPI bus directly. With that, from the SW side a fully populated CCR consists of 16 6-chip chains - which is way better controllable compared to a STM32 based proxy.
I finalized the driver for the new CCR board last week and will push it upstream as soon as I know the productization step went fine. And that is the key point here: I tested the board under lab conditions and it worked as expected (nominal + 20% over-clocking). But operating a single board attached to the PSU with chilled cooling is one thing - stuffing 8 of them into a closed case, fed by a single PSU and operated at arbitrary environmental conditions is a completely different story.
I wish I had enough confidence with the results observed to give an estimated shipping date, which most of the users here badly deserve. But I (and also Bitmine management) were tricked often enough to announce products' availability based on working prototypes - and we all know the outcome.
HTH
