<darosior>
_aj_: i don't know if it's worth the hassle of setting up pool payouts for a testnet. One easy way to achieve that would be to run ckpool in solo mode, this way miners subscribe to the pool with a Bitcoin address as their worker name (in clearnet, yolo SV1)
<darosior>
cleartext*
<darosior>
Or i just keep it all on tb1q3wds3mwyftsnszlxhru79mcq6d9fxy7uwt8p2q and redistribute that 🤷
bitcoindev1337 has quit [Ping timeout: 260 seconds]
memset has quit [Remote host closed the connection]
memset has joined #bitcoin-core-dev
mudsip has joined #bitcoin-core-dev
mudsip has quit [Client Quit]
satsfy_ has quit [Ping timeout: 264 seconds]
satsfy has joined #bitcoin-core-dev
jonatack has joined #bitcoin-core-dev
l0rinc has quit [Quit: l0rinc]
jonatack has quit [Ping timeout: 276 seconds]
satsfy_ has joined #bitcoin-core-dev
satsfy has quit [Ping timeout: 252 seconds]
satsfy_ is now known as satsfy
satsfy has quit [Read error: Connection reset by peer]
satsfy has joined #bitcoin-core-dev
<_aj_>
darosior: 0% fee solo-mining would be fine; would be nice to avoid a premine by the pool that's annointed to generate the genesis block, even if it doesn't really matter.
satsfy has quit [Read error: Connection reset by peer]
satsfy has joined #bitcoin-core-dev
satsfy has quit [Read error: Connection reset by peer]
satsfy has joined #bitcoin-core-dev
<darosior>
Sure, ok
<darosior>
Restarted the pool in solomining mode without any donation set, which should make it 0-fee. Workers now need to register with a testnet address to be accepted
<bitcoindev1337>
has anybody tried prefetching the blocks for ibd similarly to the prevoutfetch thread?
<bitcoindev1337>
i tried this (with ai stuff) and it seems to be a significant boost to ibd performance
enochazariah has quit [Ping timeout: 241 seconds]
jonatack has joined #bitcoin-core-dev
<bitcoindev1337>
and i don't see how this could cause a consensus fail, if the prefetcher doesn't have the block it falls back to the sequential logic
enochazariah has joined #bitcoin-core-dev
satsfy has quit [Ping timeout: 244 seconds]
satsfy_ has joined #bitcoin-core-dev
<darosior>
Something like this should come from running net_processing and validation in different threads
<sedited>
bitcoindev1337 I think a similar approach was discussed at some point. What order of improvement are you seeing? It might make sense to always have the next block to be processed already in memory, I don't think you need an entire thread pool for that.
__nick__ has quit [Ping timeout: 245 seconds]
<bitcoindev1337>
darosior, sure but that's an extremely difficult thing to do and this is pretty trivial it seems
<bitcoindev1337>
sedited, 35%
<bitcoindev1337>
i did this both with no threads and a thread pool and with/without doing the context free CheckBlock stuff
<bitcoindev1337>
doing the CheckBlock was an extra 2% and makes the consensus logic less obviously fine so not worth
<bitcoindev1337>
the difference between one thread and two wasn't huge but was there
<bitcoindev1337>
darosior, also this is totally useless for blocks that aren't already on disk, so steady state receiving blocks from the network dont benefit from this at all
memset has quit [Remote host closed the connection]
memset has joined #bitcoin-core-dev
<sedited>
bitcoindev1337 looks interesting. Maybe share a branch here, or polish a simple version that yields the majority of the performance benefit to a point where you can open a draft pull request.
<darosior>
bitcoindev1337: steady state does benefit from a validation / net_processing split. In fact it's in my eyes the biggest benefit. IBD time reduction is cool, but it's less clear to me how marginal reductions benefit actual users
memset has quit [Remote host closed the connection]
memset has joined #bitcoin-core-dev
memset has quit [Remote host closed the connection]
memset has joined #bitcoin-core-dev
l0rinc has joined #bitcoin-core-dev
Cory has joined #bitcoin-core-dev
<bitcoindev1337>
darosior, how could the split help steady state, you receive a block from the network and then have a series of steps to perform that cannot be done in parallel
<bitcoindev1337>
so unless your net processing is falling far behind, it's not going to improve latency
afiore has quit [Remote host closed the connection]
afiore has joined #bitcoin-core-dev
l0rinc has quit [Quit: l0rinc]
<darosior>
bitcoindev1337: avoid stalling p2p operations, like forwarding headers or block content, while performing validation. Also may allow to prioritize some things over other, like headers over unconfirmed transactions for instance.
<bitcoin-git>
[bitcoin] jonatack opened pull request #35951: doc: release note about I2P ElGamal sunset (master...2026-08-i2p-elgamal-sunset) https://github.com/bitcoin/bitcoin/pull/35951
<bitcoindev1337>
darosior, iirc cant do that because the protocol is sequential, cant skip messages from a peer to process later messages from that same peer
<darosior>
bitcoindev1337: you may have 200 different peers, and currently one of them making you do work stalls the 199 others
jonatack has quit [Ping timeout: 276 seconds]
afiore has quit [Remote host closed the connection]
afiore has quit [Remote host closed the connection]
afiore has joined #bitcoin-core-dev
<bitcoindev1337>
darosior, aren't they handled round robin?
<bitcoindev1337>
so the most a peer can stall you is one message
conman has quit [Quit: Konversation terminated!]
conman has joined #bitcoin-core-dev
ghost43 has quit [Remote host closed the connection]
ghost43 has joined #bitcoin-core-dev
Guest9897 has joined #bitcoin-core-dev
bitcoindev1337 has quit [Killed (NickServ (GHOST command used by Guest9897))]
Guest9897 is now known as bitcoindev1337
<_aj_>
bitcoindev1337: one message plus however long it takes to validate the message; hence the desire to move the validation step to a separate thread, that only pauses peers that care about that validation
<bitcoindev1337>
sure but that sounds like a lot more work :)
<bitcoindev1337>
_aj_, ^
<bitcoin-git>
[bitcoin] l0rinc opened pull request #35952: kernel: prevent dangling iterators from temporary ranges (master...l0rinc/kernel-range-iterator-lifetime) https://github.com/bitcoin/bitcoin/pull/35952
<bitcoindev1337>
l0rinc, interested what you think about prefetching blocks for ibd/reindex