Closed hackmod closed 5 years ago
Thanks for the bug report. The miner has only been used for algorithm development up until this point. The fact that "block_number" was actually the epoch number is a known limitation that we were waiting for other infrastructure changes in order to fix, which now seem to be in place.
It looks like it's crashing when compiling the kernel after initializing to a new DAG epoch, which for some reason resets the CUDA context. There's some CUDA sequencing that needs to be figured out around that.
Thanks for the changes. Pull request merged so closing this out.
this is a bug report.
since ProgPoW utilize
block_number
to get theprog_seed = block_number / PROGPOW_PERIOD(50)
, we need to obtainblock_number
from the geth/pool node.now,
eth_getWork()
support block_number info. from the latest geth 1.8.18, so I've made small fix to use block_number tocuda_init()
correctly,but the result is following,
testnet genesis with
progpowBlock: 0
test with https://github.com/ethereum/go-ethereum/pull/17731 +
eth_getWork()
fix https://github.com/ethereum/go-ethereum/pull/18038mininig command :
$ /share/ProgPOW/build/ethminer/ethminer -U --cuda-devices 0 -P http://127.0.0.1:8485
some hunks of quick and dirty patch.