Poker-AI.org

Hello.

I am currently working on a brand-new bot and progressing well on all fronts - input, output and decision-making. One thing I just finished is the pre-flop evaluation for most situations.

I implemented some of it by using look-up tables of 1326 entries (one for each combo) for various kinds of equities. One of them is pre-flop heads-up equity against one random hand. I am able to compute all possible games for two random hands in eight minutes and thirty seconds on a 3.2Ghz quad-core processor - which equates to approximately 5.5 billion 9-hand combinations effective evaluated per second. I'm simply curious how well my algorithm stacks up against others?

Are there ways to improve this even further, so you possibly can generate it on-the-fly every time on program start-up?
Does anyone do the same against two or even three random hands? How quickly do you build that table?

I just read the data from a table available from Pokerstove

You have no idea for how long I was looking for that table . Would you mind sharing?

I still want to compute it myself, as the pokerstove equity ranking is just one of many possible approaches.

I wrote a function in C, which converts hand among a number. And then to determine the winner of the 10 players do not have to all pairwise comparisons. Although the function is written in C, it is too slow for some reason (1,000,000 hands/sec).

You can find a similar thing here: http://poker-ai.org/archive/www.pokerai.org/pf3/viewtopic52b0.html?f=3&t=4644

Edit: Please link to our archive instead of linking back to the original forum since that is going down soon.

I'd be interested in the PokerStove table as well. We can probably start an LUT table thread were we compile all of the different ones that people created together.

_________________
Cheers.

Finally I'm at the right PC, so here's the file attached.

Thanks a lot. One more factor to add to my decision-making .