PH Evaluator

A Poker Hand Evaluator based on a Pefect Hash Algorithm

Overview

Efficiently evaluating a poker hand has been an interesting but challenging problem. Given two different poker hands, how to determine which one is stronger? Or more generally, given one poker hand, can we assign a score to it indicating its strength?

Cactus Kev once gave an answer for a five-card poker hand evaluation. With smart encoding, it ranks each hand to 7462 distinct values.

Still, Kev's solution is specific for a five-card hand. To evaluate a seven-card poker hand (which is more popular because of Texas Hold'em) using Kev's algorithm, one brute force solution is to iterate all 7 choose 5 combination, running his five-card evaluation algorithm 21 times to find the best answer, which is apparently too time-inefficient. Omaha poker would be even more complicated, as it requires picking exactly two cards from four player's cards, and exactly three cards from five community cards. Using brute force, it would take 60 iterations (5 choose 3 multiplied by 4 choose 2) of Kev's 5-card evaluation algorithm.

PH Evaluator is designed for evaluating poker hands with more than 5 cards. Instead of traversing all the combinations, it uses a perfect hash algorithm to get the hand strength from a pre-computed hash table, which only costs very few CPU cycles and considerably small memory (~100kb for the 7 card evaluation). With slight modification, the same algorithm can be also applied to evaluating Omaha poker hands.

Algorithm

This documentation has the description of the underlying algorithm.

C/C++ Implementation

The cpp subdirectory has the C/C++ implementation of the algorithm, offering evaluation from 5-card hands to 7-card hands, as well as Omaha poker hands.

One of the latest benchmark report generated by Google Benchmark:

2020-05-25 03:29:00
Running ./benchmark_phevaluator
Run on (2 X 2800.16 MHz CPU s)
CPU Caches:
  L1 Data 32 KiB (x1)
  L1 Instruction 32 KiB (x1)
  L2 Unified 1024 KiB (x1)
  L3 Unified 33792 KiB (x1)
Load Average: 0.84, 0.29, 0.11
-------------------------------------------------------------------
Benchmark                         Time             CPU   Iterations
-------------------------------------------------------------------
EvaluateAllFiveCards       42539892 ns     42539339 ns           16
EvaluateAllSixCards       358763068 ns    358754423 ns            2
EvaluateAllSevenCards    2712988225 ns   2712943774 ns            1
EvaluateRandomFiveCards        1924 ns         1924 ns       366811
EvaluateRandomSixCards         2031 ns         2031 ns       347350
EvaluateRandomSevenCards       2296 ns         2296 ns       306389
EvaluateRandomOmahaCards       3709 ns         3709 ns       189019

	Number of Hands	Time Used	Hands per Second	Memory Used
All 5-card Hands	2598960	42539892 ns	61 M/s	404K
All 6-card Hands	20358520	358763068 ns	56 M/s	404K
All 7-card Hands	133784560	2712988225 ns	49 M/s	404K
Random 5-card Hands	100	1924 ns	51 M/s	404K
Random 6-card Hands	100	2031 ns	49 M/s	404K
Random 7-card Hands	100	2296 ns	43 M/s	404K
Random Omaha Hands	100	3709 ns	26 M/s	404K

I didn't measure the memory properly. Basically 404K is the maximum memory used in all the evaluation methods.
The performance on random samples are slightly worse due to the overhead of accessing the pre-generated random samples in the memory.

Python Implementation

The python subdirectory has the latest python implementation, which is still in active development. Contributions are welcome.

Other Implementations