Hi folks,
sorry for being so rare, but i have to much stuff going around currently and in the next weeks.
Cool, the more nodes the better!
Do you guys think I should put Neisklar's node as bootstrap addres? Maybe it is wiser to have a number of separate p2pools so that users get fewer but higher payouts, minimizing tx fees.
Just so you know. I blocked the peer port on my pool, you can't use it as bootstrap.
At first that was during the setup, testing and so on, to be just on a save side, but later on i'm in the opinion that it's better to have more seperate p2pools.
Why?
1) If we have one big p2pool network, or have many "isolated" p2pools nodes, that is the same regarding decentralization.
The important part is (although fewer fee's for me ;-) but that's no problem, i did that mainly for fun and providing some push for the coin) to have pools operated by different people.
2) The wallet hase problems with large transactions. Large doesn't mean high amount of QRK, it means high amount of inputs.
If we have one big p2pool network the block reward is distributed about all nodes, means even more smaller payouts to slower miners.
Then when sending a amount, the wallet first tries to assemble that amount of many smaller amounts (the inputs) you received earlier, this resulting in large transactions.
If i'm wrong with my opinions, please feel free to show me my errors in my thinkings.
BTW if you have high hashing power you shouldn't use a pool. You only make it more difficult for slower miners.
Just take a litecoin calculator, take the QRK-diff DIVIDE it by 256 and use that diff in the calculator. If you see that you will get a block in 2 or 3 hours, thats ok, don't use a pool. At least in my opinion.
I'm to lazy to do proper quotations.
Next topic is miner optimizing:
Please have a look at this site:
http://bench.cr.yp.to/primitives-sha3.htmlhttp://bench.cr.yp.to/primitives-hash.htmlThere you will find DETAILLED comparision AND OPTIMIZED code of the used hashing functions for different platforms, cpu instructions sets and so on
I'm sorry to say, but currently i don't have the time to work on a quicker miner, and some people already did some really good optimizations. My last tries also produced mostly errors so i'm in the hope the cpu miner for quark is a good starting point so that more experience people when it comes to optimizing, and assembler stuff can produce some really great miners.
(Personally I would like to see one W64 with sse4 and AES support;-) )
Here is also some halfway good list of supported instructions sets by different cpu's
http://gcc.gnu.org/onlinedocs/gcc/i386-and-x86_002d64-Options.html