Stockfish 5 and Houdini

Sort:
kprawn

If Stockfish 5 turns out to have a higher rating than Houdini will the chess community just instantly switch over to it for analysis?

Or is there something else beyond the rating that people like about Houdini that makes it the number one choice for analysis?

ClutchCity

People who pay $ for chess software (specifically, computer engines) will probably keep using their commercial engine of choice. Most die-hard enthusiasts probably figured out that Houdini is not worth the money, especially given the recent rise of SF. So, I believe most chess players are using Stockfish DD/5 already (...or some other free program; many strong ones around - a patzer is unlikely to notice any difference between Critter and Stockfish).

On my machine (4 threads), the latest 64-bit SSE 4.2 compilation of SF 5 reaches 5000 kN/s easily. I am impressed.

Karthi_MVK

houdini will improve soon 

MervynS

Didn't even know Stockfish 5 was out

DiogenesDue

They released it the day after Stockfish beat Komodo +13 -6 =45 in the TCEC championship (Houdini was 3rd, just like last year).

WeakLightSquare

Stockfish did crush all oponents in the latest TCEC

nilanjanasm

Stock fish is good

MrDamonSmith

Its rating is what, 3168?

nilanjanasm

Stockfish 5

nilanjanasm

Stockfish is also very good engine

KayZed

Stockfish's analysis in normal games is brilliant, but it couldn't find correct moves in hard chess puzzles, I don't know why. Still, it doesn't make Stockfish a bad engine for analysis.

nilanjanasm

Stockfish is somtimes poor in endgame.

Synaphai
pfren wrote:
nilanjanasm wrote:

Stockfish is somtimes poor in endgame.

Not true.

Komodo is probably still the strongest in endgames, but Stockfish plays endgames way better than Houdini. Actually this is not such a difficult task- Houdini most of the times gives completely bogus evaluations on endgames. Most of my wins in engine-assisted correspondence chess (LSS) were in balanced endgames, where my opponents trusted Houdini- and lost.

I'm afraid I will have to change strategy now, since Stockfish and Komodo are not THAT bad in endgame play.

Could you please post such an endgame? I'm quite curious what it looks like.

EvgeniyZh

pfren wrote:

Kilonodes mean nothing- at all.

All three top engines have their own strong points: Stockfish is fantastic in tactical play, Houdini excels at fast time controls, and Komodo plays the endgame much better than the competition. So, it really depends on your needs.

The average Joe would surely be 100% satisfied with Stockfish, no need to look further.

Stockfish is worse than Houdini in tactics, because of very strong pruning.

ChessMN16

It's true that Stockfish relies on a lot of pruning, but I don't think it's fair to doubt its tactical ability. Look at the games it played against Houdini in the TCEC - given enough time, Stockfish can find absolutely mindblowing tactics.

maruf66076

1 month ago houdini 4 was best chess engine in the world.But After install Stockfish 5 i understand that now it is the best.Houdini 4 could not manage to win with it.And in Stocfish 5 vs Houdini 4 match is always end with a win of Stockfish 5 or a draw.
Now Stockfish 5 is the most strongest engine in the world :)

kontoleon

Ok all this but the true Question is how can culculate correct the ev. Ev +1,25 means that white is "up" 1,25 pawns. but how can culculate the posision if is +0,9 or 1,1 or something else? I see that both computer can see 25 and more moves foward! Ok stong Pruning is good and some less full moves is even good. So my only Question is Both computer program have close enought EV ?

Crappov

Stockfish 5 generally reports higher evaluation scores than Houdini 4 for a given position.  The top 3 progams usually agree on the best move but the evaluation scores can vary.

gautamgreat

whats pruning?
 

Crappov
gautamgreat wrote:

whats pruning?
 

Most modern chess engines are selective about the moves they will consider in a given position.  They do not search every possible move but only the moves that appear most promising according to the program's heuristics.  The moves ("branches") not considered are said to be "pruned" from the "search tree."  This allows the engine to look deeper (more plys) into the position.  However, there's always the possibility that this pruning will lead to an important move being missed.

I'm not an expert on this.  This is my laymans understanding of the term.