New Version 0.5.7
reduce cpu load
minor performance improvements
con: use single pool connection
con: ssl: clear session data before reconnect
nvml: handle invalid values
This release contains quite some technical work/optimizations, which results most notably in a lower cpu load - about 40% - 30%. It also contains minor solver performance improvements - most notable on an 1060 - about 0.5%, pretty small - however it's still there. ZM uses now a single connection to the pool, since some pools/routers seemed to have difficulties with multiple connections - this might help people who had regular disconnects. Pools which adjust the difficulty based on the share rate won't be able to distinguish between the GPUs because zm uses a single connections now - so the difficulty and thus the share rate might change.
People who requested failover pool support: I'm planning to support it in the next release, I'm releasing this iteration first to make sure the transition to a single connection works as expected.
Thank you for an awesome update, the overall cpu usage on my 8 GPU rig dropped from 45-50% to 25-30% which is quite significant improvement.
I can also see the 0.5% improvement on the 1060 cards, great job!