using the new .1 beta and it seems like the stratum difficulty rises too sharply... i dont know if this is ok.. but im not seeing as many shares as i have before when i had .1 diff found much quicker and many more of them.. here is a sample below
[2017-08-04 11:18:53] GPU#0: MSI GeForce GTX 1080 Ti, 48.97 MH/s
[2017-08-04 11:18:53] [S/A/T]: 0/284/285 (diff 8.966), 102.47 MH/s yes!
[2017-08-04 11:18:55] GPU#1: 2006 MHz 195.94 kH/W 269W 54C FAN 92%
[2017-08-04 11:19:25] skunk block 24286, diff 18008.502
[2017-08-04 11:19:41] GPU#0: MSI GeForce GTX 1080 Ti, 48.65 MH/s
[2017-08-04 11:19:41] GPU#0: 2017 MHz 196.99 kH/W 252W 53C FAN 85%
[2017-08-04 11:19:41] [S/A/T]: 0/285/286 (diff 16.091), 102.46 MH/s yes!
[2017-08-04 11:20:05] GPU#1: MSI GeForce GTX 1080 Ti, 52.99 MH/s
[2017-08-04 11:20:24] GPU#0: MSI GeForce GTX 1080 Ti, 49.49 MH/s
[2017-08-04 11:20:46] GPU#1: MSI GeForce GTX 1080 Ti, 53.01 MH/s
[2017-08-04 11:21:03] GPU#1: 2007 MHz 195.19 kH/W 270W 54C FAN 95%
[2017-08-04 11:21:08] GPU#0: MSI GeForce GTX 1080 Ti, 48.31 MH/s
[2017-08-04 11:21:25] skunk block 24287, diff 18335.006
[2017-08-04 11:21:52] skunk block 24288, diff 17048.673
[2017-08-04 11:22:04] GPU#0: 2020 MHz 200.74 kH/W 247W 52C FAN 85%
hi helladave ...
what is being shown there is the network difficulty - and the share difficulty that the gpu has already processed ...
what you are describing is correct ... but its not from the miner - its from the pool stratum ...
we originally had the stratum retargeting the diff every 70seconds ...
what we found was that for the higher hashers that did NOT use the hidiff port ( 6000 ) and used the lodiff port ( 2000 ) instead - the shares streamed in and flooded the stratum ... now this was not necessarily the miners fault - as CWIgm had built in card hash detection which was supposed to alleviate this ( even though it is an experimental 'feature' ) ... but due to the influx of miners all at once - the 70second retarget for change of diff was just too long ...
now the diff retarget is set at 25seconds ...
so what you are seeing is the stratum changing the diff with relation to how many shares are sent by the miner - in a time frame of 25seconds ... it will increase and decrease accordingly - and the longer you spend on the pool mining - the more accurate the stratum will be ... but as with all stratums - it will continue to fluctuate the diff to try and get a good share submission as an average ...
this doesnt mean you are losing out in anyway - it just simply means that the shares you are submitting are a higher difficulty - and as such take longer to process ...
in any case - the higher diff shares submitted - the better the chance of solving a block ... of course there does need to be a balance of share submission and difficulty of shares - and the stratum is a little quicker at doing that now ...
#crysx
