Just gave a chance to PM2.8c today on Windows 10. Sharing my first expression, observations, problems and results:
Copied the Claymore config file over. It fails out of the box. Requires a migration of configuration.
After a bit of tweaking a successful configuration file migration was in place. Got the PM2.8c running in less than 3 minutes.
The following discrepancies were observed:
PM2.8c vs CL11.5
1. config file with spaces listing multiple single option values, for example: "-cvddc 835, 835, 850, 900, 835, 835" reports an issue with invalid option 835. CL allows spaces. Some use spaces with a line above as comment with list of GPUs; GPU1 GPU2 GPU3 ... - spaces are for alignment.
2. Order of GPUs is different, this leads to incompatibility with CL existing configuration. CL lists AMD cards then adds NVidia cards. On PH2.8c all are mixed as discovered at system level The migration requires redo of the config file and custom settings for all options with individual values per GPU. For example: -cclock 0,0,0,0,1170,0,0,1160,1160,1160,1180,0,1120 - the 0 for NVIDIA cards. Migrated from -cclock 1170,1160,1160,1160,1180,1120.
3. -ethi default is different 8 vs 12.
Results/comparison as reported by miners (13 GPUs rig: 5x AMD RX580, 7x NVidia 1070):
CL11.5: Average speed (as observed in logs): 400.309 - ranging: 400.151 MH/s - 400.706 MH/s @ wall wattage: 1620W
PH2.8c: Average speed (5 min): 402.817 MH/s - ranging: 400.926 MH/s - 404.945 MH/s @ wall wattage: 1630W
Observations:
1. Can confirm increased hash rate of 0.62%. (Same settings as CL).
2. The reporting API works as advertised out of the box. Using Claymore's Monitor on Android it just works, reporting the version as: PM2.8c-ETH compared to 11.5-ETH (Claymore).
3. It appears stale shares value is at a similar level to CL. The PH reported % of stale shares is not in sync with pool report (ethermine.org): 1.26% vs 3%
4. Log file entries are detailed and easy to understand.
6. The on-screen run log/report is on a higher level compared to CL. Way more informative and rich with data. Not sure about the (!) next to Share actual difficulty when diff is in GH.
7. As a software engineer I do start counting my GPUs with 0, 1 ... but on the other hand some other SWs such as HWinfo64 are also listing GPUs from 1, 2, 3... Not sure yet how annoying if at all this is.
Will keep this rig running for a while to gather more data on stability and performance. Will report back.
Would be nice for miners to add a migration section (from other miner apps) to PH.
Definitely would be interested to buy a licensed version with devfee 0.
All in all a great alternative miner with better performance and a lower devfee. A solid ****. Excellent job.