Quote from: Buffer Overflow on July 11, 2013, 09:36:06 AM

Quote from: 8bitPunk on July 11, 2013, 09:33:34 AM

Quote from: Buffer Overflow on July 11, 2013, 09:16:19 AM

Quote from: bonez565 on July 11, 2013, 09:12:33 AM

There is no real hashrate, because the proof-of-work is not based on hashes. The best equivalent is the "primespersec" line. ~200 seems to be a level for a desktop processor.

I'm pushing out 5500 pps, but no block for 2 hours now.

With the block spacing ~12-15 seconds and the default gensieveroundlimitms set to 1 second, each miner has very little time to find a chain. Consider that: some time can be lost due to peer lag in being notified of a new block; a small amount of time is lost validating the pseudoprimality of the previous block; then the remaining time is usually lost cycling through thousands of unsuccessful sieves.

All the miners have the equal disadvantage, so this is irreverent, unless someone is using a private optimized version not released.

I also run my primecoin daemons with the datadir in ram memory to reduce orphans. Every millisecond counts to get a solved block out to the network, don't want any messing about with slow hard drives.

I'm getting a lot of orphans (~1 per hour) so I've been comparing debug.log with blockchain records based on UTC time... In most cases I'm ~1 second late so peer lag is to blame. But a few cases have been within the second, so you're absolutely right that efficiency getting solved blocks out is key.