komodo verbosity/houdini recapturing

Discussion of chess software programming and technical issues.

Moderator: Ras

yanquis1972
Posts: 1766
Joined: Wed Jun 03, 2009 12:14 am

komodo verbosity/houdini recapturing

Post by yanquis1972 »

this is hardly technical, but its quite trivial & not very intelligent, so i figured i'd stash it here?

i'm running an engine-engine match in the fritz GUI w/ a tweaked houdini set as the default engine & komodo as the kibitzer, & am manually entering komodo's moves. (i know, its a bit weird, but these are classical TC & for enjoyment, & i am trying to give the games some more kick & komodo some handicaps in terms of opening steering (so far the deepest is 9 moves) to quiet lines, settings, & clock, & i find it keeps me much more involved in the game. )

2 things have struck me, one i would consider insignificant, but possibly worth money; the other more significant but almost def worthless monetarily.

first & most important to me, re: komodo -- houdini just seems a lot more talkative than komodo; granted a)this is houdini w/ tactical mode on, which probably makes some difference, maybe a good deal). i'll frequently see H3s tasks while komodo appears to just sit there, & updates come more frequently, etc. now komodo's about as rock solid as it gets as far as engines go, so this is hardly comlpaining, but i was wondering if its feasible that komodo might chirp a bit more in the future. its also possibile im just taking a small sample

and the irrelevant to me part (& robert's most likely already rolled his eyes & avoided thread after seeing the title) is that houdini's TM defnitely doesn't -seem- ideal so far...granted, i'm only beginning my third game, & have been paying attn probably less than half the time. now i think integrated TM is pretty much bogus & there should be some kind of uniform UCI TM for ratings clubs. but you certainly notice wasted clock when playing against an engine at LTC. not taking only-move recaptures, little time spent in the opening (& i'm not sure if this is even bad or not; just that it's not how humans tend to allocate time, & again maajor sample size caveat). if an author is going to slack or ignore any part of his engine, this is the one i'd most like to see him ignore, so if houdini isn't great, sincere good on robert. but i figure it's prob one of the more irresistable areas to tune, since the elo seem all magic & free & whatnot.

and are totally worthless to the buyer, i might add, so its a bit of a scam for you guys to be wasting yr time w/ that...get that kink in the new bishop pair evaluation heuristic worked out, or whatever you guys would say. ;)
User avatar
Don
Posts: 5106
Joined: Tue Apr 29, 2008 4:27 pm

Re: komodo verbosity/houdini recapturing

Post by Don »

yanquis1972 wrote:this is hardly technical, but its quite trivial & not very intelligent, so i figured i'd stash it here?

i'm running an engine-engine match in the fritz GUI w/ a tweaked houdini set as the default engine & komodo as the kibitzer, & am manually entering komodo's moves. (i know, its a bit weird, but these are classical TC & for enjoyment, & i am trying to give the games some more kick & komodo some handicaps in terms of opening steering (so far the deepest is 9 moves) to quiet lines, settings, & clock, & i find it keeps me much more involved in the game. )

2 things have struck me, one i would consider insignificant, but possibly worth money; the other more significant but almost def worthless monetarily.

first & most important to me, re: komodo -- houdini just seems a lot more talkative than komodo; granted a)this is houdini w/ tactical mode on, which probably makes some difference, maybe a good deal). i'll frequently see H3s tasks while komodo appears to just sit there, & updates come more frequently, etc. now komodo's about as rock solid as it gets as far as engines go, so this is hardly comlpaining, but i was wondering if its feasible that komodo might chirp a bit more in the future. its also possibile im just taking a small sample

and the irrelevant to me part (& robert's most likely already rolled his eyes & avoided thread after seeing the title) is that houdini's TM defnitely doesn't -seem- ideal so far...granted, i'm only beginning my third game, & have been paying attn probably less than half the time. now i think integrated TM is pretty much bogus & there should be some kind of uniform UCI TM for ratings clubs. but you certainly notice wasted clock when playing against an engine at LTC. not taking only-move recaptures, little time spent in the opening (& i'm not sure if this is even bad or not; just that it's not how humans tend to allocate time, & again maajor sample size caveat). if an author is going to slack or ignore any part of his engine, this is the one i'd most like to see him ignore, so if houdini isn't great, sincere good on robert. but i figure it's prob one of the more irresistable areas to tune, since the elo seem all magic & free & whatnot.

and are totally worthless to the buyer, i might add, so its a bit of a scam for you guys to be wasting yr time w/ that...get that kink in the new bishop pair evaluation heuristic worked out, or whatever you guys would say. ;)
I don't understand the terminology "verbosity", are you saying that if Komodo outputs more data it will play better? What is "verbosity?"
Capital punishment would be more effective as a preventive measure if it were administered prior to the crime.