Out of the Box Performance
After pulling it out of the box and installing the GTX 1080Ti into the new rig, which is running Windows 10 Pro 64-bit, I installed the latest Nvidia GeForce drivers (version 382.33 as of this writing). I then downloaded and ran Claymore’s version 9.4 Ethereum miner and got a mediocre 30 Mh/s which is about what a RX580 can produce for much less cost. As with the GTX 1080, this card was not bought for Ethereum (Ethash) mining, but was acquired specifically for Zcash (Equihash) and related algorithms. Since Claymore does not provide his ZEC miner for Windows, I used another popular mining software for Zcash, EWBF’s CUDA Zcash Miner.
I am using a simple launch string in a bat file to launch the miner:
miner --server zdash.suprnova.cc --user Z-Yeti.GTX1080 --pass x --port 4048
I mine different Equihash algorithms depending on daily profitability, today it was Hush (formally Zdash). I have found the mining results are pretty much the same across all the equihash coins, so Zcash, Zen Coin, Zclassic and the others should all give similar results.
EWBF is very nice in that it has a built-in performance meter you can enable (use –pec in the batch file) where you can see a real-time Sols per Watt statistic. With Zcash Sols is basically the equivalent of saying hash/s, so you may hear these terms often used interchangeability. As you can see in the screenshot above, I am also using MSI’s Afterburner software to control the clocks. The default settings and performance are shown above for the EBGA GTX1080Ti and we can see we are getting around a 2.7 Sol/W efficiency rating at 100%PL (power limit). The GPU is running at 1873 Core clock and 5005 Memory clock, with the default voltage displayed at 1031mv. It is also running at a comfortable 68C with fan speed set to automatic mode.
Note: I see after taking the screenshots that something broke the voltage monitoring portion in MSI’s Afterburner. It may be the Corsair Link program interfering, I will try and troubleshoot it, but the test results remain valid as the voltage is controlled automatically by the Power Limit setting and the main thing I am concerned with is the overall wattage draw anyway.
In addition to EWBF mining program and MSI Afterburner, for this review I am also using the Corsair Link program which shows an overview of the power supply input and output wattage. In past reviews I used the Kill-a-Watt meter to measure real-time power draw from the wall, and I still do, however the Corsair Link allows me to capture all information in a single screenshot for easier correlation at a later time. I did notice the Kill-a-Watt shows a consistent 10 watts lower reading than the Corsair Link’s input wattage reading (which should be the same) but this still seems to be well within the tolerance levels both products specify for error margins. So either the Kill-a-Watt or Corsair (or both) measurements are a bit off, but still close enough to not put too much worry into as normal running power fluctuations will vary by 10 watts or more.
Cranking things up a bit just to see what it will do I adjusted the PL slider to 105%, and pushed the core to +140 and memory to +300 to get an impressive 756 to 772 Sol/s. The extra power allowed the core to adjust to 2012 MHz giving a health boost to hashing power, but the overall watt draw also jumped up to 313 for the rig, and 265W for just the GPU, resulting in the efficiency rating dropping to 2.88 Sol/W. If you have free or cheap power, this option may be more attractive to you, but I do pay a healthy amount for power so I was looking to optimize my Sol/W raring a bit more.
Also for those wondering, I did do a few experiments with running the memory clock slider higher than +300 (which results in a 5300 MHz frequency) but did not see any further improvements in hash-rate after this point. So to save power where I could and to avoid needlessly stressing the card for no performance gains, I pretty much left it at +300 for all other tests
As can be seen in the next image above, by reducing the power limit to 78% and increasing the core and memory clocks by 150 and 300 respectively, we can get right around 700 Sol/s while pulling only 240 watts (power in reading) from the wall. Assuming the system draw is around 50 watts, the EWBF program is giving a pretty close estimate of the GPU only power draw of around 195 watts and a resultant Sol/W efficiency of 3.55 to 3.65. This is a pretty significant reduction in overall wattage (55 watts) for only a modest drop (< 10%) in hash rate, and with the efficiency reading jumping up to ~3.6 Sol/W I was getting close to where I wanted to be in terms of performance per watt.
My goal was to stay at or above 700 Sol/s while reducing power consumption as much as possible. While lower settings did reduce the power consumption, I could not keep the hash rates above 700 Sol/s so I ended up sticking with the above settings as my final values.