Reading:
https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2017-April/013996.htmlIt states:
To find multiple roots with the same trailing 32-bits the miner can
use efficient collision finding mechanism which will find a match
with as little as 2^16 candidate roots expected, 2^24 operations to
find a 4-way hit, though low memory approaches require more
computation.
So does anyone know if anyone has done the statistical analysis on the number of transactions in blocks to see if there is an above expected number of blocks with 2x a prime number of transactions?
Also for this to work with publicly available kit when the miner connected to the pool it would be very obvious from the work it was sent to processes. Just a wireshark on the packets would tell you straight away.
Also I am still not clear on the maths of this. Take an antminer S9 @ 11.5TH/s. This will grind through the nonce 2600 times a second! So even if you allow +/-5 on the timestamp that still means 260 collisions you have to find a second and send to the ASIC - so to feed it 4 way hits you need 4Ghash dedicated to that. Which seems hard as this assumes high memory - so a GPU - but 4Ghash is more early ASIC speed.... So really not sure how it all adds up. Unless there is an ASIC for the collision search too.