hero member
Activity: 504
Merit: 500
Since luck comes in waves, and smaller diffs = faster finds... When one person finds a block, in his "wave of luck", thus finds 2-4 more faster than everyone, (since they built off their block faster than it was broadcast)... In those instances, you will see a lot more rejects with smaller workload sizes.
With larger workload sizes, you are still "looking", while a new block is found, and you don't attempt to submit that work, your miner just throws it away, once it is complete. With smaller workload sizes, you will potentially see more rejects, because you finish faster and submit a fraction of a second faster than those still processing a larger workload size.
This is true for solo and in pools. However, in pools, it should give you, potentially, one to three extra shares submitted, per new-block. (Over workload 256.) But again, you will also see more rejects, as there too, your work was submitted, not just thrown away as you get a "new block detected".
At the end of the day, those with more machines will notice greater advantages, while those with only a smaller single rig, may not notice any difference.
Rejects are not bad, unless they are excessive. (Fast block-time coins naturally have more rejects. BTC, for instance, has few. Due to difficulty, luck and long times between blocks.)
If errors are always excessive, it is because you are connected to a node that is usually too far away. Thus, it is taking them a lot longer to send you the new-block info... or you are connected to a rogue-node set, who is purposely rejecting any "submits" that are not their own. They do that to stall your submitted work, in hope that they find a block, which they confirm before yours. (They may also stall "new block found" data.) Just shut-down your wallet, and restart it. You should connect to faster nodes from the list you built. Otherwise you will remain on those sour nodes, until they boot you.)
Also helps if you choose a pool located in your own country/state/city/network. Hopping country-gates and country-backbones or even network-hubs, causes a lot of buffered lag that is unavoidable.
I love it when I hit like 5 blocks in a row, only seconds apart from one another. After having not found one for an hour. It is funny. As "random" as the results should be, there is an obvious pattern of waves, due to the pseudo-random numbers and the fact that everything is based off prime-numbers for the hashing. (Look at any pool and you see the same patterns... waves of 3 and 4 and 5, all from the same people, enough that you can almost predict who will hit the next wave. Just not so predictable that you could ever tell which block they will hit.)