How powerful is the machine that beat the poker player champion recently?
From the Deep Stack paper:
This seems to be for training:
For the turn network, ten million poker turn situations (from after the turn card is dealt) were generated and solved with 6,144 CPU cores of the Calcul Quebec MP2 research cluster, using over 175 core years of computation time. For the flop network, one million poker flop situations (from after the flop cards are dealt) were generated and solved. These situations were solved using DeepStack’s depth limited solver with the turn network used for the counterfactual values at public states immediately after the turn card. We used a cluster of 20 GPUS and one-half of a GPU year of computation time. For the auxiliary network, ten million situations were generated and the target values were obtained by enumerating all 22,100 possible flops and averaging the counterfactual values from the flop network’s output.
And this for actual play:
The re-solving computation and neural network evaluations are both implemented in Torch7 (53) and run on a single NVIDIA GeForce GTX 1080 graphics card.
For comparison: The distributed version of AlphaGo took 1.920 CPUs and 280 GPUs to run.