I am not a developer and wish I was. I have been mining since 2016. I would have had this miner well tested before deployment. I cannot comprehend to you the number of issues I have run into the latest Phoenix miner, DAG level, and Polaris cards when running large scale GPU farms. I manage and own a few, all running Windows Server with proper data center infrastructure management in place. It's as if Phoenix miner has only focused on development for Vega and Navi GPUs, but forgotten about Polaris as the DAG has limit increased. All of my Navi based rigs are fine for the most part. I manage many. Yes Claymore should be getting just as much backlash for not at all providing support but he is not available for commentary. Phoenix has been here and known about this encroaching issue. Please do something about this. We rely no stability.
I test the new version with -vmr 85
I am at 33.900Mhs
PH: 5.2e
amd driver 20.11.2
If you put rxboost (or vmr) too high you will have memory errors and the real hash will be half of what the miner shows. You have to check always the stats of the pool. It is good to use some memory errors tool or just watch the screen. If you see artifacts (like small blinking dots), you are above the possibility of the memory. For me, I can use like -rxboost 1 with the best hash on the pool without memory errors. It gives 32.4 Mh/s with moded timings (rx 580 8Gb).