would people prefer to not have this chat listed on getmonero?

6:24 < hyc> eh? please don't

16:25 < hyc> it's low noise now, should stay that way

yeah, was asking to make sure it was a shared opinion

ErCiccione[m], i also enjoy the low noise. i think people that want to find it will find a way.

ok. I closed the PR

thanks

I expect people who are interested probably find their way here from -dev

i wonder if randomx verification could be made cheaper by including headers for each loop

so then a hash check could be performed after performing the first loop

Nonce bits heatmap: i.imgur.com/ccbyqey.png

about what you'd expect for nonces counting up from zero

using the high byte for job separation I suppose

high byte is nicehash and xmrig-proxy

so yes, distributing work

bits 14-15 is xmrig's nonce batch size (32768)

looks like bits 15-24 are also distributing work

No, it just progresses from 0 to 16777216 in steps of size 32768

ah

not many CPUs can do 2^24 hashes in ~2 minutes

so bits 23-24 get really cold

makes sense

usually very slow 10-15 minute blocks can have these bits set to 1. Or monerod solo miners - they select start point at random

there's a spike in bits 20-22 around 450 on horizontal axis. This is EPYC servers on Azure when we had that 900 MH/s miner

heh cute