Status as of Monday, March 30
Posted: Mon Mar 30, 2015 9:49 am
Just a few updates for today:
- There will be downtime this Wednesday to release the new database changes for work restart adjustments. The reasons for the downtime are discussed in the "News" forum.
- We discovered that work restart times are also affected by our pool's effect on coins' block times. We will likely be mining many more coins simultaneously in the future. Right now, we assign miners to a coin until the block time is 3s. In the future, we will assign miners to a coin until the block time is low enough that work restarts make the coin unprofitable for every other miner in the pool. This is a surprising finding.
- Even though this release will significantly improve the system for people who have to deal with long work restart times, there is still one more optimization that can be made. In our testing, we discovered that we can also compute how many merge-mined coins it makes sense to send to a user based upon his work restart time.
- In an ideal world, miners with zero work restart times would receive constant block updates every time one of the merge mined coins gets a new block. A miner with a small restart delay may find that not only does it make sense not to mine the most profitable coin, but it also is better to only send one coin to be merge mined. A miner with a huge restart delay may find that more money is lost due to merge-mining restarts than is gained, so he should accept stale shares for the merge mined block and only get the latest merge mined blocks when a new primary coin block appears.
- We fixed a bug where variable difficulty was being computed across an entire account, not just for one miner.