So how big should a plot size be to generate 8 hour deadlines? 1PB? 2PB? 10PB? does the bus connectivity from SATA/USB and plot size read speed per round has anything to do with achieving better deadlines? I know it helps to rank better at your pool over slow miners, but does it help with your own overall resulting deadlines?
The other interesting discovery is while my plot is read 140TB @ 25/s I'm looking at the pool.poolofd32th.club screen in real-time and I see people finishing submitting their deadlines for the same round in like 1 or 2 seconds, how is this possible???? I wonder if is WAN network latency involved here, and some of this miners are very close to the pool server, say ASIA, and I'm very far AMERICA?
As an experiment, I'm looking to convert my miner to Ubuntu and use GPU CreepMiner written in C++, in theory Linux can proceses threads faster than Windows and C++ is Faster than Java (jMiner)
Change your miner setting to report all deadlines found. You'll probably notice you have dozens and dozens of deadlines that vary between days to months. If you mean, "how big should a plot size be to generate at least 8 hour deadlines"...well, I'm not even sure how to go about trying to calculate that. I mean, deadlines found on a given plot vary by block. I'm not sure there's even a way to really guarantee such a scenario of happening, regardless of plot size (well, I guess unless someone owned 100% of the network).
For submission, deadlines should be submitted to the pool the instance they are found (+latency), so its completely normal to see miners submitting deadlines a split second after a new block. Yours deadlines should be submitting in the same manner, even if it takes your miner 25 seconds to read your whole plot.