Hello,
want to make clear regarding algo-switch for NVidia GPU 970s rig. Correct me please when I'm wrong
1) e.g. I want to use the pre-defined algo set "
NVIDIA - Blake-Vanilla, Ethash, Groestl, Myriad-Groestl, NeoScrypt, Quark, Qubit, Skein, X11, X13, X15". I create a HUB-worker with the same job settings (
NVIDIA - Blake-Vanilla, Ethash, Groestl, Myriad-Groestl, NeoScrypt, Quark, Qubit, Skein, X11, X13, X15), set up the bat-file with all these algos (as shown at the sample page) to ports 12xxx and launch it
2) The ccminer shall run through all the starting commands and select THE MOST PROFITABLE ALGO (at the moment), and start mining THE MOST PROFITABLE COIN within this algo -
am I correct here?3) If no AUTOEXCHANGE option is set, then the mined coins shall store in my account, waiting till I ether withdraw it to a definite coin's wallet or convert into some other coin (e.g. BTC).
If I got the work flow right, then
HOW DOES THE POOL EVALUATE THE MOST PROFITABLE ALGO FOR MY GPU (in my case - GTX 970)? Can I affect this setting somehow? Is it done only with the "difficulty" parameter for each algo in ccminer command line? Or in some other way?
If it is "difficulty", then can anybody share the correct values for GTX970 for each of these algos (Blake-Vanilla, Ethash, Groestl, Myriad-Groestl, NeoScrypt, Quark, Qubit, Skein, X11, X13, X15)?
Thanks in advance to everyone who can help!!!
Or please PM me!!!
Hi
Sorry for late reply. I had some problem personally lately.
1) e.g. I want to use the pre-defined algo set "
NVIDIA - Blake-Vanilla, Ethash, Groestl, Myriad-Groestl, NeoScrypt, Quark, Qubit, Skein, X11, X13, X15". I create a HUB-worker with the same job settings (
NVIDIA - Blake-Vanilla, Ethash, Groestl, Myriad-Groestl, NeoScrypt, Quark, Qubit, Skein, X11, X13, X15), set up the bat-file with all these algos (as shown at the sample page) to ports 12xxx and launch it
-> That's right. You're doing correctly.
2) The ccminer shall run through all the starting commands and select THE MOST PROFITABLE ALGO (at the moment), and start mining THE MOST PROFITABLE COIN within this algo -
am I correct here?-> Actually ccminer is not selecting what coin to mine. Pool decides according to hub switch you selected.
If ccminer tries to connect to not most profitable algo(coin) endpoint, then pool just disconnect it. ccminer will try with next algo. Pool will accept connection until ccminer tries with appropriate algo endpoint.
As profitability changes time to time, pool accepts/disconnects ccminer's connection by it's own mechanism.
3) If no AUTOEXCHANGE option is set, then the mined coins shall store in my account, waiting till I ether withdraw it to a definite coin's wallet or convert into some other coin (e.g. BTC).
-> You're right.
If I got the work flow right, then
HOW DOES THE POOL EVALUATE THE MOST PROFITABLE ALGO FOR MY GPU (in my case - GTX 970)? Can I affect this setting somehow? Is it done only with the "difficulty" parameter for each algo in ccminer command line? Or in some other way?
If it is "difficulty", then can anybody share the correct values for GTX970 for each of these algos (Blake-Vanilla, Ethash, Groestl, Myriad-Groestl, NeoScrypt, Quark, Qubit, Skein, X11, X13, X15)?
-> Profitability calculation includes "mathmatical block finding probability (net difficulty of coin)" + "block reward" + "market price" + "monthly average block finding luck from pool" + "GPU hash performance for that algo (Normalization)".
What you asked is mostly about normalization. To compare each algo's profitability for each gpu performance.
Pool don't have optimized set for GTX 970 currently. But pool have hashrate performance data between different algo for AMD and NVIDIA.
Yes, only 2 types are available.
The idea is that NVIDIA gpu works similarly with other NVIDIA gpu, and AMD gpu works similarly with other AMD gpu.
I know that there is difference between each gpu product but it's hard to track and decide from pool side.
Actually optimizing for each GPU product approach is not that efficient also. I'll explain the reason.
Here's internal work of auto switching.
Auto switching only switches when target coin's block is found. So, no switching occurs until block is found. If switching occurs too frequently even block is not found then miner would have to jump all over very inefficiently.
So auto switching needs minimum hashpower to find block and switch. If every GPU uses optimized value for each of them, their hash would be spread for many coins and stuck finding their block.
If you use NVIDIA gpu, then stick to NVIDIA switches. It will work smoothly than expected.
ccminer's difficulty setting is to adjust difficulty multiplication between pool and miner. You would need to specify it with some numbers like Qubit algo, but normally you don't need to.
Thanks.
