Re: PhoenixMiner 2.5d: fastest Ethereum/Ethash miner with lowest devfee (Windows)
@PhoenixMiner - please research a bit the stale problem. With 2.5d i get 3, 4 even 5 times more stale vs. 2.4. 
Anyway, the miner is amazing, getting 6 mhs more from 192 to 198 mhs with 6 rx 580. Good job.
Also, please add to the miner cclock, mclock, cvddc, mvddc and other commands from Claymore, it would be really awesome.
Anyway, the miner is amazing, getting 6 mhs more from 192 to 198 mhs with 6 rx 580. Good job.Also, please add to the miner cclock, mclock, cvddc, mvddc and other commands from Claymore, it would be really awesome.
24hours is way too little to make any measurements, you need 4-5 times that to get a good picture - From my experience phoenixminer, even the 2.5d is giving me more shares per day on ethermine compared to claymore, I actually have a machine with 12xRX580 which consistently underperforms with claymore miner, giving me fewer shares per hour than it should, on phoenix that machine runs smooth as butter.
V2.5d appears to have some stale share issues, it is actually leveling out a bit for me, but are still producing about 50% more stales compared to 2.4 - But 2.5d solves a problem which I have been having on one of my machines so overall there is improvement to track.
While we can't see a significant change in the stale shares on our test rigs, one of the changes that we hastily made in 2.5d in an effort to prevent the ethermine connectivity problems could be causing this for some users. As we now know that that has nothing to do with the connectivity problems, we are rolling back these changes, so 2.6 should fix any unusual issues. The beta of 2.6 is already undergoing internal testing and will be released in less than 24 hours.V2.5d appears to have some stale share issues, it is actually leveling out a bit for me, but are still producing about 50% more stales compared to 2.4 - But 2.5d solves a problem which I have been having on one of my machines so overall there is improvement to track.
The hardware control commands (cclock, mclock, cvddc, mvddc, etc.) are our next priority. Version 2.7 will contain some experimental support but we would need to test a lot on a matrix of GPUs and driver versions before releasing it. Please note that only AMD provides open access to its Overdrive API via the ADL library. While Nvidia has similar capabilities via the NVAPI, the hardware control and overclocking are only available if the developers are pre-approved by them. We will try to get access but it's probably not going to happen. So, at least for now the hardware control capabilities will be limited to AMD cards.
[/quote]
Sounds good, my stale share calculations are based on the reported by the pool, not the reported by the miner which I know is not 100% - Will tinker a bit with -mi later today, but I just swapped the machine that produced most stale shares over the last 6 hours to 2.4 just to confirm that there's indeed "a problem".