Also I guess time management doesn't really support mps time controls currently, as it played too slow
for a long time? (mps tc is usually not available on servers, but it's still used by testers and one of the
main used tcs besides fischer tc)
for being slow mentioned by me and using mps tc
you can see that AXW uses always 3.32-3.33 seconds per move, but as the game starts after 6 plies
from the given start position mith movestogo 37 (40-3moves) it would fail because 37*3.33 > allowed time,
thus it seems to panic and makes moves 39-40 in zero time!
And yes your assumption is right for mps tc there is wtime/btime and movestogo sent!
I wanna just mention that most testers nowadays use start positions/books for testing
as I did above with the 6 plies start pgn, so the first session starts with TC/movestogo-X (here 37).
After the first session is over the engine gets the time added for the next session and repeating.
BTW the opponent Goyaz (xboard) probably has bug, respectively does not count the external moves
which would explain the high time usage for moves 41-43. In xboard the movestogo parameter does
not exist and they need own logic to substrct the given moves.
I selected it because I wasn't sure about current strength of AXW. 1700 (CCRL scale) was too strong for it
and Goyaz is around 1400+ and seems too weak.
Split from #2