Some benchmarking problem

[Xenooooooooo]

Hello, professor. I've recently built volepsi on my machine. When using ./frontend -perf -psi -nn 20 -v -nt 1 -t 5 for testing, the message below appeared.

t 1 fakeBase 0 n 1048576
coproto::code::remoteClosed: The connection has been closed by the remote party.
coproto::code::badCoprotoMessageHeader: Received a corrupted/invalid protocol message header
coproto::code::remoteClosed: The connection has been closed by the remote party.
coproto::code::badCoprotoMessageHeader: Received a corrupted/invalid protocol message header
coproto::code::remoteClosed: The connection has been closed by the remote party.
coproto::code::badCoprotoMessageHeader: Received a corrupted/invalid protocol message header
coproto::code::remoteClosed: The connection has been closed by the remote party.
coproto::code::badCoprotoMessageHeader: Received a corrupted/invalid protocol message header
Label     Time (ms)  diff (ms)
__________________________________
end           725.1    725.075  **********
begin         729.0      3.881  **
end           932.8    203.815  ********
begin         932.8      0.006  
end          1116.3    183.533  ********
begin        1116.3      0.006  
end          1300.7    184.335  ********
begin        1300.7      0.006  
end          1481.3    180.622  ********

23693261 10690853

There're two questions:

1. Why there're remoteClosed and badCoprotoMessageHeader message taking place?
2. The 2nd to 4th rounds are much faster than the first round. It is because these failures? (I mean, the remoteClosed and badCoprotoMessageHeader errors.)

[ladnir]

yes, you should not trust the running times. Its unclear to me what the cause is.

[Xenooooooooo]

Right, I've got that. Thanks professor.