specs:
B85 mobo
G1840
4G ram
3*gtx 1080 ti cc+120 mc-1000
1*gtx 1080 cc+150 mc -1000
ubuntu 16.04 x64
cuda 8.0
driver: 384.59

EWBF 0.3.4b:

GPU0: 454 Sol/s GPU1: 672 Sol/s GPU2: 687 Sol/s GPU3: 679 Sol/s
Total speed: 2492 Sol/s
+-----+-------------+--------------+
| GPU | Power usage |  Efficiency  |
+-----+-------------+--------------+
|  0  |    153W     |  2.97 Sol/W  |
|  1  |    203W     |  3.31 Sol/W  |
|  2  |    198W     |  3.47 Sol/W  |
|  3  |    201W     |  3.38 Sol/W  |
+-----+-------------+--------------+

yours:

>  GPU0 Sol/s: 454.4  Avg: 454.4  I/s: 240.1  Shares/s: 0.000
>  GPU1 Sol/s: 684.6  Avg: 684.6  I/s: 365.5  Shares/s: 0.000
>  GPU2 Sol/s: 683.4  Avg: 683.4  I/s: 367.8  Shares/s: 0.200  ++
>  GPU3 Sol/s: 686.0  Avg: 686.0  I/s: 365.4  Shares/s: 0.000
   ==== Sol/s: 2508.5 Avg: 2508.5 I/s: 1338.9 Shares/s: 0.000
   GPU0 Sol/s: 433.7  Avg: 444.1  I/s: 237.4  Shares/s: 0.050  +
   GPU1 Sol/s: 668.1  Avg: 676.4  I/s: 362.4  Shares/s: 0.000
   GPU2 Sol/s: 689.6  Avg: 686.5  I/s: 364.5  Shares/s: 0.100
   GPU3 Sol/s: 677.5  Avg: 681.8  I/s: 362.2  Shares/s: 0.150  +++
   ==== Sol/s: 2468.9 Avg: 2488.7 I/s: 1326.5 Shares/s: 0.150
   GPU0 Sol/s: 445.0  Avg: 444.4  I/s: 237.3  Shares/s: 0.033
   GPU1 Sol/s: 670.9  Avg: 674.6  I/s: 361.3  Shares/s: 0.067  ++
   GPU2 Sol/s: 680.8  Avg: 684.6  I/s: 363.7  Shares/s: 0.100  +
   GPU3 Sol/s: 662.1  Avg: 675.2  I/s: 361.6  Shares/s: 0.133  +
   ==== Sol/s: 2458.8 Avg: 2478.7 I/s: 1323.8 Shares/s: 0.133

How about improving the performance? I have some new ideas, however I don't have access to testing hardware this week, so I'll see how they perform next week.                                                                                               
                                                                                                                               
I was thinking about reducing the dev fee for people that are helping development, providing feedback etc. - still not sure about it.

That is just 1% difference. So it is comparable.

You can't conclude that from these numbers. The solution rate is subjected to probability, it's inherent to equihash. You have to compare a lot more number and average them to have a more robust statement, this is why I'm computing an average in my output, so you can make better judgements how e.g. over/underclocking on your system affects the performance.

Actually worst results I've ever seen.
1080 - usually gives 550-570
1080 - 730+
Even 1070 makes 460-470 Sol/s

200W is 80% power assuming default is 250W.

I'm preparing for the next update. I've implemented a temperature controller, it's a pid-controller and adjusts pretty well. People who asked for it, how do you use it? I don't want to add unnecessary configuration complexity so - do you want to be able to set the temperaturelimit for each GPU separately? Or do u usually set one global temperature limit that's is used for all GPUs?

New Version 0.3.2

mp:  improve stability
con: support more pools
con: more robust reconnection support for unreliable
  servers which often drop the connection                                                                                   
ui:  report GPU temperature