Author

Topic: A Comparison of ewbf, dstm, and bminer (Read 833 times)

newbie
Activity: 6
Merit: 0
March 03, 2018, 12:42:59 PM
#3
Which versions of all of those miners have you used?

Currently, latest versions are:
bminer 5.4,
ewbf 0.3.4b and
dstm zm 0.6
newbie
Activity: 2
Merit: 0
January 24, 2018, 12:32:36 PM
#2
Great work on the analysis! I've tried to do the same but I only have 2 1070TI so the results are noisier. Grin

Same for me tho when I tried, dstm had an error halfway while the other two were fine through the day.
newbie
Activity: 61
Merit: 0
January 24, 2018, 12:48:47 AM
#1
A Comparison of ewbf, dstm, and bminer
--Which Equihash NVIDIA miner runs faster?

Hi guys, Ewbf,Dstm & Bminer are the 3 equihash (ZEC, ZCL, & BTG) miners for CUDA GPUs. There is no systematic comparison between these 3 miners to figure out which one is the fastest, so I did it.

Here is my conclusion: Bminer is the fastest miner and Bminer is also very stable. I will explain my experimental results in detail.

FAQ
****What is equihash?
Pls visit this blog: https://z.cash/blog/why-equihash.html
****Why should I mine Zcash?
You can visit whattomine and make a simple calculation by tick your GPU and the device number you had to figure out the most profitable coins you can mine, and it will show you a list that the most profitable coins you can mine on 1080ti/1070 are ZCL(zclassical) ,Zen, BTG and Zcash(Zec).
****What Can I mine with  Equihash NVIDIA miners such as Ewbf,dstm & bminer?
Equihash miners can not only be used to mine Zcash but also zcl(zclassical),zen, and BTG; you can switch the coins you want to mine anytime.

My experimental setup

I tested three miners on my personal mining rigs in Ubuntu 16.04. Each rig contains one NVIDIA 1080ti GPU and five NVIDIA MSI 1070 GPUs. I downwatt my 1080Ti to 170W and my 1070 to 115W for power efficiency, because there is a maximum power I can draw for my house. All GPUs are in the stock settings.

This screenshot presents nvidia-smi output:
https://preview.ibb.co/dQYFGw/Pic1.png

Here are the result of my testing and How I did it:

I tested the 3 miners one by one on the same 10 machines. I ran them for more than 10 hours each and measure the reported average 6-hour on nanopool. I got the speed on the pool site for 60 GPUs so the result are convincing.


Miner   Avg latest  6-hour hashrate from pool(sol/s)
Ewbf        25476.5
Dstm       26528
Bminer    27253.7


As you can see based on the hashrate reported by nanopool, Bminer is faster than Ewbf by 7.0%. Bminer is also faster than Dstm by 2.7%.

Results screenshot:

dstm running results on one machine

https://preview.ibb.co/dQYFGw/Pic2.png

dstm results of 10 machines on nano pool:

https://preview.ibb.co/dQYFGw/Pic3.png

Bminer running results on one machine:

https://preview.ibb.co/dQYFGw/Pic4.png

https://preview.ibb.co/dQYFGw/Pic5.png


Bminer 5.1 results on pool side:

https://preview.ibb.co/dQYFGw/Pic6.png

Ewbf results:

https://preview.ibb.co/dQYFGw/Pic7.png

Ewbf results on pool side:

https://preview.ibb.co/dQYFGw/Pic8.png


Notes

About devfee: Bminer, dstm, and ewbf have 2% devfee. I tested ewbf with both normal and -fee 0 options. I see no difference in the hashrate. I cannot tell whether ewbf still collects fees with -fee 0 or not, but at least you do not benefit from using -fee 0 in ewbf.

2% devfee is pretty high. Unfortunately, it seems we have no open source alternative for Equihash that is even close to bminer and dstm.

About stability: I did not observe any crash from bminer and ewbf during my experiments. For dstm, during my first attempt, it crashed on one of my rigs after running for 4 hours. I had to restart the experiments on dstm again for collecting its data. So for me, bminer is more stable than dstm.
Jump to: