Just for grins (pun intended), I put lubuntu on an old, slow pc to try to get a very low-end data point. The test was not very successful, as the server would always get stuck (stop incrementing blockheight) after an hour or two, and eventually was always stuck. This was running the server to sync without mining. I believe Heung and Igno felt that the peer problems encountered were due to a lack of memory resources. In the end, the server remained stuck for days at a blockheight of 51,384. Earlier in the chain the server would run for awhile and then get stuck after adding exactly 1,532 blocks. During these attempts, the grinexplorer chain height was increasing from roughly 55k to 61k blocks. When the server was actively syncing blocks, syncing was taking place at about 4 times the rate of the (testnet1 small graph) chain height increase–so if it didn’t keep getting stuck, it would have eventually synced. For testnet2, this system would be worthless.
Intel Celeron 2.40GHz x 1
–prior to sync:
---- syncing & mining: sync 235 blk/hr average
---- syncing only: sync 320 blk/hr average
---- syncing & mining: (never completed sync)