In my last test ethminer (previous version) has slowly than claymore 11.9 with his fee (by real shares count on ethermine.org pool), but show more hashrate 
I didn't test new version of ethermine, but i think hasn't been improved.

Very thanks for new undervilting, all work fine on RX 588 SE!

For fine tuning i'm using polaris bios standart voltages: 800 812 825 837 850 862 875 887 900 912 925 937 950
And lock PowerStage to 0

Tytanick, please, add to dashboard advanced options for direct voltage control (two fields with parameters for --vddci and --mvdd options from wolfamdctrl) or only one checkbox that set this options equvivalent to current gpu voltage. IMHO, this help many users with "nonundervoltaged" Polaris GPU.

Yes, you should login to rig using ssh client, like PuTTY (imho best client). Use user: "miner" password: "miner".
Start file manager - "mc" (Midnight Commander).
Select dir /home/miner, highlight "amd.txt" and press F4 (edit file command)
Modify this file and save it (F2).
Then restart miner or reboot rig.
Any hints: gain root(full) access - execute "sudo su"
Reboot rig (only after gain root priv) "reboot now"
View current miner screen "screen -x miner" (only under user miner, not root)
To access to rigs from another network use computer with TeamViewer and ssh-client or configure router for port forward (dnat or netmap, dont forget lock ip's for rigs in dhcp server), last operations is potencially dangerous, but normal for short time use.

My fault, xmrig != xmrstak 
xmrstak don't have command line options, but you can specify --config /home/miner/config.txt and --amd /home/miner/amd.txt and edit this files after first start miner in autoconf mode.
In amd.txt double all indexes like this:
"gpu_threads_conf" : [
{ "index" : 0, "intensity" : 1152, "worksize" : 8, "affine_to_cpu" : false, "strided_index" : true, "mem_chunk" : 2, "comp_mode" : true },
{ "index" : 0, "intensity" : 1152, "worksize" : 8, "affine_to_cpu" : false, "strided_index" : true, "mem_chunk" : 2, "comp_mode" : true },

Hmm, possibly hang Tytanic's stat-script while parsing miner output. Miner clymore 11.6? (currently i use 11.7 and this issue not appear 24+ hours, but this not clear - i upgrade all rigs from usbsticks to ssd with flashed last release candidate version of smos)

rx580 8gb cryptonight7 xmr

Hmm, xmr-stak can use 2 threads, try this config for 6xGPU:
--opencl-platform=0 --opencl-devices=0,0,1,1,2,2,3,3,4,4,5,5 --opencl-launch=1152x8,1152x8,1152x8,1152x8,1152x8,1152x8,1152x8,1152x8,1152x8,1152x8,1152x8,1152x8 -o stratum+tcp://xmr-eu2.nanopool.org:14444 -u wallet_addr.$rigName/your@email.com -p x --donate-level=1 --print-time=5

Yes, you are not alone 
Your 'stealth' rig contain nvidia cards and mining eth?

See https://bitcointalk.org/index.php?topic=1541084.msg34169220#msg34169220

Yes, it's highly possible 
I make copy of this file and check changes after all updates.

1. Move config.txt and nvidia.txt files to persistent storage (e.g. /home/miner) from /var/tmp/miner/xmr-stak-xxxxxxx/
2. Change verbose level from 3 to 4 in config.txt - " "verbose_level" : 4, "
3. Edit your group in smos web interface - add "--config /home/miner/config.txt --nvidia /home/miner/nvidia.txt" to miner parameters
4. Reboot rig

Do you use this options?
"--noCPU --noAMD --currency monero7"

Don't remeber add to /etc/security/limits.conf
"* soft memlock 262144"
"* hard memlock 262144"

And use
"sudo su"
"mc -c"
after login to rig.
It's more simply, imho.

Hint for enable correct undervolting in smos:

Use ssh client to access farm.
Login as miner/miner.
sudo su
mc -c
Find and edit string #95 in /root/utils/oc_dmp2.sh - remove REM symbol (#) before "--memstate" and add "--vddci 900" before "--mvdd 950"
Use your own voltages if presented not adequate for your card (i use --vddci 850 --mvdd 900 for Sapphire rx588 SE with custom bios).
reboot

PS: sorry for my English