chrysophylax
I see that you finally managed to make diff setting based on Rig hashrate.
Thank you. Good job.
P.S. Still some finetuning is possible:
As I understand 100+Mh - get 0.97
And lower HR get 0.28 or 0.56 (70+Mh?)
Still 5*1060 and 6*1080 get the same 0.97diff but it is 107MH/s vs 242MH/s difference.
IMHO something like this is more adequate:
>150 = 0.96
>100 = 0.64
>70 = 0.32
>20 = 0.16
<20 = 0.08
I see that you finally managed to make diff setting based on Rig hashrate.
Thank you. Good job.
P.S. Still some finetuning is possible:
As I understand 100+Mh - get 0.97
And lower HR get 0.28 or 0.56 (70+Mh?)
Still 5*1060 and 6*1080 get the same 0.97diff but it is 107MH/s vs 242MH/s difference.
IMHO something like this is more adequate:
>150 = 0.96
>100 = 0.64
>70 = 0.32
>20 = 0.16
<20 = 0.08
the max difficulty target on the lodiff port is capped at 0.97 ...
the reason is that MANY have single card and dual card systems ...
if you need to go to the hidiff port - just remove the --lodiff parameter ... and any of the netdiff / temp control parameters ... you will find a smoother run and better processing for the gpus ...
being a vardiff stratum also - there is no manual control of the diff for the miner ... as the stratum controls that side of it ...
whereas hashrate will not be affected with difficulty of shares being processed ...
so i dont understand what the issue is with your two machines getting different hashrates ... they need to - as they are different cards pointing in the same port capped at 0.97diff ... the hidiff port goes all the way up to 7diff - as we had to cap that to 7diff from 7000diff ... that port is something that will test your systems
...#crysx
I mean that 0.97 for 6*1080 is fine (242Nh). 0.97 for 6/1070 is OK (180Mh)
But 0.97 for 5*1060 (107Mh) is not OK because - too rare shares ->big fluctuations in hashrate on the pool.
0.56 for 2*1080 (76) is not OK because no shares for 5 minutes is not OK in any way. The Rig is even lost from pool. (take a look at the picture I posted)
That's why I posted my vision for diff levels for different hashrates.
ok - understood ...
the difficulties do align - but it takes a lot of time to do so ...
the only way to do this with the stratum currently - is to set scaling diff with separate ports ...
that would be a very messy way to do it - and have more then 3 ports to solve ALL the issues ... even then - it would be upto to the user to allocate the right port - and our experience so far is dismal to say the least ... which is why the growth and development of the miner itself is very important ... not just the hashrates as SO many people are asking ... and we have been working on that ... this unfortunately will take time and a lot of errors before we can get the right adjustment ...
CWIgm is very young - not like ccminer or sgminer ... they are mature products ... which is why we wont pull CWIgm out of testing until we are confident that we can have a full release that 'ticks all the CWI boxes' ...
when diff is too high and the miner processes - you will see it drop off the pool momentarily ... its still there processing a high difficulty share and WILL submit it ... the miner is accounted for the high diff share as it carries a lot of weight compared to a miner that submit very low diff shares and many of them ...
but this sort of feedback is good - as we can adjust our short term developments to reach our goals ...
tanx ...
#crysx
