Building a Cryptocurrency Mining Rig - Part 3

tags: +blockchain +diy +hardware

This is Part 3 in a series on building a cryptocurrency mining rig.

I physically constructed a mining rig in Part 1, and made motherboard BIOS adjustments in Part 2. Now, in Part 3, I’ll discuss how I financially optimized my mining strategy.

Baseline Benchmarks

I began the optimization effort by recording baseline mining benchmarks. The following was achieved using Claymore v9.6 to dual-mine ETC and DCR:

  1. Hashrate as reported by Claymore v9.6.
Mh/s (ETC) 1Mh/s (DCR) 1Watts

Though Claymore provides for it, I did not attempt to dual-mine SC, LBRY, PASC, or PASL, because I am personally not interested in those coins from a “fundamentals” perspective.

Mining Software Comparisons

When I built my miner, I initially chose the mining software mostly arbitrarily. Wanting to rectify that, I next performed comparative benchmarking to help determine which mining software to use, and whether or not to dual-mine.

Benchmarking yielded the following data:

  1. Comma-separated values are ETC/DCR pairs.
  2. Claymore charges a "DevFee" of 1% when mining ETC, and 2% when mining ETC+DCR.
  3. "Adjusted" rates account for losses accrued by the DevFee.
  4. Wattage was measured at the wall using a Kill-a-watt.
MinerCoinMh/s 1Fee 2Mh/s (Adj) 3Watts 4Mh/s (Adj)/Watt
Claymore v9.6ETC+DCR151, 15152%148, 14859600.154, 1.557
Claymore v9.6ETC156, 00001%154, 00008500.181, 0000
Claymore v9.8ETC+DCR156, 15602%153, 15299600.159, 1.593
Claymore v9.8ETC160, 00001%158, 00008700.182, 0000
Ethminer v0.12.0ETC160, 00000160, 00008800.183, 0000

My takeaway from the above was that dual-mining DCR:

  1. reduced potential ETC yield by about 5% (Ethminer (ETC) => Claymore v9.8 (ETC+DCR))
  2. increased power consumption by about 9%

So - was mining DCR worth it?

What to Mine?

I used to run the numbers:

  1. The numbers above were calculated on 1 Sep 2017.
CoinMonthly Profit (USD)

By the numbers, yes - DCR was (barely) worth mining. However, beyond the numbers, I had some concerns.

I had been dual-mining ETC+DCR for weeks before making the above benchmarks. When I checked my mining pool, however, I was surprised to see how little DCR I had earned. Investigating why, I discovered that the pool reported my DCR hashrate to be less than half of what Claymore reported. (I still don’t know which number was correct, or how the disparity came to be.)

I had additional concerns beyond the dubious hashrate numbers.

When measuring power consumption, I noticed that dual-mining was very “peaky” - draw would swing +/-100 watts perhaps a dozen times per minute. I feared that the temperature fluctuations that (likely) accompanied those swings would eventually damage the GPUs via thermal expansion and contraction.

Lastly, I found Claymore somewhat unpleasant to use. Configuration seemed awkward to me, in that it read configuration from text files (with a proprietary syntax) from its application directory. Likewise, it continuously downloaded .bin files (containing I-know-not-what) and logged output (again into text files) into the same directory. This all felt terribly disorganized, and inconsistent with sound software engineering practices.

In light of all of the above, I decided to forego dual-mining entirely, and to simply “single-mine” ETC using Ethminer.

Thus, with a revised mining strategy, I next turned my attention to optimizing my miner’s computational performance. In Part 4, I’ll discuss how I was able to (dramatically) reduce power consumption without impacting hashrate.

Proceed to Part 4 ยป