@tutulfo: About the new feature you are developing.
I think that it will be great if MM can check and apply the best settings for a card. Currently i mine with less power and some overclocked settings (1080 Card), however some algorithms/coins can handle this better as others. Should be nice if MM can handle this fully automated, for me it is difficult to determine the best settings for all coins.
Since everyone's system is different, I don't see a way he would be able to implement that. What you can do is tweak each algo to achieve what you want (highest hash, most efficient, or some balance), save each as an afterburner profile, and then modify the json for each miner (if you wanted it by algo instead of miner, you would need to create a "custom" miner for each algo). You will put something like:
"C:\Program Files (x86)\MSI Afterburner\MSIAfterburner.exe" -profile1
into the "PrelaunchCommand":"", line, so it would look like this:
"PrelaunchCommand":"C:\Program Files (x86)\MSI Afterburner\MSIAfterburner.exe -profile1",
Disclaimer: I don't actually use this feature, so I may be a little off base on the format of the command. But it should get you started at least.
Edit to add what I meant by adding a custom miner-
For example, if C11 needed profile 1 and lbry needed profile 2, what you would do is create a file named N-ccminerAlexis-c11.json that looks like this:
{
"Path" : ".\\Bin\\NVIDIA-Alexis78\\ccminer-alexis78-ms2013-cuda7.5\\ccminer.exe",
"ExtractionPath" : ".\\Bin\\NVIDIA-Alexis78\\",
"Uri" : "https://github.com/nemosminer/ccminer-Alexis78/releases/download/ccminer-alexis78/ccminer-alexis78-ms2013-cuda7.5.7z",
"Types" : ["NVIDIA"],
"Arguments" : "-o #PROTOCOL#://#SERVER#:#PORT# -u #LOGIN# -p #PASSWORD# -R 10 -r 4 -b #APIPORT# -d #DEVICES# #ALGORITHMPARAMETERS# ",
"API" : "Ccminer",
"Fee" : 0,
"Wrap" : false,
"Dualmining":false,
"APIPort" : 4068,
"PrelaunchCommand":"C:\Program Files (x86)\MSI Afterburner\MSIAfterburner.exe -profile1",
"Algorithms": [
{"c11" : "-a c11 -i 20" },
]
}
Then add another one named N-ccminerAlexis-lbry.json that looks like this:
{
"Path" : ".\\Bin\\NVIDIA-Alexis78\\ccminer-alexis78-ms2013-cuda7.5\\ccminer.exe",
"ExtractionPath" : ".\\Bin\\NVIDIA-Alexis78\\",
"Uri" : "https://github.com/nemosminer/ccminer-Alexis78/releases/download/ccminer-alexis78/ccminer-alexis78-ms2013-cuda7.5.7z",
"Types" : ["NVIDIA"],
"Arguments" : "-o #PROTOCOL#://#SERVER#:#PORT# -u #LOGIN# -p #PASSWORD# -R 10 -r 4 -b #APIPORT# -d #DEVICES# #ALGORITHMPARAMETERS# ",
"API" : "Ccminer",
"Fee" : 0,
"Wrap" : false,
"Dualmining":false,
"APIPort" : 4068,
"PrelaunchCommand":"C:\Program Files (x86)\MSI Afterburner\MSIAfterburner.exe -profile2",
"Algorithms": [
{"lbry" : "-a lbry" },
]
}
And so on. You could also group them by profile (make a N-ccminerAlexis-profile1.json for all the algos that would use profile 1 etc...).
Automatic power limit for nvidia is working in my development enviroment, i am upgrading it, for example , allowing to switch power limit without stop miner, benchmarking only algos reaching first position, etc. I hope it will work for amd also.
How it works?, when an algo reach first position anytime, if this option is active, MM change power limit and make a benchmark for each power limit point to check (f.e. 110, 100,90, 80), recording power and hashrate, from this point each time profit is checked that algo has 4 versions, one for each power limit point where profit is calculated based in its own hashrate, power draw, power cost and revenue.
In my tests, when revenue is high or power cost low, MM set gpu to 110%,if not set to 100% or 90%.
Only one flaw i see, revenue must be adjusted to reality or over-estimations will make power limit always goes to 110%. For that reason i add a profit correction factor for each pool.