tournament poll: setLocalSpeed(0) ?
5 views
Skip to first unread message
foofoo
Aug 10, 2011, 12:45:43 AM8/10/11
to SCAI2011, ric...@jfli.itc.u-tokyo.ac.jp, alecde...@gmail.com, gabriel....@gmail.com, johan.h...@gmail.com, the....@gmail.com, bgw...@gmail.com, itays...@gmail.com, dougla...@gmail.com, alberto.ur...@gmail.com, jxy...@cs.bham.ac.uk, simon.ch...@usherbrooke.ca, dave.ch...@gmail.com, apr...@berkeley.edu, ta...@usc.edu, bia...@students.wwu.edu, euj...@gmail.com
Dave Churchill wrote:
We have run some tests with our own bot on our network, as well as the
first 4 submitted bots (which came in early). We did these tests at
setLocalSpeed(0) in order to test any crashing / gameplay issues. We
noticed that there doesn't seem to be any noticeable negative effects
from running bots at this speed (so far).
My question is then the following: Would the participants of the
tournament be comfortable with running the tournament with
setLocalSpeed(0) in order to play more games? We think we can run
approximately 15-20 times more games if we play at this speed, and
with the increase in allowed computation time (to account for network
latency) from 42ms to 55ms, we feel that any issues caused by the
faster speed will be addressed.
The real up-side of playing at this speed is that more games means
less variance in the results. Personally, as a participant I would
rather see 20x more games being run as setLocalSpeed(0) than at the
'fastest' speed, just in case my bot did something stupid in one game
and ended up losing because of it.
However, since we announced that we would run games at 'fastest', we
cannot change this rule without unanimous consent from the
participants.
Participants, please respond to this email to cda...@cs.ualberta.ca
(subject SPEEDVOTE) whether or not you would like to run the games at
setLocalSpeed(0). Your votes will be kept secret, and if
any say no, we will not do it.
Thanks.
/Michael
We have run some tests with our own bot on our network, as well as the
first 4 submitted bots (which came in early). We did these tests at
setLocalSpeed(0) in order to test any crashing / gameplay issues. We
noticed that there doesn't seem to be any noticeable negative effects
from running bots at this speed (so far).
My question is then the following: Would the participants of the
tournament be comfortable with running the tournament with
setLocalSpeed(0) in order to play more games? We think we can run
approximately 15-20 times more games if we play at this speed, and
with the increase in allowed computation time (to account for network
latency) from 42ms to 55ms, we feel that any issues caused by the
faster speed will be addressed.
The real up-side of playing at this speed is that more games means
less variance in the results. Personally, as a participant I would
rather see 20x more games being run as setLocalSpeed(0) than at the
'fastest' speed, just in case my bot did something stupid in one game
and ended up losing because of it.
However, since we announced that we would run games at 'fastest', we
cannot change this rule without unanimous consent from the
participants.
Participants, please respond to this email to cda...@cs.ualberta.ca
(subject SPEEDVOTE) whether or not you would like to run the games at
setLocalSpeed(0). Your votes will be kept secret, and if
any say no, we will not do it.
Thanks.
/Michael
foofoo
Aug 10, 2011, 1:00:54 AM8/10/11
to SCAI2011
We received one NO vote already. So the tournament will be played at
fastest setting
(24fps).
/Michael
fastest setting
(24fps).
/Michael
Reply all
Reply to author
Forward
0 new messages