17:51:29 <***> Buffer Playback...
17:51:29 <Geoge> [21:09:52]  Hi everyone I have been working on a fork of Monero that implement a new POW system called NPOW which is detailed in this PDF; https://gofile.io/d/Bpiwog
17:51:29 <Geoge> [21:10:03] A github is now live with a theoretically working fork of Monero which implements NPOW and a python miner that uses Tensorflow Keras.
17:51:29 <Geoge> [21:10:19] Would anyone be interested in getting involved with the project? It's quite a simple modification to Monero, although could still do with the audit of an experienced Monero developer, but the interesting part is building the mining software.
17:51:29 <Geoge> [21:10:27] ^ I was advised to redirect this here
17:51:29 <UkoeHB> [23:05:11] Geoge: one impression I had was you'd need to store all block templates from all nodes indefinitely, for re-validating the chain; this seems impractical
17:51:29 <gingeropolous> [23:41:41] wheres the github?
17:51:29 <Geoge> [02:04:39] @UkoeHB: you only need the block hash and the pow weights to be stored the in regular block which would be stored in the block chain as usual, the nonce is replaced by a weights array.
17:51:29 <Geoge> [02:06:00] @gingeropolous: https://github.com/monu-fork
17:51:29 <Geoge> [02:06:36] you can check the commits or I have a list of modified files in the monero readme
17:51:29 <Geoge> [02:07:21] I have left the nonce in the block_header just to make the edit easier for now, although the nonce is no longer used the nproof[] is
17:51:29 <Geoge> [02:08:19] Also updated the white paper to make it a little clearer on the streak mining topic; https://raw.githubusercontent.com/Monu-Fork/NPOW/main/npow_whitepaper.pdf
17:51:29 <Geoge> [02:10:03] at the moment it uses float32 weights at a size of ~33kb per header, but this can be reduced in the future to float16 or maybe even int8.
17:51:29 <***> Playback Complete.
19:45:20 <Ge0ge> woops just realised i had marged my changes into the wrong release of monero hence monu fork was not compiling - this is fixed now
19:45:34 <Ge0ge> i mad merged to master and not 0.17
20:08:52 <Ge0ge> I have basically circumvented the nonce but let the nonce there so that I dont have to complete remove ever reference to it in the source code - the intention here is to make the edit as simple as possible to merge into future version of monero
20:10:34 <Ge0ge> the main change was to handle_block_to_mainchain in blockchain.cpp, but by no longer utilizing the nonce do you think there might be complications in other parts of the code I have not considered? I still use the exact same difficulty variable and system