From:
http://www.cruxis.com/chess/manual/index.html?cores_and_threads_management.htm"
When using multiple threads, the Split Depth parameter defines the minimum depth at which work will be split between cores."
Imho, you have to "autotune" Houdini, but leave the other engines you have as they are. Nobody knows the settings of these engines (except their developpers, of course). One can expect that the default settings are best, except when told otherwise.
When using multiple threads, it is supposed, logically, the eficiency of the engines to be multiplied automatically by a per cent equal to 100 (i.e. one core = 100 per cent, two cores = 2x100 per cent). But no, it's not like that. Adding a second core doesn't mean you add another 100 per cent efficiency to your analysis power. It means you add efficiency of about 75 per cent to your previous 100. And these per cents diminish with every core you add to your PC configuration - all because of the fact that the cores work in parallel and can often try to work simultaneously on the analysis' lines. So, maybe for analysis it's better to have one powerful core than two weaker ones.
Well, the last is valid when trying infinite analysis. While if you use Aquarium IDeA's methods for analysis, you'll definitely need more cores even if they are weaker. But, that's another story..
As for "how long" when using 'infinite analysis' - mhmm, I think the answer is in the name of the subject.
Now seriously.. Theoretically, the infinite analysis is better when using one powerful core for two days (with maximum hash) than using two cores for one day only. And it depends on the position you analyse for how long you would allow your engine to dig in. There is a "horizon" that engines can't pass, because of the pruning. I.e. they tend to stay solid in their decisions and don't change their evaluations too easy, even they spot something 'interesting' on a very big distance from the root. The pruning at depth, say, 32 plys is enormous - so, it's not an easy decision to change the first move. A year ago, the developper of Houdini suggested (and even shared) the idea to write a special code in the engines which turn the engines to diminish the plys' depth when spotting a new opportunity and to continue from a firm ground.
Some say they never saw an engine to change its decision after 32 plys depth, but I've seen it.. So, as I already mentioned, it depends on the postion you analyse.
Besides, something you don't have to forget - engines are not designed to work in infinite mode. They are made to fight each other in a head to head matches on different time controls.. So, when you use them for infinite analysis, you definitely don't use them as expected. There are other methods that are more efficient, when trying to achieve a good analysis.
I can talk about engines all night long...so it's time to stop, I think. All the more, the self experience is better than reading someone's "how to" writings.
Pages: 
The best analysis program? (Read 154226 times)
