Gusev wrote:I see Fire on the list of "dead engines" there. Indeed, who would want to test a dead engine? But, as we see, with open source, a good engine can be "undead".
Hi all-
The open-source development of Fire has officially ended, and it is no longer available for download.
Many thanks to any and all who supported me and my efforts for all these years.
Best Regards-
kranium
Because it is running around 30,000 k memory, which means less than 30MB hash- and you cannot change it to anything higher.
Just a wild guess. Given that Vadim's parallelism is implemented via processes, not threads, could this be 30MB per process, yet reported this way? With 8 processes, this makes 240MB for all of them taken together.
Is R375 publicly available yet?
Dmitri, my expertise won't take me that far. You would have to ask someone who knows much more about this than I do. I remember when Rybka was showing 5,000 k memory in 2 or 3 different places in the list in task mgr.- it was Ray Banks who told me to just ignore all that- it was running plenty of hash. He explained the reason, but my mind did not let me follow much if any of the explanation. But when Ray says it is ok- trust me, it is ok. So I quit worrying. I just thought back to then when you mentioned this with Gull.
Not quite at the halfway mark, and I really had nothing to look back at to have much of any idea how this would go. But with either engine, I did not expect this big a spread after 93 games. It was that at which I was caught off guard.
As for Firenzina, the "large page" option is NOT being used. Too little gain for too much effort.
The above results were for 93 games out of a 100 game match! Where I came up with 200 games I haven't a clue. So a 9 game lead with 7 to play...................... My apologies. Soon I will post the final results, with elo difference, etc.
And the final results for the match- maybe I can get it right this time. Now we see that Firenzina has fairly convincingly beaten Gull R375. Of course it doesn't take Einstein to figure out what should take place now. Firenzina did well- but the question remains if it did any better than Fire 2.2b xTreme GH x64 would do in its place.................. At any rate, a satisfying victory for sure and well deserved:
George, thank you very much for running this match!
Firenzina did well- but the question remains if it did any better than Fire 2.2b xTreme GH x64 would do in its place..................
Another match could answer than nagging question. Yet, as you might guess, I am more interested in finding out where Firenzina currently stands with respect to the TCEC release of Stockfish, the celebrated open-source champ that was #2 in CCRL through a considerable part of 2012 and is currently 3rd or 4th with Rybka. And there is also IvanHoe, a close open-source relative...
Fire 2.2b xTreme GH x64 v Gull R375 x64 at the same exact controls run in above match with Gull and Firenzina. 100 games won't leave anything written in stone, but at least it might give a clue as to any improvement Firenzina has made so far.
2000 or 3000 games would be nice, but I just have other fish to fry. Testing Don and Larry's latest Komodo beta, along with testing Strelka 5.6- which is highly likely to be a beta, and about to start testing the latest Equinox beta gives me plenty to do.
It's not like I don't have the cores, but only 2 hands and part of a mind. So look forward to the results of this match that is beginning.
After Firenzina 2.2.2 xTreme x64's +31 elo win over Gull R375 x64, another 100 game match makes sense to test the possibility that Firenzina may have actually made some improvement over Fire 2.2b Xtreme GH x64. Tread lightly in "no-man's-land" when you start making observations at this point, but I would expect an increase in strength as hard as they have worked. The main point is, and always has been- how far can they take it? At any rate:
The main point is, and always has been- how far can they take it?
George,
Thank you very much for the two matches! We're closer to having a baseline. The main point is exactly the one you mentioned, I totally agree. It's a much harder question to answer than "Where are we now?"
