I set up bitcoin on a new (relatively fast) computer, connected by gigabit ethernet to another running machine, and -connect= to send it directly to that node.
It still took a really long time to bring the blocks over. It would pause for a long time at 501, 1001, etc. Every 500 blocks would have a long pause before starting again.
I'm just questioning whether it's all disk and CPU bandwidth constraints, or if there's something in the system that injects pauses or timeouts.
I had a server that was about 5 years old, but had one of those 10,000RPM SCSI arrays in it. CPU wise, it was about close to a PIV @ 2GHz (single core) with low FSB speed, etc. But when it started to download blocks, it was getting them faster than any PC I had seen before. The disk array was lit up like a Christmas display and it finished off 65k blocks in under a minute. So I knew then it was not so much the CPU or RAM, but the disk speed (or caching perhaps?) that seems to make the difference in that "first time" block download.