Trying 212.114.110.107...
Connected to winboard.nl.
Escape character is '^]'.
_ __ __ __
| | / /__ / /________ ____ ___ ___ / /_____
| | /| / / _ \/ / ___/ __ \/ __ `__ \/ _ \ / __/ __ \
| |/ |/ / __/ / /__/ /_/ / / / / / / __/ / /_/ /_/ /
|__/|__/\___/_/\___/\____/_/ /_/ /_/\___/ \__/\____/
^^__ _ _ _
/ - \_ | || || |
<| __< H.G.M's Variant Chess Server |_______|
<| \ \__ ___ /
<| \ |___|_|
<|______\ |_|___|
_|____|_ |___|_|
(________) (_______)
/________\ /_______\
Head admin : admin Complaints to : NOT_CONFIGURED
Server location: NOT_CONFIGURED Server version : 2.2.x-DEVEL
Server name : NOT_CONFIGURED
If you are not a registered player enter the login name "guest"
login: guest
Created temporary login 'GuestKXEJ'
"GuestKXEJ" is not a registered name. You may play unrated games as a guest.
(After logging in, do "help register" for more info on how to register.)
Press return to enter the FICS as "GuestKXEJ":The HD of the old server machine suffered a head crash. As a result all
accounts have been lost, and people should register anew.
__________________________________________________________________________
Do 'help' for a list of the commands that are available.
Do 'help admins' for admin information, "=admin" for a list of all admins.
Do 'help ficsfaq' for answers to frequently asked questions.
__________________________________________________________________________
Newcomers:
type 'help intro0' for a gentle introduction.
type 'help register' to get a registered account.
Foul or abusive named guests may be 'nuked' without warning!!
fics% help register
register
*** ***
*** Registration is now automated using email! ***
*** ***
To get a registered account on this server, send a message to the following
address ...
chess@caissa.onenet.net
Yes, 11 is seconds. It is the time WinBoard/XBoard thought it took the engine to move.
The ICS would only flag you when it receives the move, because you were using timeseal, so it has to examine the time stamp to know for sure you submitted the move too late to timeseal, and that it didn't come in late just because of lag. And the timestamp comes with the move.
hgm wrote:Yes, 11 is seconds. It is the time WinBoard/XBoard thought it took the engine to move.
The ICS would only flag you when it receives the move, because you were using timeseal, so it has to examine the time stamp to know for sure you submitted the move too late to timeseal, and that it didn't come in late just because of lag. And the timestamp comes with the move.
And what was the reason if not lag? Gaviota spent between 0.2-0.3 seconds since it received the move and replied, based on the timestamps I have in the log.
Apparently there was a lag between the engine writing the move into the pipe, and XBoard receiving it from there. The 11 sec is based on XBoard's own time stamps in the low-level read and write routines.
I don't know what conditions you run under (if the engine runs on the same machine as XBoard, how much you lower their priority compared to the GUI, how many cores yo leave idle to handle other tasks than the engine search threads, if there is memory contention that can cause excessive swapping...), so I cannot guess at what could cause such a lag.
hgm wrote:Apparently there was a lag between the engine writing the move into the pipe, and XBoard receiving it from there. The 11 sec is based on XBoard's own time stamps in the low-level read and write routines.
I don't know what conditions you run under (if the engine runs on the same machine as XBoard, how much you lower their priority compared to the GUI, how many cores yo leave idle to handle other tasks than the engine search threads, if there is memory contention that can cause excessive swapping...), so I cannot guess at what could cause such a lag.
I see, timeseal won't protect me against that type of lag.
The only thing I can think of is that there was much output being sent at that time. The engine was reaching very fast 64 plies in a short period of time because it was hitting the 50 move rule (which it would have been triggered in the next move).
hgm wrote:Apparently there was a lag between the engine writing the move into the pipe, and XBoard receiving it from there. The 11 sec is based on XBoard's own time stamps in the low-level read and write routines.
I don't know what conditions you run under (if the engine runs on the same machine as XBoard, how much you lower their priority compared to the GUI, how many cores yo leave idle to handle other tasks than the engine search threads, if there is memory contention that can cause excessive swapping...), so I cannot guess at what could cause such a lag.
I see, timeseal won't protect me against that type of lag.
The only thing I can think of is that there was much output being sent at that time. The engine was reaching very fast 64 plies in a short period of time because it was hitting the 50 move rule (which it would have been triggered in the next move).
I will have to pay close attention.
Miguel
Yes, that must be the problem. Gaviota records the lags in its logs, and I can see a correlation when I analyze all 9 games today. Every time there is a big lag (> 2 secs), the move was searched very deep in a very for a short period (>30 plies in a fraction of a second).