- separate semaphore for write queue (about 2% more speed) - re-enabled init optimization (avoidance of full chip resets, still experimental)
- added device.i in preparation of moving some operation sequences closer to the hardware implementation, if desired - added picture of V3 card