Derek Zimmer points out the main issues found when auditing RandomX.
There was a way to remotely crash any Monero node. So someone could remotely crash, say, all the mining pools and then execute a 51% attack on the network.
– at 12m
It's definitely, definitely going to be very hard to try to reverse engineer something into hardware that's not going to be either obscenely expensive and complex or use so much energy that it's not worth mining with it.
– at 16m25s
The main complaint that we've had from all of the audits so far is that the code is very monolithic.
– at 18m54s
So you can test things by putting numbers in one end of the application and seeing what what comes out the other end of the application, but you can't break it down into pieces.
– at 19m03s