ROI for ETH FPGA Miner


We have started an interesting column to address the Return of Investment [ROI] for FPGA miners, and lets start with Ethereum [ETH].

ETH is a memory hungry algorithm, and its difficulty is climbing through the roof. Back then, when I started mining the DAG Epoch file size is less than 100 and could make $4 per day with 1x 1060 6GB. The ETH algorithm is designed in such a way that the DAG size increases every 100 hours of 30000 blocks. Hence, it is impossible to stay on the same hardware in case of ETH mining. A lot of GPUs with higher Mem Bandwidth arrived which helps to greatly increase the difficulty and at once stage lot of low end GPUs become obsolete.

In FPGA, Mem bandwidth is too low, like 64 GB per card [some have more]. In our earlier post, it is clearly mentioned FPGA cannot produce more than 1 GH/s hashrate for ETH. Read it here: 1 GH/s ETH FPGA Miner is fake. At the time of writing this post, FPGA for mining costs $4200 per device. According to devs, ETH bitstream never crosses 150 MH/s no matter what, so ROI will be more than 3.5 years.

 FPGA required around 330W to operate at full speed, so I took 400W as the optimum power keeping in mind the CPU power and other devices. Even with intensive overclocking and 24x7 runtime, it will take more than 3.5 Years to get your investment. Within this time, 2 things will happen- Difficulty increase, POW change to POS. If we take that into account, the ROI will be further delayed. 

I will never recommend an ASIC or an FPGA for ETH mining. GPU is always the best for ETH mining. If you have more money that you want to try hands on, get an FPGA and play around it. If you are tight on budget, or look to make some reasonable money, stay away from it. 

Thanks for reading, if you like it please share.

No comments