| Your blog is excellent, but your BM1387 numbers are a hair off. BM1387 has worse performance than 0.10J/GH @ the wall in practice. Bitmain has released 3 BM1387 based miners (S9, T9, R4). Only two are intended for industrial use and should seriously be considered (home mining is dead): * The S9 is advertised as pulling 0.098 J/GH +10% at the wall on a 93% efficient PSU. [1] * The T9 is advertised as pulling 0.126 J/GH + 7% at the wall on a 93% efficient PSU. [2] However, measured performance in the field tends to be a better indicator. BlockOperations tested the T-9 and S-9 with wattmeters and discovered that the Batch 16 12TH S-9 burns 1329W at the wall and the 11.5TH T-9 burns 1352W, for an efficiency of 0.111 J/GH and 0.117 J/GH respectively at a fairly cool (60F) room temperature [3]. If the room is hotter (most mines are) the fans spin harder and the efficiency gets a tiny bit worse. Only the earliest batches of S9s that had fixed frequencies achieved 0.10 J/GH, but they had serious issues with reliability and stability so bitmain switched to variable frequency, which helped the reliability and stability but hurt the efficiency. Another thing you might want to factor in is that the actual hashrate of the bitcoin network is higher than the network difficulty indicates. Remember, roughly 2% of generated blocks are orphaned and never counted, so you should adjust your estimate upwards somewhat to account for that missing hashrate. [1] https://shop.bitmain.com/specifications.htm?name=antminer_s9... [2] https://shop.bitmain.com/antminer_t9_asic_bitcoin_miner.htm?... [3] https://blockoperations.com/power-comparison-antminer-t9-ant... |