Nope … we have not ‘given up’
In fact we have made some excellent progress
Previously we had maxed out the Ebay scrap at 150MH/s per board,
after a few modifications to the VHDL we are now at 180MH/s
There is just one issue related to a PCB that was dropped, the board edge impacted a glass table so badly that there is a 5MM chip in the PCB edge, Incidentally it is the one that has the high failure rate on hashes.
How far is too far
The results of the experiment were well worth it. Along the way we have gained an extra 30MH/s per board , which is the equivalent of adding an extra 1.5 boards to the farm.
I expect to push the boards to about 200MH/s in the next day or so.
Taking into account that the DC/DC convertors on the boards are physically tied to 0v97 ,this increase in performance can be considered a solid result.
It is unlikely that we will be able to push the boards beyond 200MH/s, this usually requires the FPGA core voltage to be around 1v0~ 1v02