Hardware control options (-tt, -fanmin, -fanmax, -cclock, etc.) are not supported on Linux yet.
Is this -tmax 80 -tstop 85 -tstart 80 didn't work in Linux too? Thanks.
-tstart and -tstop will work as they only require the card temperature to be properly measured and this is supported in Linux. -tmax won't work in Linux.
Green kernels do not work on vega 64 or radeon vii
Are they supposed to?
No, there are no green kernels (or turbo kernels) for Vega and Radeon VII.
Thank you PhoenixMiner for your response.
Yes, it is strange because only 2-3 cards have this issue, go under 32.xxx and on both rigs is the same. If it's how you said then all should have the same speed ... this is the theory

I will try to update also the driver, maybe ... I will change something or I will destroy everything

)
thanks and regards!
Try the new PhoenixMiner 4.2c - the new v3 kernels should decrease the power consumption and if the cards are hitting the power limit, this can increase the hashrate quite a bit. If this doesn't help, try
-nvkernel 2 with and without
-nvdo 1 to see if the hashrate stabilizes.
Hello all.
Rig 5cards : 2 rx480 nitro+(ubermix 3.1) and 3 gtx 1070.
Driver 18.6.1 for rx
416.94 for gtx
Since 19/03 one of my rx480 nitro+ 11260-07-20G getting incorrect shares

What i tried:
Stock clocks and bios
Change risers, now it is in the main MB slot.
-clkernel 0/1/2/3
-clnew 0
-clgreen 0/1
-mi 6-14
And it still getting incorrect shares.
Hwinfo says 0 errros
The second rx480 works fine no incorrect shares.
The miner 4.1c / 4.2a /4.2b doesnt matter.
I think this is new kernels, but don't know for 100%,.
Thanks
This has nothing to do with the kernels as this is happening in only one card. The zero memory errors that you are seeing doesn't necessarily mean that the card is OK. Most probably the card itself is going bad (either the GPU or the VRAM, most probably the VRAM) as the percent of incorrect shares is very high. If the card has 8 GB VRAM, you can try a "workaround", which may work if only one of the memory chips is bad, so there is at least some region in VRAM, which is OK - when you see the incorrect shares, disable only this card from the PhoenixMiner, then run another instance of PhoenixMiner only on this card, which will create another DAG buffer, hopefully in a healthy region of the VRAM.