Can dgbg.suprnova.cc be fixed to correctly reward work being done with ccminer?

I have a rig with 3 older AMD cards and 2 new 1070s. I'm mining on the AMD cards with sgminer-x11mod, and on the NVidia cards with ccminer-8.11-KlausT. I have a separate worker on suprnova for each of the two sets of cards.

sgminer reports a hash rate of 115 MH/s. ccminer reports a hash rate of 160 MH/s. Both miners are running at difficulty 0.139 (sgminer reports accepted shares with Diff 0.201/0.139, ccminer's last line showing stratum diff change says "Stratum difficulty set to 0.139063") and diff appears to be stable once the miners reach 0.139. ccminer and sgminer get to this diff in about the same amount of time regardless of whether I use the worker password or the "d=0.139" diff setting in place of the password.

After both miners run for 30 minutes, the AMD cards have found 318 shares, 315 accepted, 3 rejected. The NVidia cards have only found 163 shares, 160 accpeted, 3 rejected. In other words, ccminer is hashing 140% faster than sgminer, and only finding half the shares at the same difficulty in the same length of time.

Worker stats on the dashboard look like this:
     Worker             Hashrate     Difficulty
****.***AMD      123,253.49      35.6
****.***CUDA       45,790.61     35.6

I have no idea how the dashboard difficulty of 35.6 relates to the miner reported difficulty of 0.139, at least the dashboard shows difficulty is the same for both miners.

The effective hashrate for the CUDA worker is the main problem on the dashboard. The AMD worker hash rate appears to be correct, it is most often centered around the miner hash rate of 115, + or - 10, with rare outliers as low as 75 or as high as 155.  The CUDA worker hash rate is never close to the miner reported hash rate, it is most often centered around 42 (26% of the miner reported hash rate!), + or - 10 with rare outliers as low as 22 or as high as 62.  

This degraded hash rate shows up on the suprnova pool when using any of the following versions of ccminer, v2.0-x86, v2.0-x64, v2.1-tribus, and v8.11-KlausT. Playing with ccminer's -f and -m options range from no effect with settings close to 1, an even lower acceptance rate with -m 2, or a >50% rejection rate with -f 2.

Until the problem with ccminer on dgbc.suprnova.cc can be fixed, what other pools are people pointing their miners to for DGB Myriad-Groestl?