yanquis1972 wrote:much to my suprise, when i had to reinstall windows & went to download stockfish 3, there was a button to access the development versions, so it's now open to all if it wasn't before. more kudos to the stockfish team for that.
I've been using this link for quite some time now.... didn't know that others here didn't know about it !?
Word of caution/advice.... the latest version available is not necessarily the strongest ! Some of the updates actually seem to be regressive.
For instance, I and a few others have found that May versions of Stockfish are much stronger than the June versions !
This, of course, is subjective, and may change any day.
yanquis1972 wrote:excellent, carl - can't wait to try it out!
i finally got the opportunity to actually observe a full game after starting a tournament b/w houdini 3 & komodo just before i went to bed. houdini had just reeled off 5 wins in the past 8 games (with 3 draws), but komodo got the better here, handling the white pieces in a giuoco piano. i think its exemplary in that it reminds me of what i initially loved about komodo: in positions that most engines would likely struggle or find it impossible to make progress, komodo quietly prepares a break & ends up with the victory.
also, time management seems to be intelligent, for those whom that means anything to. komodo thought for around 30 seconds on the breaking move 39. f4!.
both engines are default settings except for 4 threads. houdini has access to, i believe, gaviota tablebases. i'm not sure if the chessbase GUI overrides nalimov for gaviota if a path is given, or how it's decided. i do have a few 6 man nalimov TBs, but the faster access to GTB should hopefully make up for it.
256MB hash, silver opening suite, HT enabled, -results subject to noise- as this is my only computer. also, even though it says 3503 MHz, it's clocked to 4204 MHz. so i assume chessbase is just reading the default clock.
Komodo plays very elegant chess. Sometimes top engines play very strong, but ugly chess. Komodo's play is pleasing on the eyes. Watching its games live is a really nice experience.
Re: HT enabled, this may be debatable, but most experts here recommend not using it.
It's probably better to use many threads as physical cores available. Using HT can actually help in very tactical positions, but will hurt in quiet ones. All things being equal, HT adds a lot of overhead and can degrade performance.
Having said this, I'm not yet fully convinced that *all* engines are hurt by HT.[/quote]
Interesting comment !
Which Engine in your opinion isn't really hurt by HT ?
i remember reading on a couple of occasions from harvey williamson that mark uniake optimized hiarcs for hyperthreading very shortly after the i7 was introduced. whether or not that is still the case i can't be sure. but that's the only engine i know for 100% certain that has ever had official claims to be better w/ all logical threads used.
Having said this, I'm not yet fully convinced that *all* engines are hurt by HT.
Interesting comment !
Which Engine in your opinion isn't really hurt by HT ?
If memory serves me right, I think there may have been comments of this nature re: Deep Hiarcs 14, where the default is to use as many threads as are available in the system. I'm not totally sure if this the correct interpretation of the Hiarcs readme file instructions, but I also recall recently reading this about another engine, in its respective readme text file, where the author(s) mentioned that their engine benefits from hyperthreading. I only wish I remembered the name of that engine. Maybe someone else can.
Maybe some day we may see someone produce some strong evidence that a given engine gains Elo points from HT. Otherwise, all we have is speculation regarding this subject.
Having said this, I'm not yet fully convinced that *all* engines are hurt by HT.
Interesting comment !
Which Engine in your opinion isn't really hurt by HT ?
If memory serves me right, I think there may have been comments of this nature re: Deep Hiarcs 14, where the default is to use as many threads as are available in the system. I'm not totally sure if this the correct interpretation of the Hiarcs readme file instructions, but I also recall recently reading this about another engine, in its respective readme text file, where the author(s) mentioned that their engine benefits from hyperthreading. I only wish I remembered the name of that engine. Maybe someone else can.
Maybe some day we may see someone produce some strong evidence that a given engine gains Elo points from HT. Otherwise, all we have is speculation regarding this subject.
CL
Could it be Stockfish ?
Read in a Blog somewhere that Stockfish doesn't use the parallel alpha-beta Search method as in Houdini ( don't know if its true though).
If its true though, it would mean that the inefficiency that creeps in when HT is enabled could be effectively avoided, if Stockfish is used instead of Houdini !
This would be a real boon to people like me with 6-Core i7s...we would suddenly have TWELVE threads running effectively !
Right now, we have six threads sitting idle doing practically nothing or have disabled HT in BIOS, which is a waste !
yanquis1972 wrote:much to my suprise, when i had to reinstall windows & went to download stockfish 3, there was a button to access the development versions, so it's now open to all if it wasn't before. more kudos to the stockfish team for that.
I've been using this link for quite some time now.... didn't know that others here didn't know about it !?
Word of caution/advice.... the latest version available is not necessarily the strongest ! Some of the updates actually seem to be regressive.
For instance, I and a few others have found that May versions of Stockfish are much stronger than the June versions !
This, of course, is subjective, and may change any day.
The difference between the different versions of stockfish(in may or june) is small and I do not believe that you played enough games to find improvement or regression.
Note that the stockfish team play many thousands of games before they accept a functional change when most changes are rejected.